Lieven Van Vooren

Utility of the Biolog System for the Characterization of Heterotrophic Microbial Communities

Summary The Biolog system was evaluated as a simple and rapid assay to characterize heterotrophic microbial communities. Factors influencing the substrate utilization profiles of microbial communities present in surface water and compost material were determined. Examination of five different production batches of Biolog GN microplates showed that the color development in the substrate response wells is significantly affected by the production batch. Although community Biolog profiles within the same production batch of GN microplates were reproducible, the color development in replicate control wells contained in Biolog MT microplates varied significantly. This implies that several substrate-free reference wells need to be inoculated with the same environmental sample to allow reliable data transformations of the 95 raw color responses in the GN microplates. Replicate suspensions of compost material could not be distinguished from each other using Biolog GN microplates originating from the same production batch, suggesting that the sample preparation procedure was highly reproducible. However, the data indicate that reliable use of the Biolog assay for the characterization of heterotrophic microbial communities is only possible by comparing samples of equivalent inoculum densities. Also, multiple readings over a time course of incubation are needed, to overcome the effect of nonlinear color development.

Modelling the color development in Biolog microtiter plates by the Gompertz function.

Summary The color development curves obtained from multiple readings of Biolog GN microtiter plates over a long incubation time were fitted to the Gompertz function. This yielded — for each sole-carbon source — the incubation time independent and biologically relevant Gompertz parameters A (maximal extent of color development), μ M (specific color development rate) and λ (lagtime). To evaluate the applicability of the model, the coefficients of determination, the residuals, the biological significance of the parameters, the stability of the parameter estimates and the predicting power of the model were determined. For the pure strains Vibrio alginolyticus and Pseudomonas fluorescens the model passed all the criteria. For the model community consisting of a mixture of 9 strains, all criteria were met for almost all the oxidized carbon sources. However, a high amount of measuring points in the color development curve is necessary (preferably >20) and the reading frequency needs to be the highest in the lagphase and the exponential phase to ensure a high stability and predictability of the estimates of μ M and λ, especially when the color develops very rapidly in the wells. For a few carbon sources the Gompertz model was not appropriate, as the color development curve showed clearly 2 tiers. However, it is substantiated that the second tier may be eliminated from the time series, which allowed the model again to pass all the criteria. Finally, comparison of Biolog fingerprints of environmental samples was made with principal component analysis of the estimated Gompertz parameters for each carbon source. It is shown that this approach is practicable and may yield consistent results for environmental microbial community analysis.