Firstly, a dependent claim is always completely included in its parent claim. (That answers the first question.)
Secondly, (as I'm sure you're aware) a compound having first_R2 = H and second_R2 = CH3 is included in claim 1 but not claim 2.
So, if somebody shows up with prior art having first_R2 = H and second_R2 = CH3 then that could invalidate claim 1, but claim 2 could still be good. (Hopefully that answers the second question.)
As to why someone would do it this way instead of having two separate independent claims -- in the US, 3 independent claims (and 20 claims total) are included in the USPTO basic filing fee -- you pay extra if you exceed the 3/20 count. And an extra independent claim is approximately 5x the cost of an extra dependent claim. (Search for 37 CFR 1.16(h) and 1.16(i)). So people try to conserve the number of independent claims.
Unfortunately the different ways of organizing claims aren't exactly equivalent. For instance, in your example, a prior art compound with both R2 = H would knock out both claim 1 and claim 2, leaving no patent protection for the H/CH3 compound. It would be nice if there was a claim 3 directed to the H/CH3 compound only! (But it looks like there isn't.)