Despite the consensus expressed in Agadir, I now doubt we can do without keys and references. At the very least, when we flesh out the unfinished notion of hosts, it is likely to be important that, in the descriptions of the hostee (what's the biological term?) we need a reference, not a mention. That's because many things can have the same host. If we are content with just a scientific name as the host, maybe this is unnecessary. But we know in other contexts, and probably in this one, that more information is needed, including literature citations.

Another place keys and references could appear is in references to terminology in the event that we embed that in the instance documents in some way. See GISINdefinitionsMatrix and topics referenced there.

-- BobMorris - 15 Mar 2006