Coverage is very dependent on experimental design and goals so each project can have different requirements.
Illumina suggests 100,000 reads/sample for 16S analysis. Many researchers have done much lower reads with good results. We have multiplexed 288 samples using Shoreline Biome V4 kits and sequenced on MiSeq with sufficient coverage.
For PacBio sequencing with the StrainID kit we suggest running ~96 samples on one SMRT cell. This gives sufficient reads/sample to identify bacterial strains present at 1% with reasonable confidence on the Sequel I. In general, more reads will yield greater depth per sample, enabling identification of bacteria present at lower levels in the sample. For highly diverse samples where representation below 1% is needed, more than one Sequel I SMRT cell or a single Sequel II SMRTcell may be advisable.