Due to human activities, organic pollutants are spilled to the environment where they threaten public health, often as contaminants of soil or groundwater. Living organisms are able to transform or mineralize many organic pollutants, and bioremediation techniques have been developed to remove pollutants from a contaminated site. However, fast and easy methods to document both the efficacy of bioremediation and the changes in soil microbial communities during bioremediation are not well developed. The major aim of this thesis was to develop miniaturized methods targeting the physiology of microbial communities during pollutant degradation in soil and to assess pollution-induced community changes. Furthermore, the methods should identify factors limiting efficient pollutant biodegradation. Petroleum hydrocarbons have been chosen as pollutants because these compounds are frequently spilled, readily water-soluble, partly toxic and carcinogenic, and therefore undesirable in soils and drinking water supplies. The influence of petroleum hydrocarbons on a microbial community in the vadose zone was assessed under field conditions. An artificial hydrocarbon mixture consisting of volatile and semi-volatile compounds similar to jet-fuel was emplaced in a previously uncontaminated vadose zone in nutrient-poor glacial melt water sand. The experiment included monitoring of microbial parameters and CO2 concentrations in soil gas over 3 months, in- and outside the hydrocarbon vapor plume that formed around the buried petroleum. Microbial and chemical analyses of vadose zone samples were performed on material from 9 cores drilled on 3 days at 3 distances from the buried petroleum mass to 3.3 m depth. Significantly elevated CO2 concentrations were observed after contamination. Total cell numbers as determined by fluorescence microscopy were strongly correlated with soil organic carbon and nitrogen content but varied little with contamination. Redundancy analysis (RDA) allowed direct analysis of effects of selected environmental variables or the artificial contamination on microbiological parameters. Variation in biomass and CO2 production was to 46 % explained by soil parameters and to 39.8 % by the duration of contamination. The microbial community structure was assessed by community-level physiological profile analysis using BiologTM EcoPlates. Only 35.9 % of variation in BiologTM data could be attributed to soil parameters and contamination, however, the samples with greatest exposure to hydrocarbons grouped together on RDA plots. It is concluded that at this nutrient-poor site, the microbial community was dominated by the natural heterogeneity and that the influence of petroleum hydrocarbon vapor was weak. BiologTM Eco-Plates combined with RDA were able to distinguish between samples with high and low hydrocarbon exposure. However, the method was not sensitive enough to produce consistent patterns in three replicates when extracts from subsurface soils with low cell numbers were inoculated. A more sensitive, MPN-based system was therefore developed and its ability to provide information about specific pollutant degraders was tested in alluvial sand. Eight dilutions of sand extracts were inoculated on medium- and substrate-containing polypropylene deep-well plates. To keep the concentrations of the volatile substrates constant over several days of incubation without intoxicating the soil community, a substrate-containing organic carrier phase was applied to each vial. The biochemically inert 2,2,4,4,6,8,8-heptamethylnonane (HMN) was appropriate for this purpose. Thus, the volatile substrates partition from HMN into the microorganism-containing soil water where they achieve relatively low concentrations. The sensitive watersoluble tetrazolium dye (WST-1) was added to each well after 7 days incubation to detect dehydrogenase activity. A dark yellow signal was developed for the substrates toluene, n-octane, n-dodecane, 1,2,4-trimethylbenzene and methylcyclohexane but not for isooctane, being strongest in n-alkane amended wells. With this method, more hydrocarbon degrading bacteria were detected in an alluvial sand exposed to kerosene for 72 hours compared to the community in the pristine sand. Both communities were mainly composed of n-alkane degraders. Although encouraging results were achieved, we noticed that the tetrazolium reduction was inhibited in some cases. Wells with high cell numbers, in which obvious growth occurred, did not produce a signal. In a next step, the two-liquid phase system with HMN was applied to modify a miniaturized respiration detection system using whole soil samples instead of soil extracts (Campbell et al., AEM 69, p. 3593). Soil was incubated in deep-well plates in presence of single volatile organic hydrocarbons. The carrier phase was added to the interstitial space in a bead layer on the bottom of the well (below the soil sample), and the hydrocarbons diffused into the soil-water. The soil activity was determined by means of an agar plate containing a pH indicator, which changes color as a function of the produced CO2. Physiological profiles specific to petroleum hydrocarbons in pristine and contaminated soils were assessed. The substrate concentrations, which induced highest microbial activity, as well as concentrations causing inhibition, were determined with effect-concentration curves. Community-level physiological profiles based on hydrocarbon degradation were obtained by applying multivariate analysis on the CO2 yields. The first measurement period of 0-6 hours after addition of relatively high hydrocarbon concentrations to the system yields values, which best separated soil types as well as pristine and contaminated soils. If low concentrations were applied, we obtained only significant separations after 24 hours incubation, probably after a growth step. The same micro-respiration system equipped with a 14CO2 detection plate was applied to study N and P limitation of aerobic mineralization of 14C-toluene in a soil previously exposed to petroleum. Significant nutrient limitation during short-term tests (24-48 hours) was identified and the optimized combination of nutrient addition was determined. Conclusively, miniaturized physiological methods have been proven to detect soil microbial community changes in petroleum hydrocarbon-contaminated soils. Adding an organic carrier phase to the miniaturized systems allowed assuring a volatile organic hydrocarbon supply in non-toxic concentrations in the soil water or the medium for several days. Thus simple, cost-effective systems to investigate the soil community during bioremediation and to address questions of nutrient shortage in soils have been provided with this thesis.
EPFL2004
