Plant species and temperature effects on the k-C* first-order model for COD removal in batch-loaded SSF wetlands

The modified k-C-* first-order degradation model proposed by Kadlec and Knight [Kadlec, R.H., Knight, R.L., 1996. Treatment Wetlands. Lewis Publishers, Boca Raton, FL] was fit to 192 data sets of COD concentration versus time measured in batch-loaded wetland microcosms. The time series sets were divided into four replicates of four plant species treatments; Carex utriculata (sedge), Schoenoplectus acutus (bulrush), Typha latifolia (cattail) and unplanted controls housed in a controlled-environment greenhouse in which temperature was cycled in 4 degrees C increments from 24 to 4 degrees C and back to 24 degrees C over a yearlong period. One 20-day batch incubation was conducted at each temperature setting during which seven chemical oxygen demand (COD) samples were drawn from each of 16 wetland columns. Non-linear mixed effects regression was used to fit parameters of the model. Best-fit values of the rate parameter k and the background concentration C-* varied significantly by plant species and greenhouse temperature setting. An Arrhenius relationship was used to assess the effect of temperature on these parameters. Species greatly influenced the value of volumetric rate parameter at 20 degrees C (k(20)) and ranked Carex > Schoenoplectus > Typha > unplanted control (0.925, 0.743, 0.612 and 0.366 day(-1), respectively). All displayed decreasing values for increasing temperature but species variation was much less (theta(k) = 0.945, 0.957, 0.953 and 0.936, respectively). Background COD concentration values at 20 degrees C (C-20(*)) ranked Carex < Schoenoplectus < Typha < unplanted control (42, 46, 66, 67 mg L-1, respectively) but seasonal variation was mixed (theta(C*). = 1.029, 0.999, 0.958 and 0.935, respectively), The results indicate that (1) seasonal variation of performance is plant-species specific; (2) singly, the rate constant k cannot capture the variation in performance due to temperature as much of the variation is reflected in the C-* and (3) there are strong interactions between values of k and C-*. (C) 2005 Elsevier B.V. All rights reserved.