How to determine size for sub.sample


I am unclear on how to decide on the size number for the sub.sample command. I have read up on this but confused as to how to define size number?



when I subsample my dataset I usually perform the count.groups ( command. Then I choose the sample with the fewest number of sequences across all samples. It’s also described in the 454 SOP, below in the OTU analysis section -
Additionally, some commands have the subsample=T option which chooses automatically the group with the fewest sequences as minimum number of subsamples - for example summary.single, summary.shared or tree.shared.
Hence, I would go as probably best size for subsampling with the fewest number of sequences in one group across all samples.