Is there any lower limit in the number of sites to be meaningful? Not really, but few sites will usually result in large standard errors of the indices and trends.