Conclusion This study will help to make better choice of wastewater plant system for other cities of Pakistan

because this technology is not expensive and easy to adopt. The ponds are suitable as fish farms and we can use this water for crops and farming.

CONCLUSIONS
A mediumdeep to deep DSL system (50150 cm) essentially functions as a facultative lagoon with respect to COD removal. The role of duckweed is marginal in removing COD from wastewater. DSL depth in the range of 10100 cm had no other eect than that it increased the surface load-ing ls (as kg CODlt /ha) (by increasing the CODlt mass input, in the experimental set-up). Surface-re-lated processes, notably oxygen transfer, were not strongly aected. A linear relationship was established between applied and removal COD lt surface loading. The best tting linear equation is lr=0.53 _ ls+66 with r 2=0.98. Another linear relationship was also established between applied and removal CODlt concentration. The best tting linear equation is: CODlt removal=0.82 _ (CODlt initial)25, with r 2=0.93. The concentration removal rates depended on in-itial surface loading ls, which in turn depends on initial CODlt concentration. This conrms that CODlt removal is largely determined by volume-re-lated microbial processes and not by surface-related duckweed uptake or oxygen uxes.ls determined the overall oxygen demand; at in- itial ls > 800 kg CODlt/ha the oxygen supply might become rate limiting. The COD lt removal rate followed rst-order kinetics, with a reaction constant k of 0.040.06 d 1.CODlt surface loading removal lr was in the range of 86 to 1397 kg/ha 20 days at the ls of 138 and 2185 kg/ha, respectively. E.uent quality ranged between 27 and 100 mg COD lt/l, equivalent to 1763 mg BOD5,total/l.A depth of up to 11.5 m is not likely to limit adequate DSL performance with respect to COD degradation. Thus, depth of DSL should be chosen as a function of inuent water quality and desired e.uent quality. Mixing (up to 2.3 W/m3) has a moderately benecial eect by raising DO, favoring COD removal up to 10%.

4 DISCUSSION
Copper and nickel are essential in low concentrations to the metabolism of animals and plants being present as constituent of numerous proteins, enzymes and cofactors. However, when present in excess, they may cause deleterious effect (lethal or sublethal stress). In this study, we demonstrated that copper when present in the nutrient solution at concentrations 0.3 mg/L was tolerated by L.

5 CONCLUSION
In this work, the tolerance of L. gibba L. to copper and nickel and the potential accumulation of these two metals by this genius have been investigated. The metals Cu and Ni were tolerated by L. gibba at 0.3 and 0.5 mg/L respectively. These results revealed high tolerance of this aquatic plant to the heavy metals copper and nickel. On the other hand, the high accumulation capacity of L. gibba for copper (290 mg/Kg DW) makes this genius a good candidate for the removal of low concentrations of this heavy metal from contaminated water (municipal effluents in which Cu content is relatively low). However, the low accumulation capacity of L. gibba for nickel (at concentration 0.5 mg/L) makes this aquatic species not very suitable for the phytoremediation of Ni-contaminated waters.

DISCUSSION
Depth variation in the range of 10100 cm did not aect COD removal in DSL. Similar con-clusions were reported by others albeit within much smaller depth ranges (Oron, 1988; Vroon and and Weller (1995) reported DO of <0.1 mgO2/l in higher loaded (3781410 kg COD/ha with settling) laboratory-scale experiments of 0.150.6 m depth without mixing, which is comparable to our results. E.uent DO was inuenced by initial ls and by mixing. Initial ls of below 800900 kg/ha (at zero mixing) was found to exert a suciently low oxygen demand to cause some oxygen to be in excess beyond the heterotrophic demand, and nal DO to be >0 mg/l. In reactors with ls < 800 kg CODlt/ ha, oxygen consuming substances are most likely to have been fully oxidized by the end of the exper-imental period. The pH in all reactors ranged between 7 and 8. This fact and the equally stable DO conrmed the absence of pronounced algal photosynthetic ac-tivity. The observed small quantities of algae did not signicantly interfere. Vroon and Weller (1995), Alaerts et al. (1996) and Ko rner et al. (1998) also reported similar pH ranges, as well as the absence of a diurnal prole or stratication. lr was not statistically signicantly dierent between duckweed-covered and control reactors. This suggests that the duckweed cover did not play a role in the COD lt removal. Attachment of het-erotrophic biomass to the plant roots was less sig-nicant at high depths given the high volume-to-surface ratio. The ux of oxygen into the liquid is the deciding and rate limiting factor for COD removal processes. COD lt removal rates were inuenced primarily by initial CODlt concentration, as well as by ls. Higher CODlt concentration probably stimulates heterotrophic bacterial metabolization according to rst-order kinetics. The ls, which depends on con-centration and depth, determines the overall oxygen demand exerted. At ls > 800 kg CODlt/ha oxygen tends to become rate limiting, though this did not cause detectable deviations in the linear correlation between the removal rate and the initial BOD lt concentration. The CODlt removal eciency ZCOD (after 20 days) was comparatively low at 5070% when com- pared with conventional algae-based sewage lagoons and the full-scale DSL (Alaerts et al.,1996). This can be attributed partly due to an in-adequate supply of oxygen, and to the fact that removal performance in this study was calculated against COD lt in the inuent. If removal

results were adjusted to CODtotal, overall removal rates would increase to 8090% and above 90% when calculated on BOD5,total. Oron et al. (1988) studied COD removal in DSL systems of 20 and 30 cm using a comparable settled wastewater with initial CODtotal of 318 mg/l, which corresponded to a ls,t of 636 and 954 kg/ha, re-spectively. Removal rate on COD total lr,t was 328 and 402 kg/ha 10 days. Vroon and Weller (1995) conducted outdoor batch experiments at depths of 15, 30, 45 and 60 cm with settled wastewater at CODtotal of 252, 241, 236 and 235 mg/l, which cor-responded to ls,t of 378, 723, 1062 and 1410 kg/ha. They achieved a lr,t of 280, 432, 450 and 460 kg/ha 11 days at a water temperature of 16208C, which is signicantly lower than in our experiments. Mandi (1994) in continuous outdoor experiments at 14 cm depth in summer in Marrakesh, Morocco,using raw wastewater with ls,t of 427 and 622 kg COD total/ha achieved lr,t of 270 and 324 kg/ha 7 days (water temperature was not measured during the experiments). Other researchers reported higher results. This could be attributed to higher water temperatures or longer retention times. A pilot plant (Italy) with ls,tof 7268 kg COD total/ha (316 mg/l) of settled waste- water and a depth of 2.3 m yielded a lr,t of 5796 kg/ha 16 days in summer at water temperature of 158C, but of only 3634 kg/ha 15 days in winter at a water temperature of 5108C (Bonomo et al., 1997). Alaerts et al. (1996) in a full-scale DSL of 0.51 m depth reported a lr,t of 1399 kg/ha 20 days at an in- itial ls,t of 1788 kg CODtotal/ha. All previous stu-dies used settled municipal sewage to exclude comparatively large variable amounts of particu- lates. Ko rner et al. (1998), using non-settled munici-pal wastewater at 208C, reported much higher lr,t of 38, 81 and 202 kg/ha 3 days at ls,t of 49, 104 and 251 kg CODtotal/ha, respectively. These are less representative of eld conditions because of the pre-sence of settleable matter and the very shallow water depth of 3.3 cm.The reaction constant for COD total removal in 1 m deep facultative ponds with an inuent of 504 mg COD total/l non-settled sewage at HRT of 16.8 days in Portugal is 0.35 d 1 (Gomes de Sousa, 1987), this is much higher than the k of 0.04 0.06 d 1 found here. This dierence can be attribu- ted partly to the fact that our results are based on COD lt. In addition, the kinetics appear fairly independent of the dierences in oxygen transfer (aera- tion) associated with shallow to deep reactors (10100 cm), in DSL as well as in algae-based systems. This suggests that under our experimental con-ditions oxygen transfer could be a rate limiting fac-tor.

CONCLUSIONS
A mediumdeep to deep DSL system (50150 cm) essentially functions as a facultative lagoon with respect to COD removal. The role of duckweed is marginal in removing COD from wastewater. DSL depth in the range of 10100 cm had no other eect than that it increased the surface load- ing ls (as kg CODlt /ha) (by increasing the CODlt mass input, in the experimental set-up). Surface-re- lated processes, notably oxygen transfer, were not strongly aected. A linear relationship was established between applied and removal COD lt surface loading. The best tting linear equation is lr=0.53 _ ls+66 with r 2=0.98. Another linear relationship was also established between applied and removal CODlt concentration. The best tting linear equation is: CODlt removal=0.82 _ (CODlt initial)25, with r 2=0.93. The concentration removal rates depended on in- itial surface loading ls, which in turn depends on initial CODlt concentration. This conrms that CODlt removal is largely determined by volume-related microbial processes and not by surface-related duckweed uptake or oxygen uxes. ls determined the overall oxygen demand; at in-itial ls > 800 kg CODlt/ha the oxygen supply might become rate limiting. The CODlt removal rate followed rst order kinetics, with a reaction constant k of 0.040.06 d 1.CODlt surface loading removal lr was in the range of 86 to 1397 kg/ha 20 days at the ls of 138 and 2185 kg/ha, respectively. E.uent quality ranged between 27 and 100 mg CODlt/l, equivalent to 1763 mg BOD5,total/l. A depth of up to 11.5 m is not likely to limit adequate DSL performance with respect to COD degradation. Thus, depth of DSL should be chosen as a function of inuent water quality and desired e.uent quality. Mixing (up to 2.3 W/m 3) has a moderately benecial eect by raising DO, favoring COD removal up to 10%.

RESULTS AND DISCUSSION Phytoremediation potential of duck weed (Lemna minor L.) in the removal of pollutants from refinery wastewater was studied by analyzing the refinery wastewater before and after the Phytoremediation. The value before and after the Phytoremediation and percentage changes was reported as initial and final value in Table 1. Table 1: Parameters Unit Initial Final Percentag Reduction Concentra Concentra e percentage of tion tion reduction pollutants from refinery wastewater by Lemna minor L. No 1 Water C 29 24 17.2% Temperature 2 pH 8.9 7.7 13.4%

3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18

Alkalinity Turbidity TSS TDS Sulphate Sulfide Nitrate Phosphate Oil & Grease Phenols BOD5 COD Pb Cu Zn Cd

mg.L NTU mg.L mg.L mg.L mg.L g.L g.L mg.L mg.L mg.L mg.L g.L g.L g.L g.L

-1

-1 -1 -1 -1

-1 -1 -1 -1 -1 -1

-1 -1 -1 -1

95 25 21 4130 1683 23 21 7.01 19.8 0.5 27.6 480 16 12 43 5.1

56 9 13 2110 1456 14 9 4.9 11.2 0.3 13.9 323 0. 2 0.02 12 0.02

41% 64% 38% 48.9% 13.4% 39.1% 57.1% 30% 43.4% 40% 49.6% 32.7% 98.7% 99.8% 72% 99.6%

Physiochemical characteristic of refinery wastewater before and after the phytoremediation with duckweed The result showed that water temperature was ranged between 24C and 29 C (Fig 1), which was within temperature tolerance limit for duckweed growth as mentioned by Culley et al. [18] who found that the upper temperature tolerance limit for duckweed growth was around 34C, and that may be due to Lemna minor was one of floating aquatic plants which cover the surface ofthe water andforman insulating layerprevent light penetration, leading to reducewater temperatures [19]. The Lemnaceae family is capable to survive in fairly wide pH range In nature, plant species of Lemenaceae can tolerate pH of 3.5 to 10, The optimal pH range 4.5 to 7.5 [20] also [21] has shown that some floating plants such as Eichhornia crassipes able to reduced pH value of domestic wastewater from 8.2 to7.1 during 96 daysand this was agreed with our result ,Fig (1) showed that the pH values was within the range 7.7- 8.9, that attributed to pH value affected by buffering factor such as CO2,HCO3because some plants have ability to served as buffering throughout release and take ionsfor the completion of balance within environment[1]. Total alkalinity showed a continuous gradual removal by increasing time (Fig1). Values decreased from 95 mg.L-1until reached 56 mg.L-1after 30 days, significant correlation (r =0.91, p<0.05) between pH and total alkalinity were observed in this study. Turbidity was reduced Fig(1) and this may be attributed to decrease the concentration of suspended material because of settlement and adsorption on aquarium glass and this was shown in statistical analysis , as it recorded significant correlation between suspended solids and turbidity (r = 0.94; p0.05). As evident from Table 1, total suspended solids (TSS) values decreased by increasing treatment periods, reaching minimum concentration of 13 mg.L -1after 30 days, which corroborates the findings of [22] regarding discharged duckweed treatment system in Halisahar ,Likewise [23] record a clear reduction in resuspension of sediment in Taiho lake during 41 days which covered by floating aquatic plants , and this result agreed with the study of [24]. Data in Figure (1) revealed that total dissolved solids (TDS) recorded their minimum values of 2110 mg.L-1after 30 days treatment ,this decreasing due to the plant capacity to take some organic and in organic ions, its may be absorbed high concentration of sodium ion during growth [25]. Sooknah and Wilkie [26] explained

that three types of floating plants used to treat diary wastewater reduced electrical conductivity from 2510 S.cm-1to 733S.cm-1. Results in Table (1) showed that sulfate concentration recorded 13.4% as reduction percentage during 30 days ,that may be due to plant ability to absorb different types of pollutants and accumulated in their tissues [27], On the other side Hydrogen sulfide concentrations showed a continuous gradual removal by increasing time, where its values decreased from 23 mg.L-1 until reaching 14mg.L-1 after 30 days (Fig2) , Which may be attributed to reduction in sulphate concentration that confirmed by the results of statistical analysis which reported a positive correlation between the concentration of hydrogen sulfide and sulphate (r = 0.921,p0.05). Patel and Kanungo [2] stated that in the experimental use of macrophytes for removal of nutrients from domestic wastewater, Lemna gibba removed a great amount of nitrate .In agreement with the present study, the results demonstrated that the duckweed efficiently removed nitrate from wastewater and incorporated into its biomass ,where its concentration reached 9 g.L-1in the last week (Fig 2),that was attributed to the plant's capacity to provide suitable conditions for nitrate reduction as a result of co existence with microorganism which play an important role in conversion of nitrogen or plants direct uptaking which used a large amounts of nitrogen compounds such as NO3, NH4 during growth period

Nayyef M. Azeez and Amal A. Sabbar, 2012. Efficiency of Duckweed (Lemna minor L.) in Phytotreatment of Wastewater Pollutants from Basrah Oil Refinery. Journal of Applied Phytotechnology in Environmental Sanitation, 1 (4): 163-172. 167

0 1 0

0 50 100 week 1 week 2 week 3 week 4 Total Alkalinity (mgCaCO3-1)

Control Lemna

20 30 T e

mpera tur (c)

control

Le m na

Aqu ariu m

7.5 8 8.5 9

Aquarium

0 5 10

wee wee p k1 k3 H wee wee k2 k4 Lem 1 wee na 5 k2 Aqu 2 wee ariu 0 k3 m 2 wee 0 5 k4 100 w TSS 0 e (m 200 e g.L0 k 1) 300 Cont 1 0
rol

Control Lemna

Aquarium

400 0 500 0 wee k1 wee k2 wee k3 wee k4

TD S (m g.L1)
Cont rol
Lemna

Aqu ariu m

0 5 10

15 20 25 w ee k 1 w ee k 2

week 3 week 4 Turbidty(NTU)

Control Lemna

Aquarium

Landolt and Kandler [29] reported th at Lemna sp. requires high phosphor

Fig. 1: Weekly changes of some refinery wastewater properties (Water temperature, pH, Alkalinity, TSS, TDS and Turbidity). Landlot and Kandeler [29] reported that Lemna sp. requires high Phosphorus concentration to grow in water. Perniel et al. [30] also found that Lemna minor monoculture consistently removed the largest amount of ammonia and phosphorus from storm water in 8 weeks. During the present experiment, Lemna minor used phosphate for growth and reduced its concentration until reaching 30% in the end of experiment that attributed to absorption and adsorption or direct up taking by plant [26]. Koner and Vermaat [31] also established that Lemna gibba and microorganism coexist with it reduced 75% of phosphate and plants used 52% for growth process and this agreed with study of [32] [33]. Nayyef M. Azeez and Amal A. Sabbar, 2012. Efficiency of
Duckweed (Lemna minor L.) in Phytotreatment of Wastewater Pollutants from Basrah Oil Refinery. Journal of Applied Phytotechnology in Environmental Sanitation, 1 (4): 163-172. 168

130 0 140 0 0 5 1 0

15 00 16 00 15 20 25

1700 Sulpha te (mg.L1) Nitrate( ug.L-1) Control Lemna

C o nt ro l

Lem na Aqu ariu m

0 5 10 15 20 2 4 6 8

A q u

ariu m
0

25 wee k1 wee k2 wee k1 wee k2

wee k3 wee k4 wee k3 wee k4

sulfid (mg.L-1) Control Lemna

Aquarium
Phosphate(g.L-1) Control Lemna

Aquarium

Fig. 2: Weekly changes of some refinery wastewater properties (Sulphate ,Hydrogen Sulfide, Nitrate, phosphate) Lemna minor verified its ability to reduction oil and grease concentration, until it reached 43% that may be attributed to plant capability to taking petroleum hydrocarbons , which represent one of the components of oil and grease and stored it in their tissues because hydrocarbons are lipophilic pollutants [34] as well as microbial degradation , and this agreed with [35] who confirmed that the wetland treatment systems are effective in removing pollutants by microbial degradation or direct up taking by plants. The result showed a Significant correlation between oil and grease and phenol (r = 0.92,p0.05) that may be one of causes of phenol reduction as well as plant ability to take phenol and stored in the vacuoles within cell wall as a chemical compound known 2,4-Dichloro phenyl B-D-glucopyranoside [36]. Biochemical oxygen demand (BOD), chemical oxygen demand (COD), showed a gradual removal by prolonged treatment periods (Fig. 3). Data revealed that duckweed effectively reduced BOD by 49.6% (reduced from 27.6 mg.L -1reached 13.9 mg.L-1), COD by 32.7 %( reduced from 480 mg.L-1to 323 mg.L-1). In agreement with the present research, Oron et al [37] mentioned that the duckweed contribution for the removal of organic material is due to their ability to direct use of simple organic compounds, as well as the microbial degradation processes of organic material [38]. Zimmo et al [38] found that BOD removal efficiency was higher in duckweed based ponds than in algae based ponds. Nayyef M. Azeez and
Amal A. Sabbar, 2012. Efficiency of Duckweed (Lemna minor L.) in Phytotreatment of Wastewater Pollutants from Basrah Oil Refinery. Journal of Applied Phytotechnology in Environmental Sanitation, 1 (4): 163-172. 169

0 5 1 0 0 5 1 0 1 5 2 0 2 5 3 0 w e e k 1 w e e k 2 w e e k 3 w e e k 4 B O D 5 ( m g .L
1)

15 20 wee k1 100 200 300 400 500 CO D (m g.L1) Con trol Lem na

wee k2 wee k3

wee k3 Oil &G rea

se (m g.L1)

C on tr ol

Lemna

Aquariu m
0 0.1

0. 2 0. 3

0. 4 0. 5

0.6 Phenols (mg.L-1) Control Lemna

Aquarium

Aqu ariu m

C o nt ro l L e m n a

A q u a ri u m
0

Korner et al. [39] mentioned that duckweed significantly enhanced COD removal in shallow batch systems. Pandey [22] reported that COD removal was in the range of 70% to 80% in the discharged duckweed treatment system at Halisahar. Fig. 3: Weekly changes of some refinery wastewater properties (Oil andGrease, Phenol, BOD5, COD). Figure 4 showed that Lead (Pb),Copper (Cu),Cadimum (Cd) and Zinc (Zn) reached their minimum concentrations of 0.2,0.02,0.02and 12 mg.L , respectively after 30 days, with a reduction percentage of 98.7%, 99.8%, 99.6% and 72%, respectively, that was the highest rates of reduction compared with other pollutants (Table 1) ,and this due to a plant's ability to absorb metals and accumulated in their tissues [40] [41]. Kara et al [19] referred to the aquatic plants have the ability to accumulate essential metals for their growth and development and these metals include iron, manganese, zinc and copper. Khellaf and Zerdaoui [42] have proven through a laboratory experiment the capacity of Lemna minor to tolerant high concentrations of copper, cadmium, nickel, zinc, and the results of this study agreed with the results of Other studies in terms of the capacity of aquatic plants on the accumulation of heavy metals and used it as phytoremedator and monitors of heavy metals pollution such as [43] [44] [45]. Nayyef M. Azeez
-1

and Amal A. Sabbar, 2012. Efficiency of Duckweed (Lemna minor L.) in Phytotreatment of Wastewater Pollutants from Basrah Oil Refinery. [28].

3. Results 3.1. Crowding Growth rates of L. minor decreasing with increasing density (Fig. 1). The highest growth rate of 0.30 d_1 was observed at a biomass of 10 g DW m_2. Remarkably, the lowest biomass (5 g DW m_2) had a significantly lower growth rate of 0.23 d_1 (Mann Whitney P < 0.008). The highest densities of 180 g DWm_2 and 915 g DWm_2 showed a negative net growth rate of _0.02 and _0.08 d_1, respectively. These experimental data were used to calibrate the model, assuming no nutrient limitation. The maximum growth rate, r, was S.M. Driever et al. / Aquatic Botany 81 (2005) 245251 247 obtained by extrapolation and was corrected for the optimum temperature of 26 8C (0.41 d_1). The half saturation constant, hB, (26 g DW m_2) was calculated from the equilibrium biomass. The resulting model was used to describe the net relative growth for different biomasses (Fig. 1). 3.2. Field Fig. 2 shows the field data and model prediction for each ditch. Growth rate, r, was fitted for each ditch separately assuming different limiting factors for each ditch. With an increase in air temperature (day 4247; Fig. 2), the duckweed biomass in all ditches increased, as the model predicted. For fN (N) and fP (P), the assumption was made that nutrients were not limiting (Table 1), supported by the fact that L. minor was present at the start of the sampling period. 4. Discussion Using a combination of a laboratory-scale experiment and a simple model, we obtained insight in the growth dynamics of L. minor in the field. The laboratory experiment showed the relation between crowding and growth rate. We were able to model field growth in Dutch ditches, suggesting that we captured the main processes in the model. Yet several processes were neglected or oversimplified in the model. For instance, the

model assumed that the plants were homogenous distributed over the ditch. In reality, this is usually not true because of the influence of birds and wind (Duffield and Edwards, 1981). Indeed the distribution of L. minor in the sampled ditches was very heterogeneous, also within the three strata. In the model the loss, l, was assumed to be constant for all ditches (0.05), which equals a lifetime of about 35 days, a realistic value for individual fronds (Landolt, 1986). The loss included respiration, grazing and mortality implicitly. It was neglected that respiration is 248 S.M. Driever et al. / Aquatic
Botany 81 (2005) 245251 Fig. 1. Growth rate as a function of the initial biomass of L. minor (g DW m_2). Dots indicate the measured growth rate. The solid line represents the curve described by the model at a constant temperature of 23 8C. strongly dependent on temperature. Differences between ditches in grazing

pressure by birds and invertebrates were not known and could not be accounted for in the mortality rates. The influences of temperature, crowding and decomposition on mortality are unknown. In our model, we used air temperature instead of water temperature. It is known that temperature can vary widely between water, air and within floating mats (Dale and Gillespie, 1976), so we assumed that the temperature of the fronds in the upper layers is better described by the maximum air temperature than by the water temperature. Despite these simplifications, the negative effect of crowding on growth rate was accurately described in the model. The density dependent reduction of growth rate was
S.M. Driever et al. / Aquatic Botany 81 (2005) 245251 249 Fig. 2. Biomass (g DW m_2) of L. minor in three ditches and maximum air temperature (8C) from day 0 (17 April 2003) to day 63 (18 June 2003). Field data are indicated as black squares, open triangles and black dots for Opheusden, Zetten and Sinderhoeve, respectively with standard errors (N = 10). Model output is shown as a dashed and dotted line, dashed line and solid line for Opheusden, Zetten and Sinderhoeve, respectively. Maximum air temperature is indicated by the dotted line. Table 1 Characteristics of the ditches Characteristic Sinderhoeve Zetten Opheusden Length (m) 40.0 45.5 48.0 Width (m) 3.0 1.83.0 0.7 Depth (m) 1.6 0.5 0.4 Sediment Sand Clay Clay NNH4 + (mg l_1) 0.07 (_0.03) 0.08 (_0.01) 0.14(_0.10) NNO3 _ (mg l_1) 0.05 (_0.05) 0.04 (_0.03) 0.99(_1.38) PPO4 3_ (mg l_1) 0.42(_0.56) 0.51 (_0.68) 1.62 (_0.65) The nutrient levels are mean values for the last three sampling dates.

clearly not linear, like in logistic growth, but it decreased approximately logarithmically. It could well be described as a Monod-type growth limitation by biomass. The mechanism causing this density dependency is not completely clear. In a dense mat, fronds are piled up in several layers. One could expect that in such case layers can be subdivided into two parts: an upper part with nutrient limitation and a lower part with light limitation (Clatworthy and Harper, 1962) or CO2 limitation. However, since we observed only decompositon in the lower parts, it seems not very likely that nutrient limitation of the upper part was a major factor. The laboratory experiments suggested that there is an inverse density dependence at low densities. In waters partly covered by L. minor (biomass below 9.5 g DW m_2), the relative growth rate increased significantly with increasing Lemna density. A likely explanation for this effect seems to be the increase of temperature in a closed deck, due to solar radiation (Dale and Gillespie, 1976). In this way, there is facilitation at low densities once there is full coverage, i.e. the higher temperature within the mat will increase the relative growth rate. For other floating plants, a similar facilitation may occur. For instance, the tropical floating Salvenia molesta is known to increase temperature locally (Room and Kerr, 1983). Acknowledgements We thank Gertie Arts and Dick Belgers from Alterra

(Wageningen, NL) for useful suggestions and discussions and Frank van Herpen, who helped a great deal with the practical work.

Conclusion Metals and metalloids are removed to varying degrees depending on the constructed wetland type. This review confirms previous findings by Kropfelova et al. (2009) that the removal rates of metals and metalloids in constructed wetlands decrease as follows: Hg > Mn > Fe Cd > Pb Cr > Zn Cu > Al > Ni > As (Table 3). Generally, the removal rate is higher than 70% particularly in systems dominated by P. australis, P. canadensis, Potamogeton spp., Acorus spp., L. salicaria, I. pseudacorus, Schoenoplectus spp., E.crassipes, H. reticulatum, C. demersum, and P. stratiotes. These results may be biased by the dominance of certain species, e.g. Pragmites australis, not necessarily because they are the best choice, but because they are widely available and easy to grow. Few, if any, systems have been planted with particular species because of a deliberate choice regarding their efficacy. The field of phytoremediation needs to set up standardized protocols for validation and better quantify the roles of substrate, microorganisms, macrophytes, and their interactions in constructed wetlands. In order to optimize designs for constructed wetlands for metal removal one of the first stages should be the selection of the plant species and/or ecotype selection, in conjunction with a substrate suitable for plant growth and a balanced supply of organic matter and pH buffering capacity. Studies to underpin such selection can be carried out in a micro- or mesocosm scale experiment. These would not aim at exactly mimicking natural ecosystems but allow visualization of plant responses to the conditions prevailing in constructed wetlands, such as metal concentrations, redox conditions and pH. Experiments aimed at testing metal removal should use the following criteria: (1) measurement of system efficiency via the assessment of a mass balance of water and pollutants (for example by measuring concentration differences between inlet and outlet, assuring the system is hydrologically isolated, and taking into account effects of evapo-transpiration), (2) each treatment unit must be properly replicated, and (3) an unplanted control must be included. Concerning the roles of the substrate, studies using so-called Bio-Racks, which are completely soil free, offer new avenues for research (Valipour et al., 2009). Proper experimental design is often neglected in phytoremediation studies, but is essential because it makes it possible to accurately quantify the effects of the variousexperimental conditions, including the choice of plants and substrates. Adoption of the relative treatment efficiency index (RTEI) proposed here would achieve quantification and standardization of these various conditions and their effects. Acknowledgements This work was supported by Axa foundation (PhD grant of L. Marchand) and ADEME, Department Polluted Site and Soils, Angers, France.

Results and Discussion

The system has been operated as advanced treatment system. The results relate only to two sampling; at the influent and effluent. The system treatment efficiency was high. The removal efficiency of wastewater between the two reactors was similar. Both municipal and industrial wastewater showed pH in the influent 7.20.2 and at the final effluent around 8.00.2. Water depth was 8 cm which is optimal level for the systems. Although there are still different ideas about the temprature requirements for duckweed growth, the production of duckweed will decrease when the temprature is below 17oC or above 35oC (Smith and Maolyawati, 1998). Both for municipal and industrial wastewater temprature was 210.5oC. COD removal: DWWT is designed on the basis of volumetric COD loading due to possible anaerobic process underneath the duckweed mat. Mandys experiment proved that DWWT ponds
Time (Day)
2468

TN (mg/L)
0 10 20 30 40 50 Municipal Wastewater Industrial Wastewater

Fig. 3: Influent and effluent TN concentration of municipal and industrial wastewater

3 Results and Discussion Cr and Co from the water were very efficiently removed by Pistia stratiotes (Fig. 1). It is clear from the figure that Cr was almost completely removed in 48 hours and non detectible level was present after 72 hours. Similar results were observed with the removal of Co from water. The present finding is in consonant with other findings and is supported by literature.
Fig. 1 Concurrent removal of Cr and Co by Pistia stratiotes Fig. 2 Removal of Cr and Co by Pistia stratiotes

When the removal of heavy metals Cr and Co from water was done separately by Pistia stratiotes then different results were observed (Fig. 2). It can be seen clearly that almost all Cr is removed by Pistia stratiotes in 48 hours and there was negligible amount of Cr in water after 48 hours which cannot be detected. However, Co was not completely removed in 4 days. The main reason for this may be the toxicity of Co to Pistia stratiotes.
137 Proceedings of the International Academy of Ecology and Environmental Sciences, 2012, 2(2):136-138 IAEES www.iaees.org Table 1 Percent age removal of heavy metals by Pistia stratiotes Days % age removal of heavy metals by Pistia stratiotes Cr Co 100 2 30 70 3 100 85 4 100 86

It can observe from the Table 1 that percentage removal efficiency of Pistia stratiotes for the two toxic heavy metals differs. The capacity of Pistia stratiotes for Cr removal is high as compared to Co from water. Other researches also have similar findings and the reasons are also similar i.e. high toxicity of Co as compared to Cr for Pistia stratiotes.

It can be concluded from the present study that aquatic macrophytes Pistia stratiotes can be used for phytoremediation of water bodies polluted with toxic metals Cr and Co in a sustainable way. Acknowledgments Authors are thankful to the Department of Botany, School of Life Sciences, Guru Ghasidas Vishwavidyalaya, C.G. for providing the various facilities. References

Results and discussion 3.1. Arsenic and its speciation in mine tailing waters Arsenic concentration in tailing waters and that accumulated in L. gibba L. at different sampling points are summarised in Table 1. The results revealed that on average the tailing waters in samples from both Lengenfeld and Neuensalz-Mechelgrun had significantly higher arsenic content than did waters from the reference site. Large standard deviations are due to temporal variations in Neuensalz- Mechelgrun speciation calculation with geochemical modelling software, PhreeqC+ 2.8.0.0 Alpha version (Parkhurst and Appelo, 1999), which indicate that in ambient systems, arsenic occurred as arsenate (HnAsO4 3 _ n) or arsenite (HnAsO3 3 _ n) complexes. Over the pH value range of 6.08.1 found in tailing waters, the predominant species of As (V) was H2AsO4 _ and that of As (III) was H3AsO3, of which the reduced form is the more mobile. The content of arsenic at the Lengenfeld sampling points differed significantly (data not shown). Due to the short sampling period, the temporal variation was not significant. Differences in arsenic water content from sampling points in Neuensalz-Mechelgrun were both temporally and spatially significant. The content of arsenic was the lowest in water between July and August 2002. The following factors might have contributed to this: the year 2002 had the highest rainfall recorded in the area for a decade, and dilution took place; high temperatures were also recorded during the period. This was also the productive period for macrophytes in the tailing ponds (data not shown). Aquatic plants play a key role in remobilizing arsenic (Hallberg and Johnson, 2003). Ferrous iron oxidised to ferric form, which leads to precipitation of iron oxyhydroxides in the rhizosphere (iron

plaque) (Godowski et al., 1995; Arienzo et al., 2002). As a result, there is a decreasing concentration gradient of dissolved iron towards the plant roots. The iron oxyhydroxides consequently bind arsenic (Robinson et al., 2003b). Table 1 Arsenic concentrations in water and L. gibba samples from reference sites, Lengenfeld and Neuensalz-Mechelgrun Site Sample Point [As] in surface mine water (Ag l_ 1) As accumulation (mg kg_ 1 dry biomass) Reference RW1 7.55F3.40*** 77.83F35.05*** RW2 3.04F2.13*** 29.71F21.54*** RW3 5.61F3.11*** 44.98F28.91*** Lengenfeld LTW1 47.01F11.2** 519.61F64.75*** LTW2 62.96F20.77** 937.62F253.03* LTW3 265.42F228.31 1543.22F238.29 Neuensalz- MTW1 106.71F30.46 1296.90F241.17 Mechelgrun MTW2 109.84F72.31 1107.78F312.87

MTW3 167.92F90.97 1447.96F522.03 MTW4 64.03F21.11** 1035.74F381.43 MTW5 57.50F21.71** 707.26F310.09 MTW6 53.48F22.98** 687.43F395.75 Sampling periods: reference site and Lengenfeld between August and October 2001, n = 31; Neuensalz-Mechelgrun between February and December 2002, and n = 108. Values are means of four repeated measuresFstandard deviation. * p = 0.05. **p =0.01. ***p = 0.001. 84 M. Mkandawire, E.G. Dudel / Science of the Total Environment 336 (2005) 8189 3.2. Arsenic accumulation in L. gibba samples from the field 3.2.1. Influence of the milieu arsenic concentration Arsenic accumulation in L. gibba L. from the tailing water ranged from 0.54 to 110.8 mg kg_ 1 in fresh mass, and 61.7 to 1966.48 mg kg_ 1 in dry biomass. Mean accumulations are presented in Table 1. Effect of arsenic on L. gibba growth and yield has been studied (Mkandawire et al.,in press, 2004) Uptake and accumulation of arsenic in L. gibba in both Lengenfeld and NeuensalzMechelgrun increased with arsenic concentration in the milieu tailing waters fitting in a sigmoid regression (r2 = 0.93 in Lengenfeld, r2 = 0.89 in Neuensalz- Mechelgrun; p < 0.05 at 95% confidence interval; Fig. 2). The accumulation pattern in the fronds suggests that the water uptake by L. gibba induced arsenic migration via mass flow into the frond until the fronds were saturated with arsenic. Thus, arsenic might have entered the fronds, either via the symplastic or via the apoplastic pathways where some active or passive filtering might occur (Blinda et al., 1997; Parker and Pedler, 1997; Samardakiewicz and Wozny, 2000). Mkandawire et al., in press, 2004 found that under laboratory condition L. gibba tolerated arsenic toxicity in the range of 10500 Ag l_ 1 and suddenly shows toxicity. Robinson et al. (2003b) showed that plants that tolerate high concentrations of toxic elements have an active uptake mechanism for the nonessential elements. Such plants have constant u over a narrow concentration range (Robinson et al., 2003b) similar to the behaviour demonstrated by L. gibba in the laboratory investigation (see below). As the element concentration increases in the milieu, a sudden increase in the plant element concentration occurs. This happens when the uptake control mechanisms break down, due to overload of the regulatory mechanism (Zhao et al., 2002). When this phenomenon occurs, the plants show toxicity symptoms and biomass production is reduced (Wenzl et al., 2001). The current result and results from the previous studies on arsenic toxicity to L. gibba(Mkandawire et al., in press, 2004) attest to the general phenomenon. 3.2.2. Influence of milieu phosphorus concentration Fig. 3 shows that accumulation of arsenic in L. gibba at both sites correlated negatively to the P content of the tailing waters (r2 = 0.916; at p = 0.05 and 95% confidence interval). This phenomenon has been observed in other plant species too (Creger and Peryea, 1994; Carbonell-Barrachina et al., 1998; Peryea, 1998; Cao et al., 2003). Phosphates have long been reported to suppress plant uptake of arsenate (Pickering et al., 2000). For instance, phosphorus addition resulted in a reduction of arsenic uptake by 5572% over the control in Indian mustard (Brassica juncea) (Pickering et al., 2000). Most of the studies which

reported this relationship were conducted in either amended soil or hydroponic culture experiments in laboratory (e.g. Bissen and Frimmel, 2000; Jackson and Bertsch, 2001; Cao et al., 2003). This study reports results that demonstrate the behaviour in the natural aquatic system. However, the effect of phosphates on arsenate uptake seems Fig. 2. Accumulation of arsenic in L. gibba in relation to arsenic concentration in milieu tailing waters.M. Mkandawire, E.G. Dudel / Science of the Total Environment 336 (2005) 8189 85 to differ from one plant species to another, and the medium used. A few variations have been reported in the literature where phosphorus increases arsenic uptake by plant in soils (e.g. Marin et al., 1993; Elvira et al., 2003). Much as L. gibba has demonstrated the relationship between arsenic uptake and phosphorus, it should not be taken for granted that most macrophytes in the tailing ponds in the sites behave in the same way. Hence, investigations of ther macrophyte species in the tailing waters are necessary. 3.3. Accumulation of arsenic in controlled phosphate Supply At the completion of the laboratory trials, the arsenic concentrations in all the plants had reached equilibrium, indicated by a negligible decrease in the solution arsenic concentrations. Fig. 4 shows that the levels for arsenic in the fronds of L. gibba at the end of the laboratory experiments ranged from 0.02 (in 40 mg l_ 1 PO4 3 _) to 106 mg kg_ 1 dry biomass (in 0.0136 mg l_ 1 PO4 3 _) and 0.00187.79 mg kg_ 1 Fig. 4. Accumulation of arsenic in L. gibba under variable initial PO4 _ and AsO4 _ concentrations in nutrient solution. The values are means of four replications and the error bars are standard deviations. Fig. 3. Relationship between arsenic accumulation in L. gibba to concentration in tailing water. 86 M. Mkandawire, E.G. Dudel / Science of the Total Environment 336 (2005) 81 89 fresh weight. Toxicity behaviour of arsenic in the\ concentration range in nutrient solution has been reported in Mkandawire et al. (2004, in press). Arsenic concentrations in L. gibba L. dry biomass have a strong positive correlation with the arsenic concentrations of the milieu water (r2 = 0.96; p < 0.05 at 95% confidence interval), as demonstrated in the field samples (see above). The arsenic concentration ratio of plant/solution increased exponentially as the arsenic concentration in the solution increased. These results resemble the work of Robinson et al. (2003a) who described this type of increase to be associated with active exclusion; viz. at low concentrations, plants are able to exclude the toxic element, and the barrier breaks down at higher concentrations. The accumulation of arsenic in L. gibba decreased significantly with increasing phosphate concentration at all six levels of test arsenic concentration in the nutrient solution. These results agree with earlier results by Mkandawire et al. (2004), Meharg and Macnair (1990), and Robinson et al. (2003a). Speciation modelling with PhreeqC predicted increasing desorption due to similarities in thermodynamic and kinetic properties between AsO4 3 _ and PO4 3 _. AsO4 3 _ is a sorption analogy of PO4 3 _ and, PO4 3 _ competes with AsO4 3_ for sorption sites (Mkandawire et al., 2004). Therefore, it should be expected that more arsenic should be desorbed in the solution with increasing phosphate. In spite of this, the uptake of arsenic by L. gibba was affected. Because arsenate is the dominant form of arsenic under aerobic conditions and is an analogue of phosphate, they compete for the

same uptake carriers in the plasmalemma (Smith and Read, 1997). Thus, arsenate uptake in L. gibba is obviously suppressed by phosphate, as shown by the much lower arsenic uptake rates in both laboratory and mine tailing plants at high P supply, supporting the view that arsenate uptake is mediated by phosphate transporters.3.4. Arsenic interaction with L. gibba The results of arsenic accumulation by L. gibba in the field samples and in laboratory experiments showed an increase in the arsenic accumulation with increasing arsenic concentration in the milieu. This indicates that the uptake of arsenic by L. gibba occurs through active exclusion mechanisms. The results also showed that in the field, arsenic accumulation correlated negatively to increasing P concentration in the milieu and reduced arsenic accumulation in the laboratory with PO4 3 _ concentration. This has been related to As (V) uptake through PO4 3 _ uptake pathway. A few works argued that active exclusion is not consistent with arsenic uptake via the phosphate uptake pathway (Smith and Read, 1997; Robinson et al., 2003a). Our laboratory results show that L. gibba exhibited arsenic uptake through active exclusion and at the same time phosphate uptake. This suggests that it is important to consider arsenic concentrations together with phosphorus or, indeed, with other elements like sulphur, manganese, iron, etc., that may affect speciation in the milieu when investigating arsenic uptake by L. gibba. 3.5. Comparison of accumulation in laboratory and field samples Generally, the arsenic concentrations in L. gibba samples from tailing waters were approximately twofold less than similar arsenic concentrations in solutions. This is because L. gibba in the surface mine waters was exposed to arsenic for an undefined period, whereas samples used in laboratory trials were certified to contain a belowdetection amount ofarsenic and were exposed to arsenic only for 21 days. However, the mean bioaccumulation coefficients for the field L. gibba samples were twofold higher in the laboratory than in the field. This might be attributed to the fact that nutrient solutions had limited and defined amount of ions, whereas tailing waters had an undefined and unlimited amount of ions that definitely competed with arsenic for uptake sites. The ease with which elements enter the plants symplast is affected by the amount of ions present in water (Keltjens and Van Beusichem, 1998; Samardakiewicz and Wozny, 2000), as well as by other factors such as pH, Eh, and temperature (Del Castilho and Chardon, 1995). These factors determine chemical speciation in the milieu. 3.6. Arsenic and uranium accumulation in L. gibba The mean background concentrations of uranium in tailing waters found in our earlier studies were 186.0F56.31 and 277.01F10.18 Ag l_ 1 in Lengenfeld and Neuensalz-Mechelgrun, respectively (unpublished data). The mean uranium concentrations were M. Mkandawire, E.G. Dudel / Science of the Total Environment 336 (2005) 8189 87 significantly higher than the arsenic background concentrations at both sites. Uranium accumulations in L. gibba were within the ranges of 514.47F83.71 and 612.36F143.6 mg kg_ 1 U dry biomass in Lengenfeld and Neuensalz-Mechelgrun, respectively (unpublished data). Comparison of the bioaccumulation coefficients for uranium found in an earlier study (Mkandawire et al., in press) with those for arsenic revealed that the values for uranium were significantly lower than the values for arsenic. Accumulation of arsenic and uranium in L. gibba increased with an

increase in the milieu concentration but differed in relation to PO4 3 _. Previous studies showed that uranium accumulation increased within a short range with PO4 3 _ concentration before levelling off. Speciation calculation predicted precipitation of uranyl phosphates at higher concentrations of both UO2 2+ and PO4 3 _ in the aquatic system. Hence, much as phosphate increases U uptake, U bioavailability is generally reduced (Mkandawire and Dudel, 2002). 4. Conclusions The results reveal that arsenic accumulates considerably in L. gibba growing in tailing ponds of abandoned uranium mines at Lengenfeld and Neuensalz- Mechelgrun. This indirectly proves that arsenic contamination exists in abandoned uranium mine sites. The accumulation is directly influenced by the milieu concentration of arsenic and phosphorus. If the accumulation in other macrophytes is similar to that observed in L. gibba, arsenic could transfer more easily from contaminated waters to higher trophic levels than uranium. Thus, arsenic may pose more risk than uranium. The results also suggest that L.gibba should be used as an indicator for arsenic and for arsenic transfer from contaminated waters to plants. The high arsenic bioaccumulation coefficients and the ability of L. gibba to reduce arsenic on average by 40.3% in the solutions show that the species has potential for arsenic phytoextraction from contaminated waters. Unfortunately, L. gibba is a small floating macrophyte that can be transported to uncontaminated areas with flowing water. Hence, investigations on the remobilisation and biomineralisation mechanisms of arsenic in L. gibba are required.Further investigations are also required on L. gibbas defence mechanism against arsenic toxicity, e.g. biomethylisation, assimilation, or compartmentalisation. The influence of fungi and other microorganisms that enhance phosphate acquisition by the host plant on L. gibba accumulation of arsenic is worth investigating because of the interaction between phosphate and arsenate. The study has also opened an area that requires further investigation to clarify whether active exclusion and the phosphate pathway uptake mechanism are simultaneously employed in Lemna species and in other organisms. Acknowledgements The German Federal Ministry of Education and Research (BMBF) supported the study under Project Grant No. 02WB0222. Arndt Weiske did ICP-MS analyses, with laboratory assistance from Karin Klinzmann and Annet Jost. Phosphate analysis with ICP-OES was done at the Landesforstprasidium of the State of Saxony, Graupa. Dmitry Tychinin (IPBBM of the Russian Academy of Sciences) read the manuscript and made valuable contributions to the language.

RESULTS AND DISCUSSION PERFORMANCE OF L. MINOR 8627 IN A CONTROLLED ENVIRONMENT

Results from the in vitro tests demonstrated that the duckweed efficiently removed nitrogen and phosphorus from the synthetic medium (SAM) and incorporated them into its biomass. Figure 1 shows nitrogen and phosphorus removal from different dilutions (100%, 75%, 50%, and 25% strength) of SAM by growing L. minor 8627 and the duckweed growth during the in vitro tests. There were clearly two phases of nutrient uptake by the duckweed for both nitrogen and phosphorus in all the tests: an initial slow uptake, followed by a rapid uptake (figs. 1a and 1b). The removal of NH4N and oPO4P from the media followed the same pattern as that of TKN and totalP, respectively (fig. 2). However, the duckweed growth experienced three phases: a lag phase, an exponential growth phase, and a linear growth phase (fig. 1c). The lag phase lasted for approximately 100 hours in all the tests, and the duckweed weight was almost constant in this phase. During the exponential growth phase, approximately from 100 to 336 hours, there were only small differences in duckweed growth rates in different dilutions of SAM. Beyond 336 hours, most nitrogen and phosphorus were removed from the media, and a linear duckweed growth was observed at about the same rate in all treatment conditions. The values of the COD, TOC, TKN, NH4N, TP, and oPO4P concentrations and pH in the control boxes without duckweed were almost constant during each in vitro test (data not shown). This indicates that biological activity other than duckweed growth can be neglected in these tests, and nitrogen and phosphorus removal from the media during the tests was solely a result of nutrient uptake by the duckweed. The apparent lag phase during the first 100 hours was most likely due to the abrupt change in growth conditions for the 1006 TRANSACTIONS OF THE ASAE TKN, mg/l 0 50 100 150 200 250 300 350 400 (a) Total P, mg/l 0 5 10 15 20 25 (b) Time, hours 0 100 200 300 400 500 600 Duckweed Dry Weight, g 0.0

0.2 0.4 0.6 0.8 1.0 (c) Initial SAM strength: 100%, 75%, 50%, 25% Lag Phase Exponential Growth Phase Linear Growth Phase Figure 1. Removal of (a) total Kjeldahl N (TKN) and (b) total P from different dilutions of a synthetic swine lagoon liquid (SAM) by growing Lemna minor 8627, and (c) the duckweed growth in a growth chamber (temperature = 23C; photon flux density = 40 _mol m2 s1; photoperiod = 16 hours/day; medium volume = 150 mL; surface area = 25.8 cm2). AmmoniumN, mg/l 0 50 100 150 200 250 300 350 400 (a) Time, hours 0 100 200 300 400 500 600 oPhosphateP, mg/l 0 5 10 15 20 25 100%, 75%, 50%, 25% (b) Initial SAM strength: Figure 2. Removal of (a) ammoniumN and (b) orthophosphateP from different dilutions of a synthetic swine lagoon liquid (SAM) by growing Lemna minor 8627 in a growth chamber (temperature = 23C; photon flux density = 40 _mol m2 s1; photoperiod = 16 hours/day; medium volume = 150 mL; surface area = 25.8 cm2). duckweed from the SH medium to SAM. For example, the majority of nitrogen in the SH medium was provided in the nitrate form, while SAM contained mostly ammonium nitrogen. The SH medium also contained organic components, such as myoinositol and vitamins, that were not present in SAM. These differences probably necessitated

physiological adjustments, and may have hindered duckweed growth. It is notable that there was a slight N and P reduction in the media during the lag period. This phenomenon indicates that the duckweed accumulated N and P in its cells without increasing its weight during the lag phase, resulting in a higher N and P contents of the duckweed. Landolt (1986) indicated that N and P contents of duckweed increase with the increase of N and P concentrations in the medium. Our observation agrees well with Landolts conclusion.After the duckweed was acclimated to the new media, it began to grow and rapidly remove nutrients from the media. Generally, higher rates of nitrogen and phosphorus uptake and duckweed growth occurred in the media with higher nutrient concentrations (fig. 1). The highest TKN uptake rate was observed in the 100% strength SAM at around 240 hours and was 0.014 mg cm2 h1 or 3.36 g m2 day1. The highest TP uptake rate was observed in the 100% strength SAM at around 192 hours and was 8.2 104 mg cm2 h1 or 0.20 g m2 day1. After the exponential growth, during which most N and P were removed from the media, duckweed growth rates in all different dilutions of SAM were similar (fig. 1c). The average duckweed growth rate during the linear phase was 1.19 g (dry) m2 h1 or 28.6 g m2 day1. NUTRIENT REMOVAL FROM SWINE LAGOON LIQUID IN OUTDOOR TANKS WITH GROWING L. MINOR 8627 Two batch experiments were conducted to study the nutrient dynamics of a duckweed based swine wastewater renovation system in outdoor tanks under natural climate conditions of Raleigh, North Carolina, in the spring and fall of 1999. Different dilutions (50%, 33%, 25%, and 20%) of the swine lagoon liquid were used as media for growing duckweed, and the average N and P concentrations and other characteristics of each dilution are listed in table 2. Generally, the P:N ratio in the initial media in the fall experiment (0.47 to 0.62) was much higher than that in the spring experiment (0.30 to 0.34). Orthophosphate, in particular, was nearly nonexistent in the spring wastewater, while it was relatively high in the fall lagoon liquid (table 2).The pH in the testing cells of outdoor tanks with diluted swine lagoon liquid was fairly stable within a range of 7.1 to 8.1 during both experiments, indicating that the swine lagoon liquid had a strong buffering capacity. This is very different from SAM to which NaOH had to be added to maintain the neutral pH. The strong buffering capacity of the swine lagoon liquid is very important for maintaining duckweed growth because growth tends to lower the pH value of the media quickly without a buffer (from approximately 7.0 to approximately 5.0 within 24 hours). Significant COD and TOC reduction was observed in both experiments. COD and TOC were reduced by 51% to 74% and 61% to 67%, respectively, in the spring experiment, and by 62% to 76% and 52% to 73%, respectively, in the fall experiment. This indicates that bacteria were quite active in the testing cells during the experiments.Linear or close to linear rates of N and P removal and duckweed growth were observed in both field experiments (figs. 3 and 4), unlike in the controlled environment described in the preceding section. Considerable amounts of nutrients were still present in the more concentrated waste Vol. 45(4): 10031010 1007

Table 2. Initial conditions of diluted swine lagoon liquid used as media for growing L. minor 8627 in batch field experiments in Raleigh, North Carolina, in 1999. Nutrient Concentration (mg/L)[b] COD[b] TOC[b] Dilution of Swine Lagoon Liquid as Medium[a] TKN NH4N TP oPO4P (mg/L) (mg/L) pH[b] Spring Experiment 50% 140 83 47 0.4 783 221 7.50 33% 100 58 34 0.2 550 142 7.38 25% 73 42 24 0.1 403 102 7.25 20% 54 31 16 0.0 302 77 7.12 Fall Experiment 50% 196 112 92 43 1,058 434 7.93 33% 86 51 45 27 507 149 8.05 25% 48 40 26 21 215 57 8.00 20% 34 26 21 18 178 46 7.83 [a] The actual initial nutrient concentrations may not be exactly the same as intended because of the deviations caused in the dilution operation in the field tanks. [b] Each datum is an average of the results obtained from two replicate experimental cells. (b) TKN, mg/l 0 25 50 75 100 125 150 175 200 (a) Total P, mg/l 0 20 40 60 80 Date 5/24/99 5/31/99 6/7/99 6/14/99 6/21/99 6/28/99 7/5/99 7/12/99 7/19/99 7/26/99 Duckweed Production (dry), kg 0.0 0.2 0.4 0.6 0.8

1.0 1.2 1.4 (c) Lagoon Liquid Dilution: 50%, 33%, 25%, 20% Figure 3. Removal of (a) total Kjeldahl N (TKN) and (b) total P from different dilutions of a swine lagoon liquid by growing Lemna minor 8627, and (c) the duckweed production under natural climate conditions in spring (May 24 through July 26) 1999 in Raleigh, North Carolina. water at the end of the experiments. In the field tests, no lag phase in duckweed growth was observed, probably because the seed duckweed was acclimated to the nutritional environment of swine lagoon liquid during the greenhouse production period. Perniel et al. (1998) also suggested that an acclimation period would help duckweed adjust to drastic changes within a system. The field experiment results for N and P reduction and duckweed growth in various levels of diluted swine lagoon liquid show that the nutrient removal rates were generally higher in more concentrated wastewater, while the duckweed growth rate was higher in more diluted wastewater (table 3), which was different from the in vitro results. TKN, mg/l 0 25 50 75 100 125 150 175 200 Total P, mg/l 0 20 40 60 80 (a) Date 8/9/99 8/16/99 8/23/99 8/30/99 9/6/99 9/13/99 9/20/99 9/27/99 10/4/99 10/11/99 Duckweed Production (dry), kg 0.0 0.2 0.4 0.6 0.8 1.0 1.2 1.4 (b)

(c) Lagoon Liquid Dilution: 50%, 33%, 25%, 20% Figure 4. Removal of (a) total Kjeldahl N (TKN) and (b) total P from different dilutions of a swine lagoon liquid by growing Lemna minor 8627, and (c) the duckweed production under natural climate conditions in fall (August 9 through October 11) 1999 in Raleigh, North Carolina. The mass balance analysis presented in table 4 indicates that a large portion of removed nutrients was not taken up by the duckweed, especially in the 50% dilution of the fall experiment, which agrees with the growth data (table 3). Despite low rates of duckweed growth and nutrient uptake, the rate of overall nutrient removal was the highest in cells containing the most concentrated wastewater. This suggests that wastewater with higher nutrient concentrations may be more favorable to microbial growth than to duckweed growth. Some nitrogen was probably lost through ammonia volatilization, algal and microbial assimilation, and nitrification/ denitrification (AlNozaily et al., 2000). Bonomo et al. (1997) reported that at pH values lower than 8.0, ammonia 1008 TRANSACTIONS OF THE ASAE Table 3. Nutrient (N and P) removal and duckweed growth rates in batch experiments with L. minor 8627 growing on diluted swine lagoon liquid under the natural climate condition of Raleigh, North Carolina, in 1999. Dilution of Swine Lagoon Liquid as Nutrient Removal Rate (g m2 day1)[a] Duckweed Growth Rate[a] Medium TKN NH4N TP (g m2 day1) Spring 50% 1.24 0.59 0.26 17.6 Experiment 33% 1.09 0.60 0.29 21.3 25% 0.95 0.58 0.27 25.0 20% 0.73 0.47 0.18 28.5 Fall 50% 2.11 1.29 0.59 4.3 Experiment 33% 1.00 0.59 0.30 13.5 25% 0.70 0.61 0.25 12.7 20% 0.48 0.40 0.20 15.7 [a] Each datum is an average of the results obtained from two replicate experimental cells. volatilization in duckweed wastewater treatment systems was negligible, indicating that it could be neglected in our experiments. NH4N removal from the media followed the same pattern as TKN removal in both field experiments (figs. 5a and 6a). At the beginning of the spring experiment, there was almost no oPO4P in swine lagoon liquid (table 2), and its level fluctuated during the test (fig. 5b). This was probably due to bacterial degradation of organics in the wastewater and subsequent phosphorus uptake by the duckweed. During the fall experiment, oPO4P also fluctuated but generally decreased during the experiment (fig. 6b). In addition to duckweed uptake, phosphate removal may have also occurred by microbial assimilation, precipitation with some minerals, and adsorption onto clay and organic matter (AlNozaily et al., 2000).

Overall, duckweed production rate was much lower during the fall experiment. This is likely due to a combination of climatic factors. Both light intensity and temperature were lower during the fall experiment (fig. 7) and day length was shorter, less favorable to duckweed growth. Differences in nutrient concentrations were probably not as important because the most diluted wastewater with the highest duckweed growth in the fall experiment did not support Table 4. Mass balance of total N and P removal and the nutrient uptake by duckweed, Lemna minor 8627 from the diluted swine lagoon liquid in two batch field experiments under the natural climate condition of Raleigh, North Carolina, in 1999. Total Nutrient Removal from the Media[a] Nutrient Uptake by the Duckweed [a] Dilution of Swine Lagoon g % g % of Total Removal Liquid as Medium N P N P N P N P Spring 50% 102.4 24.9 63.1 45.7 50.3 11.9 49.1 47.8 Experiment 33% 90.5 26.8 78.0 68.0 63.0 16.5 69.6 61.6 25% 76.0 23.5 89.7 84.4 66.4 18.6 87.4 79.1 20% 57.4 15.8 91.6 85.1 51.6 14.0 90.0 88.6 Fall 50% 178.1 46.7 78.3 43.8 16.8 6.0 9.4 12.8 Experiment 33% 86.1 25.6 86.3 49.0 41.7 13.4 48.4 52.3 25% 52.1 16.9 93.6 56.0 37.3 11.8 71.6 69.8 20% 36.2 14.8 91.8 60.8 35.9 11.5 99.2 77.7 [a] Each datum is an average of the results obtained from two replicate experimental cells. AmmoniumN, mg/l 0 25 50 75 100 125 150 175 200 Date 5/24/99 5/31/99 6/7/99 6/14/99 6/21/99 6/28/99 7/5/99 7/12/99 7/19/99 7/26/99 oPhosphateP, mg/l 0 10 20 30 40 50 60 70 80 90 Lagoon Liquid Dilution: 50%, 33%, 25%, 20%

(a) (b) Figure 5. Concentrations of (a) ammoniumN and (b) orthophosphateP in diluted swine lagoon liquid for growing Lemna minor 8627 under natural climate conditions in spring (May 24 through July 26) 1999 in Raleigh, North Carolina. AmmoniumN, mg/l 0 25 50 75 100 125 150 175 200 Date 8/9/99 8/16/99 8/23/99 8/30/99 9/6/99 9/13/99 9/20/99 9/27/99 10/4/99 10/11/99 oPhosphateP, mg/l 0 10 20 30 40 50 60 70 80 90 Lagoon Liquid Dilution: 50%, 33%, 25%, 20% (a) (b) Figure 6. Concentrations of (a) ammoniumN and (b) orthophosphateP in diluted swine lagoon liquid for growing Lemna minor 8627 under natural climate conditions in fall (August 9 through October 11) 1999 in Raleigh, North Carolina. Vol. 45(4): 10031010 1009 nearly as much duckweed growth as the most concentrated wastewater, the least productive environment in the spring experiment (figs. 3c and 4c). Significant shortterm temperature fluctuations were observed in both spring and fall field experiments (fig. 7b). This is due to a relatively small amount of liquid in each tank. Normally, this kind of temperature fluctuations would not be expected in lagoons or ponds for animal wastewater treatment or storage. Clearly, duckweed yields were lower when grown in the presence of higher nutrient concentrations. This observation does not agree with the in vitro study, in which duckweed yields and nutrient concentrations showed a positive correlation up to 345 mg/L TKN and 24 mg/L TP in the 100% strength SAM. This could be due to a number of factors, such as the presence of undefined chemical components as well as other organisms such as microbes and insects in the field condition.

Furthermore, duckweed grown in vitro depended mostly on sucrose in SAM for its energy source, while fieldgrown duckweed depended solely on photosynthesis, leading to very different physiological states. Nevertheless, the highest duckweed growth rate in the outdoor experiment (28.5 g m2 day1) was very close to the duckweed growth rate in the lab test (28.6 g m2 day1). These numbers are almost twice as high as a reported value of 15 g m2 day1 in municipal wastewater (Oron et al., 1988). While both nitrogen and phosphorus uptake by the duckweed was the sole mechanism of the nutrient removal from SAM in the in vitro tests, it played only partial role in nitrogen and phosphorus removal from the swine lagoon liquid in the field experiments. Furthermore, relatively concentrated wastewater appears to inhibit growth and nutrient uptake by the duckweed under the field conditions.This effect is probably not due to the high nitrogen concentration because much higher nitrogen level in SAM is not inhibitory to duckweed growth or nutrient uptake.Reasons that cause lower duckweed growth rate in more concentrated swine lagoon liquid need to be further investigated. Effects of high phosphorus content or other components, such as other organisms, need to be determined in the future. Light Intensity, umol/m2s 0 200 400 600 800 1000 1200 Date 5/24/99 6/21/99 7/19/99 8/16/99 9/13/99 10/11/99 Temperature,oC 0 5 10 15 20 25 30 (a) (b) Figure 7. Light intensity (a) and water temperature (b) during the two field experiments on nutrient removal from a swine lagoon liquid by growing Lemna minor 8627 under natural climate conditions of spring (May 24 through July 26) and fall (August 9 through October 11) 1999 in Raleigh, North Carolina. Data reported here indicate that wastewater should be diluted to at least below 100 mg/L TKN and 50 mg/L TP to maintain rapid growth and nutrient uptake by the duckweed. In the design and operation of a duckweed pond for secondary treatment of swine wastewater, these data would be very useful to the initiation of the duckweed system. To establish a duckweed mat in the pond, diluted swine lagoon liquid should be used for initial

rapid growth of the duckweed. Once the duckweed mat is established, the duckweed would remove nutrients from the liquid and lower the nutrient concentrations in the liquid. When the nutrient concentrations are reduced to a desired level, the duckweed pond would be ready to take pretreated swine wastewater continuously. The pretreated swine wastewater usually contains high nutrient concentrations, which are not favorable for rapid duckweed growth and nutrient uptake. To avoid nutrient shock loading on the duckweed, we recommend that some of the duckweedtreated liquid be recycled and mixed with the influent wastewater to lower the influent nutrient concentrations.In the normal operation of the duckweed pond, wastewater would flow into the pond continuously or semicontinuously, and the duckweed would be harvested periodically. The duckweedtreated liquid can be used for crop irrigation. The influent wastewater flow rate (or the size of the duckweed pond) and the amount of duckweed to be harvested should be determined by the duckweed growth and nutrient uptake rates. Lemna minor 8627 has shown its capability of removing nitrogen and phosphorus from swine lagoon liquid under field conditions. Nevertheless, it is important to establish duckweed as a useful product to apply to infield swine wastewater treatment systems. Although research reports have shown that duckweed is a good source of proteins for livestock, poultry, and fish feed, as mentioned earlier in this article, a market for utilizing duckweed has yet to be developed. Economic analysis of the whole duckweed wastewater polishing system is necessary to determine the costeffectiveness of the system. CONCLUSIONS In summary, the following specific conclusions may be drawn from the work presented here: _ Lemna minor 8627 grew well on synthetic and actual swine lagoon liquid and effectively removed N and P from the wastewaters. The highest observed duckweed growth rate was close to 29 g m2 day1 in both systems. The rates of N and P uptake by the duckweed growing in the synthetic media were as high as 3.36 g m2 day1 and 0.20 g m2 day1, respectively. _ Preacclimation of the duckweed with swine lagoon liquid could prevent the lag phase of duckweed growth, allowing efficient nutrient uptake throughout a treatment period. _ The relatively concentrated wastewater inhibited the duckweed growth and nutrient uptake from swine lagoon liquid. _ High light intensity and longer periods of warm temperature during the spring experiment resulted in a higher growth rate for the duckweed on diluted swine lagoon liquid. 1010 TRANSACTIONS OF THE ASAE ACKNOWLEDGEMENT All the chemical analyses were conducted in the Environmental Analysis Laboratory (EAL) of the Biological and Agricultural Engineering Department at North Carolina State University. The authors would like to thank the staff of the EAL (Rachel Huie, Chris Hayes, and Indira Thillai) for their assistance.

3. Results and discussion

3.1. Expression of endoglucanase E1 Fifteen independent transgenic duckweed lines with stable kanamycin resistance were established and conWrmed to have genomic integration of multiple copies of the recombinant E1 construct. Only Wve of these transgenic duckweed lines (Cel25IX-1, 6, 12, 13, and 15) consistently showed signiWcant clearing zones on CMC detection plates. While some clearing was observed at 37 C, zones of hydrolysis were signiWcantly more pronounced at 70 C, indicating that the CMC-degrading activity was due to the expression of active recombinant E1 enzyme. One of the lines, Cel25IX-12, was found to have CMC-degrading activity in the growth medium supernate, suggesting that some autolysis of the plant was occurring. There did not appear to be any notable morphological diVerences between recombinant cell lines and the untransformed duckweed control. While all transgenic duckweed lines appeared normal, there were slight growth rate variations among them. However, variations in growth rate did not seem to have any correlation with levels of CMCdegrading activity or gene copy number (data not shown). Spectrophotometric CMCdegrading activity assays clearly showed that the Cel25IX- 15 line routinely had the highest level of expression and therefore was chosen for all further analysis. Growth studies with this particular transformant failed to show any diVerences in fresh weight or number of plantlets produced over a two-week period, when compared to the untransformed control (Table 1). Immunodetection of Y. Sun et al. / Bioresource Technology 98 (2007) 28662872 2869 the E1 protein produced by the transformed duckweed showed that it co-migrates with the puriWed E1-Cat protein

(Fig. 1), which has an approximate molecular weight of 42,000Da (Sakon et al., 1996). While the plasmid pCel25IX contains the sequence encoding the 72,000Da holoenzyme, cleavage within the linker region of the E1 holoenzyme has often been observed with other plant expression systems (Dai et al., 2000a,b; ZiegelhoVer et al., 2001) and is presumed to be the result of proteolytic digestion. However, it is unknown whether this degradation occurs in vivo or during processing of the protein. While both holoenzyme and truncated enzyme were observed in the same extracts for these studies, no intact E1 holoenzyme protein was detected in the transgenic duckweed extract. Since the cellulosebinding domain from E1 is capable of binding crystalline cellulose, it is possible that the holoenzyme protein was removed from the soluble protein fraction as a result of the binding to the insoluble plant matter. Binding of the cellulose- binding domain to plant cell wall fragments is commonly observed, and this may explain the absence of the holoenzyme in duckweed extracts (Shpigel et al., 2000; Tomme et al., 1998). The estimated amount of the E1-Cat protein in the duckweed extract was 0.24% of total soluble protein or 3.5 _g/g fresh duckweed and its CMC-degrading activity was 0.20 units mg1 of protein at 65 C (Table 1). The extract from wild type L. minor 8627 did not contain any protein cross-reacting with the E1 antibody (Fig. 1, lane 1) or detectable CMC degrading activity (Table 1). 3.2. Temperature and pH responses of E1 Endoglucanase activity increased with the increase of temperatures from 60 C to 80 C at the corresponding Ph value and decreased dramatically at 95 C (Fig. 2). The recombinant E1 exhibited the maximum CMC-degrading activity when pH value was about 5, which corresponded to the reported optimum pH of A. cellulolyticus E1 enzyme (Tucker et al., 1989; Himmel et al., 1994). The CMCdegrading activity dropped rapidly when pH changed from 5 to 4 at each reaction temperature. At 60 C and 70 C, the CMC-degrading activity exhibited a broad pH range. When pH changed from 5 to 7, the activity dropped only 8%. However, the CMC-degrading activity was lost by 30% at 80 C and 86% at 90 C when pH increased from 5 to 7. The pH response curves indicate that the E1 protein produced by duckweed was more sensitive to pH changes with temperatures of approximately 90 C. Table 1 Growth, CMC-degrading activity, total soluble protein, and E1-Cat amount in the protein extracts of transgenic duckweed line Cel25IX-15 and wild type L. minor 8627 a Data are means SD of two replicates. b Data are means SD of three replicates. c NDDnot detectable.Duckweed Growth CMC-degrading activitya Total soluble proteinb (mg/g fresh duckweed) E1-Cat protein Fresh weight (g)/Xask Plantlets/Xask (units/g fresh duckweed) (units/mg total soluble protein) Transgenic duckweed Cel25IX-15 0.12 0.03 57.8 0.29 0.240.012 0.20 0.057 1.337 0.257 3.5 Wild type L. minor 8627 0.07 0.01 53.0 0.97 NDc ND 1.157 0.202 ND Fig. 1. Immunodetection of endoglucanase E1 in transgenic duckweed line Cel25IX-15. Lane 1 contained 15 _l of protein extract from wild type control L. minor 8627. Lanes 35

contained 50, 20, 5 ng of E1-Cat. Lanes 69 contained 20, 15, 10, 5 _l of protein extract from transgenic duckweed L. minor line Cel25IX-15. Western blots were performed according to the instruction of ECL western blotting kit (Amersham Biosciences, Piscataway, NJ). The amounts of E1 protein were estimated using Molecular Dynamics Personal Densitometer SI (Molecular Dynamics, Sunnyvale, CA). Fig. 2. EVects of temperature and pH on E1 enzyme activity with crude cell free extracts. Duckweed protein extracts prepared using 50mM sodium citrate buVer (pH 4.8) were diluted in four volumes of 100mM phosphate-citrate buVer at pH 4, 5, 6, and 7. The CMC-degrading activity assay was then performed at temperatures between 60 and 95 C. Each data point is the meanSD of three replicates. 0 0.1 0.2 0.3 0.4 0.5 3.5 4 4.5 5 5.5 6 6.5 7 7.5 pH CMC degrading activity (units/g fresh duckweed) 60C 70C 80C 90C 95C 2870 Y. Sun et al. / Bioresource Technology 98 (2007) 28662872 The CMC-degrading activity in the duckweed extract after 6-h heating at 60 C was almost the same as the original activity, while its activity suVered a 27% reduction after 30-min heating at 70 C and remained constant during the rest of the experiment (Fig. 3). When heated at 80 C, CMC-degrading activity declined gradually to 38% of its maximum activity within one hour and almost lost all of its activity after three hours. CMC-degrading activity dropped remarkably at 90 C with no activity detected after 45min heating at 90 C. These results for the duckweed-expressed recombinant E1 protein are similar to the native E1 enzyme activity. Tucker et al. (1989) reported that the CMCdegrading activity of concentrated A. cellulolyticus growth supernatant was reduced by 15% when heated at 75 C and a complete activity loss occurred after 4-h preincubation at 90 C. 3.3. EVects of buVers and heating on E1 extraction BuVer chemicals, pH values, and extraction conditions are known to inXuence the protein extraction eYciency, and buVers with higher pH (6.57.2 or higher) are often used to extract proteins from transgenic plant tissues (Hatti-Kaul and Mattiasson, 1996). Our results show that the HEPES buVer at pH 8 extracted much more soluble proteins (Fig. 4b) and E1 protein (Fig. 4d) than citrate and acetate buVers. However, the highest CMC-degrading activity per total soluble protein was obtained with citrate buVer because signiWcantly less total soluble protein was extracted with this buVer (Fig. 4b and c). Heating dramatically

reduced the soluble proteins in the extracts (Fig. 4b), while the amount of E1 did not show much divergence between heat and no heat treatment (Fig. 4d). Heating during the extraction at 65 C for 5min denatured a large amount of soluble protein. The total amount of CMCdegrading activity in the extract increased slightly after heat treatment (Fig. 4a), possibly due to improved degradation of the cell wall. The citrate buVer aVorded the best removal of soluble proteins (Fig. 4b), which could be due to the precipitation of proteins at the low pH of citrate buVer. Fig. 4c shows that the duckweed extract prepared with citrate buVer and heat treatment provided the best enrichment of CMC-degrading activity per mg of soluble protein.Although HEPES extracted more E1 protein, the total extracted proteins also increased, which may necessitate Fig. 3. Heat stability of E1 protein in crude cell free extracts. The duckweed protein extracts were heated at diVerent temperatures for the time indicated on the graph. The CMC-degrading activity assay was performed at 80 C at pH 5. The activity was expressed as percentages of the initial activity. Each data point is the mean SD of three replicates. 0 20 40 60 80 100 120 0 50 100 150 200 250 300 350 400 Heating time (min) % E1 maximum activity 60C 70C 80C 90C Fig. 4. Extraction and expression of E1 protein in transgenic duckweed line Cel25IX-15 under diVerent extraction conditions. Total soluble protein was extracted using 50mM sodium citrate at pH 4.8 (1, 2), sodium acetate at pH 5 (3, 4) and HEPES at pH 8 (5, 6) with no heating (1, 3, 5) or heating at 65 C for 5min (2, 4, 6). After centrifuged at 0,000g for 10min at 4 C, the supernatant was used for E1 protein and enzyme assays. (a) CMCdegrading activity, expressed as units g1 fresh duckweed (SD of two replicates). (b) Total soluble protein extracted (SD of two or three replicates). (c) CMC-degrading activity, expressed as units mg1 total soluble protein (SD of two replicates). (d) Percentage of E1 extracted, expressed as % of E1 extracted by HEPES (50mM, pH 8) with heat treatment at 65 C for 5min. CMC degrading activity (unitsl/g fresh duckweed) 0.0 0.2 0.4 0.6 0.8 123456 Total soluble protein

(mg/g fresh duckweed) 0 3 6 9 12 CMC degrading activity (units/mg total soluble protein) 0.0 0.3 0.6 0.9 1.2 123456 % Maximum E1 extracted 0 25 50 75 100 Y. Sun et al. / Bioresource Technology 98 (2007) 28662872 2871 extensive puriWcation of E1 protein and result in increased cost for downstream processing. In addition, the use of HEPES buVer produces a protein extract with pH 8, which needs to be adjusted to the optimal pH of 5 for subsequent enzyme reaction. 4. Conclusions This study clearly demonstrates that the endoglucanase E1 gene from the bacterium A. cellulolyticus can be successfully expressed in duckweed L. minor 8627 without obvious observable eVects on morphology or rate of growth. The recovered E1 protein expressed in duckweed was approximately the same molecular weight of the E1-Cat and was biologically active. The E1 enzymatic activity from the transgenic duckweed retained similar properties with respect to pH, temperature, and heat stability to the native E1 holoenzyme. HEPES buVer (50mM, pH 8) provided a better recovery of total soluble protein and E1 than sodium citrate buVer (50mM, pH 4.8) and sodium acetate buVer (50mM, pH 5). Extraction in citrate buVer (50mM, pH 4.8) with heat treatment enriched the E1 protein in the duckweed extract because heating denatured a large fraction of endogenous proteins, while total CMC-degrading activity remained stable. Compared to the reported expression of E1 gene in other transgenic plants, the accumulation of E1 protein in the transgenic duckweed line Cel25IX-15 was somewhat low, especially compared to the expression of the catalytic domain of E1 in tobacco targeted to apoplast (ZiegelhoVer et al., 2001). While these results are encouraging as a Wrst step towards production of cellulase in an aquatic plant, considerable improvement will be necessary before industrial scale-up is feasible. This becomes especially obvious when one considers that only 0.2 units/mg protein, or 48 units/kg duckweed, was recovered from our best transgenic duckweed line. The CaMV promoter used in this

study may not be optimal for duckweed expression, since it is more suited for expression in dicotyledons. Targeting the protein to the apoplast or to vacuoles may signiWcantly increase the E1 expression levels in transgenic duckweed as demonstrated in previous studies using other host plants. Additionally, the cellulose-binding domain of the recombinant E1 holoenzyme could bind to insoluble plant matter during extraction, causing the 72,000Da intact enzyme to be trapped in the pellet during centrifugation while only the 42,000Da catalytic domain may be detected in the extract. It is anticipated that this duckweed protein expression system will be substantially improved with further manipulation of the promoter and targeting sequences and possibly expressing the catalytic domain alone. The addition of exoglucanase and _-glucosidase would be expected to improve the overall crystalline cellulose- degrading activity through synergism (Baker et al., 1994). These additional enzymes could be added through supplementation or ideally through the further transformation of duckweed. Acknowledgements This paper is dedicated in memory of Shelby N. Freer for his contributions to this work and for his enthusiastic desire to always lend a hand to those working in the Weld of biomass conversion. Funding from the OYce of the Biomass Program (OBP) of the United States Department of Energy is gratefully acknowledged. This work was partially supported by the United States Department of Agriculture- National Research Initiative Grant No. 9800906.