The problem of how much sampling is enough to validly characterize a living
molluscan fauna is not unlike the same problem palentologically.  In the
first case, you might measure collecting hours, or number of collecting
trips, or person/days, etc.  In the second case, it is usually
pounds/tons/cubic feet of sediments to yield a statistcally defendable
sample.  Seems to me someone published about this problem in regard to the
Chesapeake Bay Miocene - perhaps Earl Shapiro?