Sampling design and enumeration statistics for bacteria extracted from marine sediments.

The spatial and temporal distributions of marine bacteria were studied at both a muddy and a sandy subtidal site in North Inlet, S.C. The sampling design was hierarchical, since subsampling (by a dilution series) of the sediments was necessary to count bacterial cells using acridine orange epifluoresence microscopy. The cell count data fit a log-normal distribution. The abundance of bacteria was 10 g (dry weight) of mud and 10 g (dry weight) of sand. Variance component analyses demonstrated that variation due to the subsampling procedures was always statistically significant. Thus the common practice of counting 20 fields from one filter preparation is inadequate for estimating the true bacterial population variance in marine sediments. It is recommended that replication of the subsampling level be performed. Standardization of data (by dry weight of sediment) decreased sampling variance at the mud site but not at the sand site, implying that bacteria are more homogeneously distributed in sand than in mud.