Succession following restoration can be stochastic, and ecosystem function does not always return to a desired level due to ecological constraints. Wetlands are important for the biological removal of nitrate by denitrification, but restoration of this service is not always successful. This may be associated with development of the microbial community responsible for this function. We investigated whether denitrification was successfully restored in four Illinois wetlands between 6 and 16 years post restoration. We determined whether communities followed a trajectory of change during this period, and further, whether communities in the restored wetlands were becoming more similar to those in adjacent reference wetlands. To monitor community changes, bacterial 16S rRNA genes and a suite of denitrification genes were sequenced by Illumina MiSeq. Potential denitrification rates were determined using acetylene block assays. Changes in composition were assessed by calculating Bray-Curtis values between the restored wetland in each year to the community in the adjacent reference wetland during the final year of the study, as well as for each pair of restored and reference wetlands in each year individually. If development of denitrifying services in restored wetlands is dependent on microbial succession, then we expect similar trends in composition and function.
Results/Conclusions
Potential denitrification rates in the restored wetlands did not increase smoothly over time, and they were greatest during a year that experienced drastic flooding in the middle of the study. Denitrification rates were typically greater in reference than in restored wetlands. Microbial communities in the restored sites did not follow a linear trajectory, and they did not become more similar to their respective adjacent reference wetlands. Neither the total bacteria nor the denitrifiers showed significant trajectories over time. Across all years, all four paired sites showed differences in composition between the restored and reference community, as demonstrated by significant PERMANOVA R2 values, and this was true for total bacteria and denitrifiers. A Mantel correlation was used to compare differences in community composition and rates across the entire study, and showed that there is no clear relationship between denitrification and either total bacterial or denitrifier composition. Denitrification rates were strongly influenced by wetland hydrologic conditions (e.g. flooding), so function did not follow similar trajectories as the community. Most notably, however, this study found that the restored and reference wetlands did not achieve comparable rates more than a decade following restoration, which coincided with significant differences in bacterial and denitrifier community composition.