Response of Wetland Soil Carbon to Groundwater Conservation: Probabilistic Outcomes from Error Propagation

Abstract Water loss compromises functions performed by wetland ecosystems. Efforts to rehabilitate wetland function typically begin with attempts to reestablish hydrology. These activities are often not monitored, so tools to extract information from them could partly offset the lost opportunity to learn from whole-ecosystem hydrological manipulation. In 2002, groundwater abstraction was lessened by 35% throughout 1700 km2 of west-central Florida (USA). I assembled a pathway of correlations to project how this hydrological manipulation affected water levels and soil carbon (C) storage in overlying wetlands. Parameter values and residual error in these statistical models were resampled from known variances, thereby propagating uncertainty through the pathway of relationships, and expressing the response of soil C probabilistically. Projected soil C probability distributions were most distinguishable between full and moderate (30% less) abstraction. With more severe abstraction cutbacks, gains in projected soil C became more marginal and uncertain, suggesting that wetland soil C pools are not notably impacted by low-volume groundwater abstraction. Reducing uncertainty in projected soil C will require better understanding the dynamic response of soil C to increases in the amount of time that wetland soil is inundated. The step-by-step error propagation routine presented here is a platform for assimilating information from diverse sources in order to project probabilistic responses of ecosystem function to wetland restoration attempts, and it helps identify where further certainty is most wanted in a pathway of cause-effect relationships