I just had a though as to whether this could be anything to do with the Wiznet5100 issue over not releasing the SPI bus properly?
I don't seem to be having any issues on a transmitting node that doesn't have an Ethernet shield.
A friend of mine is developing in parallel and he has seen the same error after a period of receiving.
EDIT: It looks like I have a clone of V5 of the Ethernet shield. It does look like it works OK as it happily receives data from the nRF and posts it to a webpage for insertion into a database. This happens about 50-60 times before it fails with the address error.