Banks have two sets of inputs:
(1) Equity (i.e. the owners of the bank put up capital. This is great because it requires them to have 'skin in the game')
(2) Incoming debt (i.e. depositors give them money.).
They have two outputs:
(3) Outgoing debt (i.e. the loans the bank makes out to people, in the form of mortgages, small business loans, lines of credit, etc.)
(4) Reserves (money they have to keep on hand)
Notice that as an accounting identity, we must have that (1) + (2) = (3) + (4).
Leverage limits essentially limit the ratio (2):(1). For example, a leverage limit of 9:1 means that a bank lends out $100 has to source at least $10 of that from equity, and at most $90 of that from incoming debt.
Fractional reserve limits essentially limit the ratio (3):(4). As "user downandout shows in a sister thread to this one, this can be used to determine a "multiplier" for the economy as a whole, since for a nonzero limit, there is a closed form fixed point solution maximum multiplier ratio in the economy as a whole. (Although in practice the "true" multiplier can be lower since people might stuff their mattresses with money, or have lots of loose change lying around, etc.).
Note that the fractional reserve limits are something that is enforced on banks by the government to limit the multiplier. This restriction doesn't exist on P2P lenders (as of yet), and so if any thing, it means that P2P lenders are even freer to do bad things (such as have no reserves on hand at all).