Analytical approaches to the characterization of samples of microbial communities using patterns of potential C source utilization

Profiles of potential utilization of 95 separate C sources by microbial communities can be readily generated from direct incubation of environmental samples in BIOLOG microplates. Color formation from a redox indicator dye is used to quantify the degree of C source utilization. I aimed to examine different analytical approaches for classifying microbial communities based on these profiles. Specifically, the relative effects of average rate of color development versus the pattern of relative C source utilization on the classification of rhizosphere samples from different crop types were evaluated. The average rate of color development was correlated to the density of total (acridine-orange) bacterial cells ( R 2 = 0.52) and active (5-cyano-2,3-ditolyl tetrazolium chloride) bacterial cells ( R 2 = 0.70) inoculated into the plate. Analysis of plates with different rates of color development after a specific incubation period resulted in samples with variation in the overall extent of color development (expressed as average well color development, or AWCD). Classification of these samples using principal component analysis was significantly influenced by the variation in AWCD, resulting in the classification of samples based on the density of inoculum rather than the pattern of C source utilization. The effect of variation in AWCD was eliminated by normalizing data prior to ordination, or by using an alternative ordination technique, detrended correspondence analysis. Variation in AWCD can be limited through multiple-plate readings and subsequent selection of plates with a common reference point in AWCD. The specific AWCD used for analysis does not appear important for classification purposes; consistent discrimination of rhizosphere samples from different crop types was apparent for analysis across a wide range of AWCD (0.25–1.00 abs. units). The specific differences in C source utilization between rhizosphere sample types did depend on the set point used for analysis due to the differences in the rate of color formation among wells. Results suggest that single-plate readings can be used to classify samples, but only if potential differences in AWCD are accounted for in the data analysis. Repeated plate readings will provide a more complete understanding of differences in C source utilization among samples.