A Bayesian hierarchical modeling approach is introduced to pool data properly from multiple flow-through wetlands to estimate wetland-specific long-term phosphorus retention capacity. By pooling data from multiple wetlands, we overcome the difficulties in estimating the effectiveness of using constructed and natural wetlands for nutrient reduction. The Bayesian hierarchical modeling approach reduces estimation uncertainty by shrinking wetland-specific estimates towards the overall average of the same quantity from multiple wetlands, facilitating information sharing across sites, thereby reducing the demand on sample sizes from individual wetlands and avoiding several common pitfalls of using large data (i.e., from multiple systems) induced by Simpson's paradox. In this paper, we develop a sequential updating framework to alleviate the computational burden of compiling and modeling data from multiple wetlands. We then demonstrate the sequential updating process to estimate retention capacity of a suite of wetlands in Ohio, USA. A total of four wetlands, representing both natural and constructed wetlands, were used. The estimated total phosphorus retention capacities range less than 0.01 to well over 1 ton per year per system. As wetland restoration initiatives expand around the Laurentian Great Lakes and nationally, this model serves as an important initial step in developing tools to meet nutrient reduction goals and standards. Extending this work, we have developed a publicly accessible on-line open computation platform that can help natural resource specialists better plan for wetland efficacy in the future.