Sie sind auf Seite 1von 16

Nutrient removal from aquaculture wastewater

using a constructed wetlands system


Ying-Feng Lin
a,
*
, Shuh-Ren Jing
a
, Der-Yuan Lee
a
,
Tze-Wen Wang
b
a
Department of Environmental Engineering and Health, Chia-Nan University of Pharmacy and Science,
Tainan 717, Taiwan
b
Department of Pharmacy, Chia-Nan University of Pharmacy and Science, Tainan 717, Taiwan
Received 20 October 2000; received in revised form 15 August 2001; accepted 3 September 2001
Abstract
Nutrient removal is essential for aquaculture wastewater treatment to protect receiving waters
from eutrophication and for potential reuse of the treated water. A pilot-scale wastewater treatment
system consisting of a free water surface (FWS) and a subsurface flow (SSF) constructed wetlands
arranged in series was operated for around 8 months. The study was conducted to examine system
start-up phenomena and to evaluate system performance in removing inorganic nitrogen and
phosphate from aquaculture wastewater under various hydraulic loading rates (1.8 to 13.5 cm
day
1
). The wetlands system showed rapid start-up behaviors in which process stabilities were
achieved in the following sequence: phosphate removal in the SSF without an adaptation period,
nitrogen removal in the SSF after 1 month, nitrogen removal in the FWS after 2 to 3 months,
phosphate removal in the FWS after 3 months, and vegetation cover in both wetlands after 7 months
of operation. Nitrogen removals were excellent, with efficiencies of 86% to 98% for ammonium
nitrogen (NH
4
N) and 95% to 98% for total inorganic nitrogen (TIN). Removal efficiencies were
affected little by the hydraulic loading trials. Phosphate removal of 32% to 71% occurred, with the
efficiencies being inversely related to hydraulic loading. The FWS wetland removed most inorganic
nitrogen, whereas the SSF wetland removed phosphate at a rate equal to or even greater than the
FWS. Removal of ammonium and nitrite (effluent concentrations < 0.3 mg NH
4
N l
1
and 0.01 mg
NO
2
N l
1
) were sufficient for recycle in the aquaculture system without danger of harming the
fish. D 2002 Elsevier Science B.V. All rights reserved.
Keywords: Constructed wetlands; Aquaculture; Wastewater treatment; Nitrogen; Phosphate
0044-8486/02/$ - see front matter D 2002 Elsevier Science B.V. All rights reserved.
PII: S0044- 8486( 01) 00801- 8
*
Corresponding author. Tel.: +886-6-266-4911-304-16; fax: +886-6-266-7323.
E-mail address: yflin@ms12.hinet.net (Y.-F. Lin).
www.elsevier.com/locate/aqua-online
Aquaculture 209 (2002) 169184
1. Introduction
Aquaculture is an important industry in Taiwan and one which requires large
quantities of water. Because existing surface water sources are widely polluted, ground-
water is the main fresh water source for aquaculture. Consequently, several areas have
faced ground subsidence as a result of over-withdrawal of groundwater. In addition, the
accumulation of feed residue and fish excreta during cultivation often causes water
quality deterioration in fishponds, resulting in toxic effects to the fish. Aquaculture farm
discharges contain considerable quantities of organic matter, nitrogen, and phosphorus
and can further degrade the water quality in receiving waters. Therefore, it is apparent
that an appropriate wastewater treatment process is helpful for sustaining aquaculture
development in Taiwan.
A number of physical, chemical, and biological methods used in conventional
wastewater treatment have been applied in aquaculture systems. Solids removal is
accomplished by sedimentation, sand filtration, or mechanical filtration. Biological
processes such as submerged biofilters, trickling filters, rotating biological contactors,
and fluidized bed reactors are employed for oxidation of organic matter, nitrification, or
denitrification (Van Rijn, 1996). These above methods do help with phosphorus removal;
however, little work has been focused on aquaculture wastewater. Conventional treatment
systems have the disadvantages of sludge production, high-energy demand, and frequent
maintenance requirements.
Natural treatment systems, including constructed wetlands, have grown in popularity
for wastewater treatment since the early 1980s (Reed et al., 1995). Constructed wetlands
have been used to treat acid mine drainage, storm water runoff, municipal wastewater,
industrial wastewater, and agricultural effluent from livestock operations. Researchers
have demonstrated that treatment wetland systems can remove significant amounts of
suspended solids, organic matter, nitrogen, phosphorus, trace elements, and microorgan-
isms contained in wastewater (Kadlec and Knight, 1996). Constructed wetland systems are
characterized by the advantages of moderate capital costs, low energy consumption and
maintenance requirements, and benefits of increased wildlife habitat (International Water
Association, 2000).
Use of constructed wetlands for aquaculture wastewater treatment is increasing
(Zachritz and Jacquez, 1993; Schwartz and Boyd, 1995; Panella et al., 1999). Various
biotic and abiotic processes regulate pollutants removal in wetlands (Kadlec and Knight,
1996; Reddy and DAngelo, 1997). Microbial mineralization and transformation (e.g.,
nitrificationdenitrification) and uptake by vegetation are the major biotic processes.
Abiotic processes include precipitation, sedimentation, and substrate adsorption and may
be particularly important for phosphorus removal. In wetland systems, some removal
processes require only brief periods (e.g., the abiotic processes) in which to become
fully operative, while others can require months (e.g., microbial community establish-
ment) or years (e.g., plant litter development) to reach stability (International Water
Association, 2000). These adaptation periods may depend on antecedent soil or gravel
properties, and the initial hydrological and vegetation conditions. Much research has
been targeted at the performance results of constructed wetlands for long-term operation,
whereas little work has been conducted on start-up phenomena. For these reasons, we
Y.-F. Lin et al. / Aquaculture 209 (2002) 169184 170
set up a two-stage pilot-scale system comprised of a free water surface (FWS) and a
subsurface flow (SSF) constructed wetlands (Fig. 1). System performance was moni-
tored with respect to removal of ammonium, nitrite, nitrate, and phosphate from
aquaculture wastewater under various hydraulic loading rates ranging from 1.8 to
13.5 cm day
1
.
2. Materials and methods
2.1. Constructed wetlands system
The pilot-scale constructed wetland system was built adjacent to a 0.2-ha earthen
pond for production of milkfish (Chanos chanos). The milkfish pond is located in
Tainan County, Taiwan. The pilot-scale system consisted of a free water surface (FWS)
and a subsurface flow (SSF) constructed wetlands arranged in series (Fig. 1). The
wetlands were cast-iron vessels, each measuring 5 m by 1 m by 0.8 m (length, width,
height), and lined with impermeable plastic liners. The FWS wetland contained a 30-cm
layer of local soil (Jender silt loam, 25j37VN, 166j55VE) at bottom and 40 cm of free
water surface above the soil layer. The SSF wetland included 60 cm of river gravel
(nominal diameter 10 to 20 mm), providing a porosity of 45%, and 40 cm of subsurface
water flow within the gravel layer. The elevation level of the FWS wetland was 30 cm
higher than the SSF wetland. The water level remained constant for the duration of the
study.
Because the groundwater source used in the aquaculture farm contains about 0.5%
salinity, the plants for the constructed wetland system had to be salt tolerant. The FWS
Fig. 1. Layout of the pilot-scale FWSSSF series constructed wetland system for treating fishpond water. (A)
Sampling location for the influent; (B) sampling location for the FWS effluent; (C) sampling location for the SSF
effluent.
Y.-F. Lin et al. / Aquaculture 209 (2002) 169184 171
wetland was planted with water spinach (Ipomoea aquatica) in the front half and with a
native weed (Paspalum vaginatum) in the second half. The SSF wetland was planted
with common reed (Phragmites australis). During initial plant establishment (January
1999), the wetland water was kept static, with a water depth of 5 cm for the FWS and 40
cm for the SSF. The planting densities were 12% of wetland cover for the FWS and four
plants m
2
for the SSF. These three kinds of aquatic plants grew rapidly to colonize the
wetlands since influent was continuously added. Plants were not harvested during this
study.
2.2. System operation
Outdoor cultivation of fish in Taiwan normally begins in spring with harvest in the late
fall to avoid losing fish to the low water temperatures of the winter. The constructed
wetlands system in this study was operated from March 24, 1999 through November 20,
1999. This period was also the optimal season for vegetation growth and pollutant
processing in the wetlands.
The aquaculture wastewater was continuously fed into the FWS wetland directly from
the fishpond by a gear pump and then passed through the SSF wetland via gravity flow
(Fig. 1). A lateral perforated distribution pipe for inflow and a lateral trough-shaped
collector for drainage were installed at the inlet end and the distal end of the FWS tank,
respectively. Another lateral perforated pipe served as a collection drain and was installed
at the bottom of distal end of the SSF wetland. Sampling ports were set up at the inlet and
outlet of the FWS wetland, and at the outlet of the SSF wetland (Fig. 1). Five hydraulic
loading rates and associated retention times were monitored; each operating for 1 to 2
months (Table 1).
Table 1
Hydraulic conditions for operating the constructed wetlands system in various stages trials
Trial
nos.
Operating date Q
(m
3
day
1
)
q
(cm day
1
)
t for overall
system (day)
t for FWS
wetland (day)
t for SSF
wetland (day)
Stage 1 24 Mar. 1999
31 May 1999
0.18 1.8 12.8 8.4 4.4
Stage 2 1 June 1999
9 Aug. 1999
0.23 2.3 10.0 6.5 3.5
Stage 3 10 Aug. 1999
6 Oct. 1999
0.34 3.4 6.8 4.4 2.4
Stage 4 7 Oct. 1999
25 Oct. 1999
0.68 6.8 3.4 2.2 1.2
Stage 5 26 Oct. 1999
20 Nov. 1999
1.35 13.5 1.7 1.1 0.6
Q= average rate of inflow and outflow.
q =hydraulic loading rate, which is the average flow rate (Q) divided by surface area of wetland(s).
t =nominal hydraulic retention time, which can be computed as surface area times water depth times porosity of
wetland(s) divided by average flow rate. The porosity or fraction of the space available for water to flow through
the wetland was assumed to be 0.75 in FWS wetland and 0.4 in SSF wetland in this study.
Y.-F. Lin et al. / Aquaculture 209 (2002) 169184 172
2.3. Performance evaluation
Pollutant loading rate (g m
2
day
1
) was calculated by multiplying the hydraulic
loading rate (m day
1
) by the influent pollutant concentration (mg l
1
). Pollutant removal
rate (g m
2
day
1
) was defined as hydraulic loading rate times the difference in
concentration between the influent and the effluent. The first-order plug flow kinetic
model was applied accordingly:
C
0
C
i
expKt 1
Where C
i
= influent pollutant concentration, mg l
1
; C
o
= effluent pollutant concen-
tration, mg l
1
; t = nominal hydraulic retention time, day; K= first-order removal rate
constant, day
1
.
Because average temperatures of the influent and effluent of the constructed wetlands
system in each experimental stage ranged from 24.3 to 29.6 jC (Table 2), we omitted the
temperature effect on K and considered the K determined as an apparent reaction rate
constant.
Table 2
Water qualities (mean Fstandard deviation) at sampling locations for various stages trials
Sampling
location
Temperature
(jC)
NH
4
N
(mg l
1
)
NO
2
N
(mg l
1
)
NO
3
N
(mg l
1
)
TIN
(mg l
1
)
PO
4
P
(mg l
1
)
Stage 1
Influent 27.0 F2.2 0.16 F0.18 0.030 F0.085 0.26 F0.027 0.45 F0.53 2.39 F0.91
FWS effluent 27.3 F2.5 0.59 F0.47 0.041 F0.091 0.30 F0.16 0.90 F0.56 3.47 F2.35
SSF effluent 27.0 F3.2 0.40 F0.51 0.005 F0.004 0.19 F0.18 0.60 F0.70 0.73 F0.28
Stage 2
Influent 29.6 F1.6 3.31 F3.65 0.432 F0.450 0.74 F1.38 4.48 F5.48 7.44 F2.56
FWS effluent 29.3 F1.8 0.12 F0.12 0.006 F0.004 0.18 F0.21 0.30 F0.28 4.98 F1.72
SSF effluent 29.5 F1.7 0.08 F0.06 0.004 F0.002 0.13 F0.19 0.21 F0.25 2.15 F0.71
Stage 3
Influent 29.6 F1.2 1.30 F0.98 0.647 F0.450 0.94 F0.40 2.88 F1.83 10.45 F3.49
FWS effluent 29.0 F1.2 0.13 F0.11 0.006 F0.003 0.02 F0.01 0.16 F0.13 8.52 F3.83
SSF effluent 29.0 F1.2 0.08 F0.05 0.003 F0.002 0.01 F0.005 0.09 F0.06 5.72 F3.42
Stage 4
Influent 27.7 F1.8 1.46 F0.34 0.474 F0.339 2.26 F1.33 4.19 F2.01 8.57 F2.8
FWS effluent 27.1 F1.4 0.12 F0.10 0.003 F0.002 0.02 F0.001 0.14 F0.10 6.48 F1.87
SSF effluent 26.9 F1.6 0.07 F0.04 0.003 F0.002 0.01 F0.005 0.08 F0.05 5.26 F0.90
Stage 5
Influent 25.3 F1.7 0.80 F0.56 0.423 F0.134 2.66 F0.63 3.88 F1.32 5.19 F1.64
FWS effluent 24.5 F1.2 0.18 F0.21 0.005 F0.003 0.21 F0.06 0.39 F0.22 4.30 F1.07
SSF effluent 24.5 F1.6 0.11 F0.13 0.003 F0.002 0.07 F0.06 0.18 F0.17 3.53 F0.92
NH
4
N= ammonium nitrogen; NO
2
N=nitrite nitrogen; NO
3
N= nitrate nitrogen; TIN= total inorganic
nitrogen; PO
4
P=ortho-phosphate phosphorous.
Y.-F. Lin et al. / Aquaculture 209 (2002) 169184 173
2.4. Sampling and analysis
Water samples were taken twice each week for the duration of the study from the
influent of the system, effluent of the FWS wetland, and effluent of the SSF wetland. Such
sampling was usually carried out at around 10 A.M. in each sampling date. These samples
were analyzed for temperature, ammonium nitrogen (NH
4
N), nitrite nitrogen (NO
2
N),
nitrate nitrogen (NO
3
N), and ortho-phosphate phosphorus (PO
4
P) according to the
Standard Methods (American Public Health Association, 1985). Total inorganic nitrogen
(TIN) was calculated as the sum of NH
4
N, NO
2
N, and NO
3
N.
3. Results
3.1. Start-up behaviors
Continuous-flow operation of the constructed wetlands system was initiated with a low
hydraulic loading rate of 1.8 cm day
1
in Stage 1. Wetland plants grew actively in this
stage, but the vegetation cover was still sparse (Fig. 2). During Stage 1 (Fig. 3), reductions
of inorganic nitrogen (ammonium, nitrite, and nitrate) and phosphate through the FWS
wetland were negligible, although the influent nutrient concentrations and hydraulic
loading rate were low. Nutrient concentrations in the effluent of the FWS wetland were
higher than influent concentrations, resulting in negative removal efficiencies (Table 2).
Significant removal of inorganic nitrogen and phosphate in the FWS wetland did not occur
until around day 70 (nitrogen) and day 90 (phosphate) of operation (Fig. 3). Thereafter,
effluent concentrations of inorganic nitrogen in the FWS wetland were consistently low,
and effluent phosphate concentrations followed the trend of the influent concentrations
under various hydraulic loading rates.
Fig. 2. Time course of vegetation growth in FWS and SSF wetlands.
Y.-F. Lin et al. / Aquaculture 209 (2002) 169184 174
Conversely, significant reductions in concentrations of inorganic nitrogen through the
SSF wetland were achieved after approximately 30 days of operation (Fig. 3). Sub-
sequently, the SSF wetland showed consistently low nitrogen concentrations in its effluent.
Furthermore, phosphate concentrations in the SSF effluent were stable and consistently
lower than in the FWS effluent during Stage 1, and it followed the trend of the FWS
effluent and gradually increased because of an increase in the phosphate loading rate.
Fig. 3. Performance transitions during start-up period and the stationary state influent effluent concentration data
of nitrogen and phosphate for the pilot-scale FWSSSF series constructed wetlands system.
Y.-F. Lin et al. / Aquaculture 209 (2002) 169184 175
The wetlands system required approximately 7 months to attain a vegetative cover of near
80% (Fig. 2). Because of the unstable behavior, treatment results in Stage 1 were not
included for evaluation of system performance.
3.2. Nitrogen removal
Nutrient concentrations in the fishpond increased as feed residue and fish excreta
accumulated. In Stages 2 through 5, the influent concentrations in the constructed
wetlands system ranged from 0.12 to 14.7 mg NH
4
N l
1
, 0.02 to 1.5 mg NO
2
N l
1
,
Fig. 4. Relationships between effluent nitrogen concentrations (C
o
) and loading rate (LR) for NH
4
N and TIN in
the constructed wetlands system.
Y.-F. Lin et al. / Aquaculture 209 (2002) 169184 176
0.01 to 5.3 mg NO
3
N l
1
, and 3.1 to 17.7 mg PO
4
P l
1
. As shown in Fig. 3, the
wetlands system effluents in Stages 2 through 5 were consistently much lower in nitrogen
concentrations than the influents. Based on data in Table 2, the average removal
efficiencies of the wetlands system were 86% to 98% for NH
4
N, > 99% for NO
2
N,
82% to 99% for NO
3
N, and 95% to 98% for TIN. These efficiencies were extremely high
and were only slightly affected by hydraulic loading rate. The observed decreases in
inorganic nitrogen in the FWS wetland were significantly greater ( P< 0.01) than those in
the SSF wetland (Table 2). Effluent concentrations of NH
4
N and TIN in the constructed
wetlands (Fig. 4) were positively correlated with loading rates (r = 0.462 for NH
4
N and
0.114 for TIN). However, effluent concentrations were usually < 0.3 mg NH
4
N l
1
, 0.01
mg NO
2
N l
1
and 0.38 mg NO
3
N l
1
in Stages 2 through 5. Even at high hydraulic
loading rates (Stage 5), effluent values remained low (0.11 F0.13 mg NH
4
N l
1
,
0.003 F0.002 mg NO
2
N l
1
, and 0.07 F0.06 mg NO
3
N l
1
).
Removal rates for TIN showed a linear relationship to the loading rate (Fig. 5), with a
maximum removal rate of 0.55 g N m
2
day
1
occurring in Stage 5 when the loading rate
was increased to 0.59 g N m
2
day
1
. The first-order removal rate constant was
determined by linear regression analysis of the C
o
/C
i
(calculated by mean values in Table
Fig. 5. Comparison of TIN removal rates and loading rates observed in the present study with those demonstrated
from two comparable studies in the literature.
Y.-F. Lin et al. / Aquaculture 209 (2002) 169184 177
2) vs. t in Stages 2 through 5. Ammonium reduction followed first-order kinetics, with
removal rate constants of 0.434 day
1
(r =0.849) for the overall system, 0.573 day
1
(or
17.2 cm day
1
, r = 0.874) for the FWS wetland, and 0.17 day
1
(or 5.1 cm day
1
,
r = 0.812) for the SSF wetland. TIN data were unsatisfactory to fit the design model
because of low correlation coefficients in linear regression; but, a high rate constant could
still be obtained by computation directly using Eq. (1), with an average value of 1.175
day
1
for the overall system, 1.452 day
1
(43.6 cm day
1
) for the FWS wetland, and
0.665 day
1
(10.6 cm day
1
) for the SSF wetland.
3.3. Phosphate removal
The SSF wetland reduced phosphate levels at a rate equal to or even greater than the
FWS wetland in Stages 2 through 5 ( P= 0.92). Average overall removal efficiencies for
phosphate decreased markedly from 71.2% to 31.9% as the hydraulic loading rate
increased from 2.3 to 13.5 cm day
1
(Table 2). As a result, phosphate concentrations in
the effluent were exponentially correlated (r = 0.82) with the phosphate loading rates (Fig.
6). Average phosphate removal rates rose with the increase in phosphate loading, gradually
Fig. 6. Relationship between effluent phosphate concentrations (C
o
) and phosphate loading rate (LR) in the
constructed wetlands system.
Y.-F. Lin et al. / Aquaculture 209 (2002) 169184 178
reaching a plateau of about 0.23 g P m
2
day
1
(Fig. 7). A maximum removal rate of 0.46
g P m
2
day
1
was recorded in Stage 5 as the loading rate was increased to 0.9 g P m
2
day
1
. Phosphate removal data followed the first-order removal model, with rate constants
of 0.117 day
1
(r = 0.924) for the overall system, 0.063 day
1
(or 1.89 cm day
1
, r = 0.74)
for the FWS wetland, and 0.215 day
1
(or 3.44 cm day
1
, r = 0.945) for the SSF wetland.
4. Discussion
4.1. System start-up
The wetlands system required 2 to 3 months to reach a consistent nitrogen removal
performance level for the FWS wetland and 1 month for the SSF wetland. This is a much
shorter period than that reported in a study of the Tres Rios treatment wetland, Arizona,
USA, in which total nitrogen removal became efficient only after 1 year of operation
(International Water Association, 2000). The SSF wetland appeared to achieve stable low
nitrogen levels in the effluent more quickly than in the FWS wetland, probably because the
Fig. 7. Comparison of PO
4
P removal rates and loading rates observed in the present study with those
demonstrated from two comparable studies in the literature.
Y.-F. Lin et al. / Aquaculture 209 (2002) 169184 179
presence of gravel in the SSF wetland provided more specific surface area for biofilm
growth.
The FWS wetland required about 3 months to achieve significant phosphate removal.
In contrast, the SSF wetland showed efficient phosphate removal as soon as the wetlands
began to receive flow, even when vegetation was sparse. This implies that gravel
adsorption determined initial efficient phosphate removal. However, the ability of
adsorption may decrease with time as sorption sites on the gravel become saturated.
The brief start-up period for nutrient removal in this study can be attributed to factors such
as uniform hydrological conditions in a pilot-scale system and high water temperature
during the initial phase (27.6 jC averaged at that time). During the start-up period, a net
increase in nitrogen levels in both the FWS and SSF wetlands and in phosphate levels in
the FWS wetland occurred because the nutrient removal processes had not yet completely
developed and could not balance the nutrient release resulting from decomposition of
existing plant litter, leaching by the antecedent soil, or incomplete nitrificationdeni-
trification.
One growing season (from March to October) was required to reach about 80%
vegetation coverage in our study. International Water Association (2000) documented that
a period from 3 to 4 months up to 2 years is required for complete plant cover in treatment
wetlands. In this study, nutrient removal became stable 4 to 5 months before full
vegetation coverage was reached, suggesting that stable performance for pollutant removal
may be achieved without complete vegetation being established.
4.2. System performance
Removal efficiency for inorganic nitrogen was rather high after Stage 1 (86% to 98%
for NH
4
N, 95% to 98% for TIN), which might have resulted from the low nitrogen
concentrations generally found in aquaculture wastewater. The FWS wetland removed
most of the ammonium, nitrite, and nitrate once its performance became stable. This result
suggests that sufficient nitrification and denitrification proceeded concurrently in the FWS
wetland. Soil in the FWS wetland can provide ideal environments for microbial processes
including nutrient mineralization, nitrification and denitrification due to a variety of
microbial population and wide range of oxidation states present in soil (International Water
Association, 2000). The SSF wetland did not remove nitrogen as efficiently as the design
objective (although it did display quick start-up behavior) because the SSF wetland
received consistently lower load of nitrogen, thereby kinetically limiting nitrogen removal.
Therefore, it is not suitable to compare the performance data between the FWS and SSF
wetlands in such series combination system in which the loading rate to each wetland were
considerably different.
The removal rate of TIN ( < 0.55 g N m
2
day
1
) was relatively low as compared to
other study (Tanner et al., 1995; Fig. 5). Nitrogen removal rate is dependent on nitrogen
loading rate or hydraulic loading rate (Fig. 5). Although a high mean hydraulic loading
rate up to 13.5 cm day
1
was examined in this study, the maximum loading rate of
inorganic nitrogen was only 0.59 g N m
2
day
1
due to the characteristic of low strength
aquaculture wastewater. Therefore, the low removal rates in this study were apparently
limited by the low nitrogen loading rates. Both nitrogen removal efficiency and rate in this
Y.-F. Lin et al. / Aquaculture 209 (2002) 169184 180
study were higher than those demonstrated in a study of a constructed wetlands system
receiving channel catfish pond effluents (Schwartz and Boyd, 1995; Fig. 5), probably due
to the smaller wetland scale used and warmer climate in our work.
Phosphate removal was substantial and was about equal in the FWS and the SSF
wetland, indicating that phosphate was not a kinetically limited substrate in either wetland.
Phosphorus removal rates in relation to loading were similar to those documented for two
systems for treating fish farm wastewater (Schwartz and Boyd, 1995) and dairy farm
wastewater (Tanner et al., 1995) (Fig. 7). We believe that vegetation uptake and deposition
in soil and gravel are the two main mechanisms for phosphate removal in the study.
Sustainable phosphorus removal processes involve accretion of new wetlands sediments
and soils (International Water Association, 2000).
By combining different types of constructed wetlands in series, improved pollutant
removal can be achieved through a greater variety of treatment mechanisms. Similar to this
study, Schwartz and Boyd (1995) observed most of the nitrogen oxides, but not
ammonium, being removed in the first wetland of a series FWSFWS system from
low-concentration catfish pond effluents. However, in another study, Kadlec et al. (1997)
found that most nitrification occurred in the third SSF wetlands and most denitrification in
the final FWS wetland, whereas reduction of suspended solids and COD and ammoni-
fication took place in the first two FWS wetlands when treating high-concentration
wastewater with a four-stage FWSFWSSSFFWS system. These previous findings
and this study suggest that wastewater strength can affect the site (or retention time) of
pollutant removal in a series-constructed wetlands system. High concentrations of COD
and organic nitrogen (as in potato-processing wastewater) require longer retention times to
remove ammonium and nitrate, whereas wastewater with lower concentrations (as in
aquaculture wastewater) requires a shorter retention time.
Many researchers have applied the first-order reaction model for removal of ammo-
nium, nitrate, total nitrogen, and phosphate in FWS and SSF wetlands (Reed et al., 1995;
Kadlec and Knight, 1996). Ammonium removal rate constants for the FWS wetland (0.573
day
1
or 17.2 cm day
1
) and the overall system (0.434 day
1
) in this study are both higher
than the 0.218 day
1
reported by Reed et al. (1995) and the 4.932 cm day
1
by Kadlec and
Knight (1996), both for the FWS wetland. This result might be caused by the high
ammonium removal efficiency in our system, even under a high hydraulic loading rate.
Phosphate removal rate constants for the SSF wetland (0.215 day
1
or 3.44 cm day
1
) and
the overall system (0.117 day
1
) are both similar in magnitude to the 2.73 cm day
1
reported by Reed et al. (1995) and the 0.14 day
1
by Tanner et al. (1995), both for SSF
wetlands.
The present constructed wetlands system also performed well with respect to the
removal of chemical oxygen demand (2555%), suspended solids (4786%) and
chlorophyll a (7695%) from the fishpond effluent for the duration of this study
(unpublished data). Nutrient mineralization cannot be evidenced directly from the
influent effluent results because organic nitrogen and phosphorous were not measured
in this study. However, the consistent removal of pollutants (COD, SS, algae, inorganic
nitrogen and phosphate) implies that a considerable amount of particulate organic nitrogen
and phosphorous in fishpond effluent (generated from uneaten feed residue, fish excreta
Y.-F. Lin et al. / Aquaculture 209 (2002) 169184 181
and phytoplankton biomass) was retained in the constructed wetlands and subsequently
mineralized and removed from the fishpond effluent by various biotic and abiotic processes.
4.3. Area requirement of constructed wetland
The main disadvantage of constructed wetland is the requirement of large area. The
total area of a FWSSSF wetlands system (A
total
) required for treating a given flow rate of
fishpond effluent can be calculated by rearranging Eq. (1) and using the following
equations:
t
lnC
o
=C
i

K
1
t
Aeh
FWS
Aeh
SSF
Q
2
A
FWS
A
SSF
3
A
total
A
FWS
A
SSF
4
Where Q= flow rate of fishpond effluent, m
3
day
1
; e = porosity of wetland, assuming
0.75 in FWS wetland and 0.4 in SSF wetland in this study; h = water depth of wetlands,
being 0.4 m for both FWS and SSF wetland; A
FWS
= area required for FWS wetland, m
2
;
A
SSF
= area required for SSF wetland, m
2
.
Assuming fishponds have a daily regular effluent averaging 5% of the pond volume,
the depth of fishpond is 1.5 m, ammonium removal rate constant for the overall system is
0.434 day
1
, and the annual production of milkfish is normally 20 ton ha
1
year
1
, then 1-
ha fishpond would require 1.2 ha of wetlands or 1 ton of annual production of milkfish
would require 0.06 ha of wetlands to remove 80% of ammonium from fishpond effluent.
Schwartz and Boyd (1995) also estimated a similar wetland area of 0.72.7 times pond
area for treating catfish farm effluent.
Constructed wetland also can be potentially used for treating the recycle water with
operating at higher hydraulic loading rate and consequently with lower removal efficiency
in a recirculating intensive aquaculture system (Zachritz and Jacquez, 1993; Panella et al.,
1999). However, additional works on higher hydraulic loading rates and their effects on
water quality and fish growth in the recirculating aquaculture system are needed.
5. Conclusion
By observing the performance transitions and growth of wetland plants, we demon-
strated that FWS and SSF constructed wetlands showed quick start-up behaviors for
treating aquaculture wastewater, with the SSF wetland achieving stable performance
more rapidly than the FWS wetland. The start-up period around 3 months for nutrient
Y.-F. Lin et al. / Aquaculture 209 (2002) 169184 182
removal in this study seemed to be shorter than those that occurred in other systems cited
from the literature. The reason for this might be that a pilot-scale constructed wetland
system can develop stable removal processes more quickly than a large-scale constructed
system or a natural wetland. Because the aquaculture wastewater has low nitrogen
concentrations, removal of inorganic nitrogen was extremely efficient under various
hydraulic loading trials (2.3 to 13.5 cm day
1
), and the removal rate constants for
ammonium and nitrogen oxides were typically high. Phosphate concentrations in the
aquaculture wastewater were moderately high, and removal efficiencies for phosphate
decreased significantly as hydraulic loading rate increased. However, hydraulic loading
rates and operating time were confounded because we cannot evaluate the influence of
time vs. the changes in loading rate. The subtropical climate prevailing in Taiwan
provides for suitable conditions for the constructed wetlands ecosystem, which also
might help to explain the rapid start-up phenomena and efficient nutrient removal in this
wetlands system. Levels of ammonium and nitrite in wetlands discharge ( < 0.3 mg
NH
4
N l
1
and 0.01 mg NO
2
N l
1
) were obtained that rendered the water harmless to
fish, so that the water potentially could be reused and recycled in the aquaculture system.
Acknowledgements
We would like to express our appreciation to the National Science Council of the
Republic of China (Project Number: NSC-88-2621-Z-041-001) for funding support for
this project, and to Dr. James P. Kaetz, Associate Professor, Department of English,
Auburn University and Editor of National Forum, for help in editing this paper.
References
American Public Health Association, 1985. Standard Methods for the Examination of Water and Wastewater.
APHA, AWWA, and WPCF, Washington, DC, 1268 pp.
International Water Association, 2000. Constructed Wetlands for Pollution Control. Processes, Performance,
Design and Operation. IWA Publishing, London, 156 pp.
Kadlec, R.H., Knight, R.L., 1996. Treatment Wetlands. CRC Press, Boca Raton, 893 pp.
Kadlec, R.H., Burgoon, P.S., Henderson, M.E., 1997. Integrated natural systems for treating potato processing
wastewater. Water Sci. Technol. 35, 263270.
Panella, S., Cignini, I., Battilotti, M., Falcucci, M., Hull, V., Milone, N., Monfrinotti, M., Mulas, G.A.,
Pipornetti, G., Tancioni, L., Cataudella, S., 1999. Ecodepuration performances of a small-scale experimental
constructed wetland system treating and recycling intensive aquaculture wastewater. Ann. N. Y. Acad. Sci.
879, 427431.
Reddy, K.R., DAngelo, E.M., 1997. Biogeochemical indicators to evaluate pollutant removal efficiency in
constructed wetlands. Water Sci. Technol. 35, 110.
Reed, S.C., Crites, R.W., Middlebrooks, E.J., 1995. Natural Systems for Waste Management and Treatment.
McGraw-Hill, New York, 433 pp.
Schwartz, M.F., Boyd, C.E., 1995. Constructed wetlands for treatment of channel catfish pond effluents. Progr.
Fish-Cult. 57, 255267.
Tanner, C.C., Clayton, J.S., Upsdell, M.P., 1995. Effect of loading rate and planting on treatment of dairy farm
wastewaters in constructed wetlands: II. Removal of nitrogen and phosphorus. Water Res. 29, 2734.
Y.-F. Lin et al. / Aquaculture 209 (2002) 169184 183
Van Rijn, J., 1996. The potential for integrated biological treatment system in recirculating fish culturea
review. Aquaculture 139, 181201.
Zachritz, W.H., Jacquez, R.B., 1993. Treating intensive aquaculture recycled water with a constructed wetlands
filter system. In: Moshiri, G.A. (Ed.), Constructed Wetlands for Water Quality Improvement. Lewis Publish-
ers, Boca Raton, pp. 609613.
Y.-F. Lin et al. / Aquaculture 209 (2002) 169184 184

Das könnte Ihnen auch gefallen