The present study sampled every month from June to November in 2016, aimed to assess the effect of oxidoreductase activity on the removal rate of pollutants from constructed wetlands and to determine whether matrix enzyme activity can be used to assess the effectiveness of overall removal rate. Integrated vertical-flow constructed wetlands that incorporated two substrates (bio-ceramics and pebbles) and three plant species () were constructed to investigate the activity of four oxidoreductases (dehydrogenase, polyphenol oxidase, peroxidase, and nitrate reductase) and to explore the relationship between enzyme activity and removal rates of total nitrogen (TN), nitrate nitrogen (NO₃^--N), total phosphorus (TP), orthophosphate (PO₄^3--P), and COD_Mnfrom aquaculture wastewater. Oxidoreductase activities were significantly different in the experimental operation cycle. Dehydrogenase and peroxidase activities initially increased and then decreased, and the enzyme activities reached maximum levels during the peak of plant growth. Polyphenol oxidase and nitrate reductase activities were not significantly different from June to August and reached maximum levels when the plants stopped growing. Dehydrogenase, polyphenol oxidase, and peroxide activities exhibited similar vertical distributions (i.e., surface > middle > lowest), whereas nitrate reductase activity exhibited the opposite trend. There were significant positive correlations between nitrate reductase activity and the removal rate of NO₃^--N (=0.0444) and between dehydrogenase activity and the removal rate of COD_Mn (=0.0439). The activities of polyphenol oxidase and peroxidase were not significantly correlated with the removal rates of TN, NO₃^--N, TP, PO₄³-P, and COD_Mn. Accordingly, dehydrogenase can be used as an indicator of the effectiveness of COD_Mn removal in constructed wetlands, and nitrate reductase activity can be used as an indicator of the effectiveness of NO₃^--N removal.