Microbial community analysis for denitrifying biofilters.

摘要

Denitrifying biofilters are a promising and low-maintenance technology for removing nitrate from agricultural drainage, capable of removing 50--80% of annual nitrate loads. Because increased riverine nitrate concentrations are correlated with indicators of eutrophy and degradation of coastal waters, including a large hypoxic zone in the Gulf of Mexico (Turner and Rabalais 1994; Lohrenz et al. 1997; Rabalais and Turner 2001), denitrifying biofilters have the potential to substantially improve surface water quality. However, our understanding of ecological factors influencing biofilter performance was limited. Therefore, in this study, the microbial community of the biofilter was described using several different techniques. Denitrifying enzyme assays with inhibition showed that denitrification is primarily mediated by bacterial populations, but that fungi are also indirectly important. These assays also showed that denitrification occurs both on the surface of the biofilter woodchip media and in biofilter water, and that washing woodchips in buffer solution was able to remove cells for further study. Sample preparation methodology for use with FISH was developed using vortexing of samples with glass beads to aid in removing cells from woodchip debris. Community fingerprint techniques using ARISA and nosZ t-RFLP were also developed to provide high throughput bacterial and denitrifying community data.;The spatial structure of microbial communities in a biofilter was characterized using mapping of ecological metrics, MDS plots, and a new geostatistical method---ANOSIM-GS---developed for this research. ANOSIM-GS provides a robust way to characterize the variation of ecological communities with separation of space or time. Using these tools, significant spatial variability in overall bacterial (ARISA) community spatial structure were observed across sampling depth and in the direction of biofilter flow, but not in the direction of cross-flow. Bacterial community correlation distances of 6.1 m at the 0.76 m biofilter depth, and 10.7 m at the 1.52 m biofilter depth were calculated. No significant spatial structure in the denitrifying (nosZ t-RFLP) community was observed.;Time-series data were collected from three biofilter sites beginning in November 2008, including performance, microbial community, and environmental data. Using ANOSIM-GS, bacterial and fungal communities (ARISA and fARISA) were shown to have temporal structure. Bacterial communities in all three biofilters showed a correlation time of approximately 125 d and signs of annual cyclic patterning. Correlation times in fungal communities were more variable, between 100--200 d, and annual cyclical patterning was less. Communities were also structured by space in the time-series data. Analysis of the relationships between community, performance characteristics, and environmental variables yielded several results. A subset of 39 bacterial and fungal populations accounted for 80% of the community variation both between and within biofilter sites, and several of these populations were significantly associated with variation in nitrate removal. Microbial community composition was found to be structured by changes in temperature, inlet nitrate, pH, moisture content, and depth (distal controls on denitrification). Additionally, nitrate removal was also significantly affected by COD, DO, flow, temperature, and moisture, apart from the influence of any of these parameters on microbial community structure (proximal controls on denitrification). These results suggest that inoculation of well-performing species, or changes in the biofilter environment, either to restructure community composition or to improve the denitrification rate of those populations already present, may improve biofilter performance.