Abstract Seven constructed wetlands (0.05–0.69 ha), situated in agricultural catchments (22–267 ha) in the south of Sweden, were studied for two years with two aims: to (i) quantify their function as sinks for particles and phosphorus (P) lost from the catchments, and (ii) investigate to what degree catchment and wetland characteristics and modeled loads (using hydrochemical catchment models) could be used to explain differences in retention between the wetlands. The wetland areas ranged from 0.04 to 0.8% of the respective catchment area, and they were situated in areas dominated by fine-textured soils with relatively high P losses and the main proportion of P transported in particulate form. Net P and particle retention were estimated during two years from annual accumulation of particles on sedimentation plates (40 × 40 cm) on the bottom of the wetlands.

There was an annual net retention of particles and P, but with a large variation (for particles 13–108 t ha⁻¹ yr⁻¹ and for P 11–175 kg ha⁻¹ yr⁻¹), both between wetlands and between years. The difference between the two years was larger than the difference in mean P retention between the seven wetlands. There was a positive relationship between P and particle retention and three catchment factors, i.e. P status (P-AL) of agricultural soils, average slope in the catchments and the livestock density, and a negative relationship with the agricultural soil clay content. In addition, there was a positive relationship with the wetland length:width ratio. Contrary to expectations, neither the modeled hydraulic load nor P load was significantly correlated with the measured particle and P retention. There was also a positive relationship between P concentration in the sediment and soil P status in the catchment. The results imply that considerable errors are introduced when down-scaling modeled regional nutrient losses to estimate the P loads to small wetlands in agriculturally dominated catchments. A more qualitative approach, using catchment characteristics for identification of hot-spot fields, may be equally good to identify suitable locations for constructed wetlands to reduce diffuse P loads. © 2014 Elsevier B.V.