Sei sulla pagina 1di 29

Aquacultural Engineering 26 (2002) 81 109

www.elsevier.com/locate/aqua-online

Constructed wetlands as recirculation filters in


large-scale shrimp aquaculture
David Rogers Tilley a,*, Harish Badrinarayanan b,
Ronald Rosati c, Jiho Son b
a
Biological Resources Engineering, Uni6ersity of Maryland, 1449 An. Sci./Ag. Engr. Bldg.,
College Park, MD 20742, USA
b
Department of En6ironmental Engineering, Texas A&M Uni6ersity-Kings6ille, Kings6ille,
TX 78363, USA
c
College of Agriculture and Human Sciences, Texas A&M Uni6ersity-Kings6ille, Kings6ille,
TX 78363, USA

Received 10 March 2001; accepted 4 March 2002

Abstract

Effluent waters from shrimp aquaculture, which can contain elevated levels of phosphorus,
ammonia, nitrate, and organics, must be managed properly if shrimp aquaculture is to
achieve sustainability. Constructed wetlands are ecologically beneficial, low cost treatment
alternatives proven capable of reducing suspended solids, biochemical oxygen demand
(BOD), nitrogen, phosphorus and heavy metals from wastewater of many sources. The goal
of this study was to determine how well a constructed wetland performed as a filter in a
full-scale shrimp aquaculture operation. A 7.7 ha (19 ac) mesohaline (3 8 ppt) constructed
wetland treating 13,600 m3 per day (3.6 MGD) of effluent from 8.1 ha (20 ac) of intensively
farmed shrimp ponds at the Loma Alta Shrimp Aquaculture Facility (LASAF), located
along the coast of the Gulf of Mexico in semi-arid South Texas, was found to reduce
concentrations of total phosphorus (TP), total suspended solids (TSS) and inorganic sus-
pended solids (ISS) by 31, 65 and 76%, respectively, during recirculation, and maintained
consistently low levels of mean BOD ( B9 mg l 1), total ammonia ( B 1.8 mg N l 1) and
nitrate ( B 0.42 mg N l 1). Determination of parameter values for the k C* wetland design
model for ISS or TP showed that mean target levels could be achieved, given expected
influent concentrations, when the ratio of pond surface to wetland surface was 12. Con-
structed wetlands can perform satisfactorily as recirculation filters in large-scale shrimp

* Corresponding author. Tel.: + 1-301-405-8027; fax: + 1-301-314-9023.


E-mail address: dtilley@umd.edu (D.R. Tilley).

0144-8609/02/$ - see front matter 2002 Elsevier Science B.V. All rights reserved.
PII: S0144-8609(02)00010-9
82 D.R. Tilley et al. / Aquacultural Engineering 26 (2002) 81109

aquaculture operations, reducing the impact of effluent on local water bodies, conserving
large quantities of water and providing valuable ecological habitat. 2002 Elsevier Science
B.V. All rights reserved.

Keywords: Constructed wetlands; Shrimp aquaculture; Water quality; Sustainable farming; Ecological
engineering

1. Introduction

1.1. Statement of the problem

The worlds annual supplies of fish and shellfish produced through aquaculture
grew from 1.6 kg per capita in 1991 to 2.1 kg in 1998 and were predicted to provide
half of world demand by 2030 (UNFAO, 2000) if not earlier (Paez-Osuna, 2001).
Shrimp aquaculture was popular worldwide and gained ground in U.S. agriculture
to meet growing domestic demand for high-end protein (National Aquaculture
Association, 1998). From 1993 to 2001, U.S. imports of shrimp increased by 27%
from 273,000 MT valued at $2 billion to 346,000 MT worth $3.8 billion (USDA,
1998, 2001). Still, U.S. shrimp farms produced only 0.6%, by weight (1900 MT) and
value ($12.2 million), of the 1993 imports (USDA, 1998).
Shrimp aquaculture can be a highly profitable venture when smart resource
management is coupled to the appropriate natural conditions. Internationally, the
industry has been devastated by poor water quality management. For example,
virus epidemics greatly diminished recent harvests in South America (Funge-Smith
and Briggs, 1998). The industry discharges large quantities of brackish water that
must meet water quality standards set forth by laws (e.g. U.S. Clean Water Act)
and regulations (e.g. U.S. National Pollutant Discharge Elimination System). Thus,
to maintain and encourage sustainable shrimp aquaculture that meets state water
quality standards, low-cost, low-maintenance technologies are needed which reduce
water demand and improve the quality of discharged water (Lin, 1995). In
particular, the water quality parameters of most concern in shrimp aquaculture
have been inorganic suspended solids (ISS), total suspended solids (TSS), biochem-
ical oxygen demand (BOD), dissolved oxygen (DO) and ammonia (Paez-Osuna,
2001). Additionally, total phosphorus (TP), nitrate and the pH of discharge waters
are of concern for the health of receiving waters.
Unionized ammonia (NH3) is the toxic form of ammonia. Juvenile penaeid
shrimp, such as Litopenaeus 6annamei, exhibited an exposure-sensitive LC50 (me-
dian lethal concentration), ranging from 2.9 mg NH3 N l 1 at 24 h to 1.6 mg
NH3 N l 1 at 96 h exposure (25 ppt salinity, 8.05 pH, 23 C; Lin and Chen,
2001). Total ammonia (NH3 plus NH+ 4 ), on the other hand, was much less toxic
with LC50s from 66 to 35 mg total ammoniaN l 1 for the same conditions (Lin
and Chen, 2001). Nitrogen is often the limiting factor in marine ecosystems (Day et
al., 1989); therefore its discharge from shrimp aquaculture in large amounts can
create unhealthy eutrophication in natural coastal waters. Phosphorus is also
D.R. Tilley et al. / Aquacultural Engineering 26 (2002) 81109 83

required for eutrophic blooms so its discharge must be controlled. Elevated


loadings of organic carbon increase the BOD of receiving waters, which often
interferes with ecological oxygen cycles. High concentrations of ISS discharged to
estuaries in warm climates suitable for shrimp farming, can increase turbidity,
which deteriorates the habitat for seagrass beds (Longstaff and Dennison, 1999).
Constructed wetlands are a viable technology for treating and recirculating the
water used in shrimp aquaculture production (Negroni, 2000). As natural systems
driven in large part by free energies such as sunlight and wind, constructed
wetlands have proven financially smart investments in controlling inorganic and
organic solids, nutrients, and pathogens when employed in municipal (Nerella et
al., 2000), industrial (Vrhovsek et al., 1996) and agricultural (Knight et al., 2000)
applications as wastewater treatment filters (Kadlec and Knight, 1996). Kivaisi
(2001) reviews the technology for application in developing countries.
A pilot-scale gravel bed, subsurface-flow constructed wetland located in Thailand
and planted with mangrove fern (Acrostchum aureum), treating high salinity shrimp
pond effluent reduced the concentration of suspended solids, BOD, total organic
carbon, total nitrogen and TP by 84, 91, 46, 48 and 31%, respectively (Sansanayuth
et al., 1996). Samocha and Lawrence (1995) recommended constructed wetlands
when they reviewed treatment alternatives for shrimp aquaculture. Costa-Pierce
(1998) used tertiary treated municipal wastewater to produce tilapia, carp and
crayfish in an integrated aquaculture-wetland ecosystem and found that the wetland
reduced ammonia from 8.9 to 0.7 mg N l 1 and decreased nitrate from 3.7 to 0.6
mg N l 1.
Zurayk et al. (1997) showed that increased residence time in pilot scale con-
structed wetlands increased phosphorus uptake by reed, cattail and water hyacinth.
Serodes and Normand (1999) demonstrated the ability of cattail to accumulate P in
a constructed four-cell, surface-flow wetland in Quebec, receiving agricultural
wastewater with inflow concentrations around 20 mg P l 1.
Verhoeven and Meuleman (1999) showed that plant species had little impact on
N removal in a 3-year study of a constructed wetland vegetated with wool grass
(Scirpus cyperinus) and cattail (Typha latifolia). They also found that both NH+ 4
and TKN concentrations decreased exponentially with longer residence times.
Removal ranged from 18 to 39% for ammonium and from 31 to 46% for TKN.
Bachand and Horne (2000), studying three free-surface marsh vegetation treatments
(bulrush, cattail and a mixed stand of macrophytes and grasses), observed cattail to
remove 565 mg N m 2 per day, bulrush 261 mg N m 2 per day and mixed 835 mg
N m 2 per day. Bacterial denitrification rather than plant uptake was found to be
the main mechanism for nitrate removal. Studies on ammonium reduction by
treatment wetlands have yielded variable results with some wetlands demonstrating
reduction while others actually released more ammonium than in influent (Kadlec
and Knight, 1996). Gearheart (1990) showed that ammonia reduction in pilot
wetlands tended to reach a steady state concentration around 35 mg l 1.
Gersberg et al. (1989) witnessed the reduction of viral counts and other patho-
genic organisms in wastewaters treated by created wetlands. Verhoeven and Meule-
man (1999) found that a constructed wetland in Holland, receiving sewage from 800
84 D.R. Tilley et al. / Aquacultural Engineering 26 (2002) 81109

people-equivalents, reduced influent Escherichia coli and F-specific virus counts


by 4 logarithms. Perkins and Hunter (2000) saw a 8594% decrease in fecal
coliform and fecal streptococci in pilot-scale Typha wetlands.
A municipal treatment wetland in Queensland (Greenway and Woolley, 1999)
produced effluent with suspended solids less than 22 mg l 1 and BOD less than
12 mg l 1. A wetland designed to treat high COD (\ 3500 mg l 1) wastewater
from a food processing plant in Slovenia reduced COD by 92% (Vrhovsek et al.,
1996).

1.2. Plan of study

In an effort to test sustainable solutions to water management on full-scale


shrimp farms, a 7.7 ha (19 ac) mesohaline (salinity of 315 ppt) wetland was
created at the Loma Alta Shrimp Aquaculture Facility (LASAF), located on the
H.P. El Sauz Ranch near Port Mansfield, TX. The wetland served as the water
filtration system in the re-circulating agroecosystem. During the year of this
study (2000), the LASAF contained 8.1 ha (20 ac) of shrimp ponds that pro-
duced 1015 MT of L. 6annamei at harvest.
The ability of the mesohaline constructed wetland to improve water quality
during both the recirculation and harvest periods was evaluated. Water collection
stations were established along a transect, parallel to the hydraulic gradient of
the LASAF. Samples were collected weekly during recirculation and daily during
harvest. Analytes included total dissolved solids (TDS), pH, DO, total ammonia
(NH3 plus NH
4 ), nitrate (NO3 ), nitrite (NO2 ), TP, reactive phosphorus (RP),
TSS, volatile suspended solids (VSS), ISS, and 5-day biochemical oxygen demand
(BOD5). Time series plots of the quality of water supply, wetland influent and
wetland effluent were made to see overall treatment effect of wetland. Paired
t-test were conducted to check for statistical differences between locations
throughout sampling period. Additionally, the mean seasonal value of each water
quality parameter was plotted as a function of wetland area to determine effect
of treatment area.
Shrimp aquaculturists implementing constructed wetlands as their main filtra-
tion system will need to know wetland space requirements. Key parameters for
constructed wetland design equations given by Kadlec and Knight (1996) were
determined based on data collected. The design equations define the wetland
area required to treat a given influent concentration (e.g. ISS or TP) to a
prescribed target level. Furthermore, the design equations were charted so the
relationship between wetland treatment area, inlet concentration and effluent
concentration could be visualized. The required wetland treatment area was nor-
malized to the contributing pond surface area and reported as the ratio of pond
to wetland (P:W). Research in this area will be useful in evaluating the feasibility
of implementing the constructed wetlands technology as a recirculation filter on
individual large-scale shrimp aquaculture facilities.
D.R. Tilley et al. / Aquacultural Engineering 26 (2002) 81109 85

2. Materials and methods

2.1. Description of system studied

The 2-year-old 7.7 ha constructed wetland that served as the recirculation filter
for the LASAF (26 28% 20 N and 97 28% 09 W; see Fig. 1) was the study area.
The climate is classified as semi-arid (NWS, 2001). During the sampling period
(July 170 mm of precipitation fell on 18 days (NOAA, 2002). Wind speeds on an
average day range between 6 and 7 m s 1 during the shrimp growing season
(NOAA, 2002). Soils are acidic sands and sandy loams (TPWD, 2001).

Fig. 1. Layout of the LASAF at the H.P. El Sauz Ranch in South Texas, USA.
86 D.R. Tilley et al. / Aquacultural Engineering 26 (2002) 81109

Fig. 1 shows the layout of the LASAF wetland. Water was taken from the
Hidalgo Drainage Ditch, which drains cropland planted with sorghum, cotton, and
sugar cane and discharges to the central Laguna Madre. Water supply was pumped
to the water delivery canal where it was gravity-fed to four equally sized shrimp
ponds (2 ha 1.5 m). Individual forebays received each ponds discharge by
gravity-feed prior to the effluent entering the wetland. A circuitous route for water
flow in the wetland was created using Earthen berms inside the wetland to promote
plug-flow and extend the hydraulic residence time (HRT). Thus, the wetland was
divided into five segments defined by sampling locations (Fig. 1). All pond
discharge entered section A of the wetland (Fig. 1) and proceeded to section B via
the opening where sampling point W1 was placed. Having conducted similar
monitoring in the previous growing season (Valenti, 2000), it was known that
several water quality parameters declined to a near-steady state within sections B
and C. Thus, sampling point W2 was placed to better detect the change. From
section B, water flowed around another Earthen berm through section C and on to
section D. Sampling point W4 was placed where water rounded the final interior
berm and flowed on through section E to the final wetland sampling point W5.
Originally, the wetland was planted with 10 species of salt tolerant wetland
plants, including widgeongrass (Ruppia maritima), chara (Chara spp.), pithophora
(Pithophora spp.), fragrant waterlily (Nymphaea odorata), southern watergrass
(Hydrochola carolinensis), cattail (T. latifolia), soft rush (Juncus effusus), rattlebush
(Sesbania drummondii ), sea-oxeye (Borrichia frutescens) and black mangrove (A6i-
cennia germinans) (Valenti, 2000). However, by the second year of operation the
year of this studythe wetland had undergone self-design whereby T. latifolia
became the overwhelmingly dominant plant species, occupying nearly two-thirds of
the wetland area with other planted species occupying much less.
Operation of the wetland consisted of three distinct phases: fill, recirculation and
harvest. At the beginning of the growing season (May) the wetland was devoid of
surface water so the fill phase was undertaken until at least 0.15 m of standing
water was present in the wetland at which time the recirculation phase began.
During recirculation water treated by the wetland was pumped to the water delivery
canal for reuse in the shrimp ponds (Fig. 1). Water supply was withdrawn from the
Hidalgo Drainage Ditch to compensate for losses due to evapotranspiration. The
flow rate averaged 13,600 m3 per day (3.6 MGD), which provided a HRT in the
wetland on the order of 24 h. Depth of water was maintained between 0.15 and
0.45 m (618 in.). The recirculation period began on July 24, 2000 and continued
until October 9, 2000.
On October 9, an early season cold front passed and dropped the temperature of
the pond water below 10 C. This forced an emergency drawdown of all ponds so
harvest could commence before shrimp mortality reached dire levels. This resulted
in a tremendous pulsed discharge of 136,000 m3 over a 36 h period. During this
period the wetland water depth reached a maximum of 0.69 m (27 in.) 48 h after
pond draining commenced. After peaking, the water depth decreased following a
first-order decay rate (k = 0.009 h 1), reaching 0.59 m (23 in.) at 72 h and 0.10 m
(4 in.) at 266 h (11 days). Although the shrimp harvest was complete by October 12,
the wetland continued to discharge until October 23.
D.R. Tilley et al. / Aquacultural Engineering 26 (2002) 81109 87

Table 1
Analytical methods used to characterize water quality at the LASAF

Parameter Analytical method Equipment

Field methods
pH EPA 150.1 pH electrode method EXTECH Oyster
pH/conductivity meter
Dissolved oxygen Oxygen electrode method YSI Model 55 Handheld DO
system
Temperature Temperature electrode method YSI Model 55 Handheld DO
and temperature system
Laboratory methods
Salinity Standard methods 2520 B EXTECH Oyster
electrical conductivity method pH/conductivity meter
Nitrogen, ammonia EPA approved Hach Method Hach DR-2000
8038 Spectrophotometer
Nitrogen, nitrate EPA approved Hach Method Hach DR-2000
8171 Spectrophotometer
Phosphorus, total EPA approved Hach Method Hach DR-2000
8190 Spectrophotometer
Phosphorus, reactive EPA approved Hach Method Hach DR-2000
8048 Spectrophotometer
Total, inorganic and EPA 160.2 Filtration and drying apparatus
VSS
Carbonaceous EPA approved Hach Method Incubator
biological oxygen 8043
demand

2.2. Field methods

During recirculation water samples were collected between 09:00 and 11:00 h
every Monday. During harvest in October, samples were collected on a daily basis.
Water samples were collected at mid-depth using 1 l grab bottles, which were
cleaned and labeled accordingly prior to collection. Samples were collected at the
five (5) wetland locations, pond c2, the water delivery canal, and the Hidalgo
Drainage Ditch (supply water) (Fig. 1). Sample bottles were placed on ice and
transported to the environmental engineering water chemistry lab at Texas A&M
University-Kingsville for analysis. Table 1 shows the field methods used to measure
temperature, pH, and DO.

2.3. Analytical techniques

Table 1 lists the analytical techniques that were used. Ammonia, nitrate, nitrite,
RP, and TP were determined using USEPA approved methods for the Hach
DR2000 Spectrophotometer (APHA, 1992; Hach, 1991). TSS, VSS and ISS were
determined by drying and weighing. The BOD5 was determined using a USEPA
approved Hach method (Hach, 1991). Samples (300 ml) were incubated at 20 C
88 D.R. Tilley et al. / Aquacultural Engineering 26 (2002) 81109

for 5 days. The initial and 5-day DO concentrations were recorded and used to
estimate BOD5.

2.4. Data analysis

The concentration of each water quality parameter in the water supply (Hidalgo
Drainage Ditch), wetland influent (pond outlet) and wetland effluent was plotted as
a time series that included both recirculation and harvest. Paired t-tests were
performed between each of the three bodies of water for each parameter.
The effect of the wetland on water quality was evaluated by plotting mean
seasonal concentration of each constituent along with 9 1 standard deviation as a
function of wetland area for both recirculation and harvest data. The mean and
standard deviation of the seasonal concentration for a specific constituent at a
specific sampling point was calculated from all samples taken at that point (eleven
samples for recirculation and nine for harvest).

2.5. Design equations

Kadlec and Knight (1996) proposed the two-parameter k C* model to describe


the removal of pollutants by wetlands along the direction of flow. The model,
which assumes steady plug flow conditions, can be written:
dC/dx = k/qw(C C*), (1)
where C is the concentration of the constituent of concern (mg l 1), x is the
fraction of the area through the wetland along the flow path, k is the areal rate
constant (m per day), qw is the hydraulic loading rate of the wetland (m per day)
and C* is a background concentration (mg l 1). Solution of the equation provides
an exponential decrease in concentration along the flow path. Considering the
hydraulic loading from a given sized shrimp pond, the equation can be manipulated
to solve for the wetland treatment area required to meet a given target outlet
concentration for a given inlet concentration and shrimp pond discharge rate:
Aw = ln[(C0 C*)/(Ci C*)]qpAp/kz, (2)
where Aw is wetland treatment area (m2), C0 is the target concentration (mg l 1),
Ci is the inlet concentration (mg l 1), qp is the hydraulic loading rate from the
shrimp pond (m per day), Ap is shrimp pond area (m2), and kz is areal rate constant
for constituent z (m per day). Determination of kz for a specific wetland and
constituent is the key to making use of Eq. (2). The other important determination
is on the choice of C*, the background concentration. A kz was determined for each
of TP and ISS during the recirculation period using Eq. (3) (Kadlec and Knight,
1996). Since the equation assumes steady flow, its utility for assessing treatment
requirements during the harvest period, when flow was dynamic, was deemed
insufficient.
kz = ln[(C0 C*)/(Ci C*)]qw/x. (3)
D.R. Tilley et al. / Aquacultural Engineering 26 (2002) 81109 89

Kadlec and Knight (1996) suggested that C* can be assumed to equal zero for
TP based on the presumption that phosphorus is often a limiting nutrient that
will be taken up rapidly and conserved in aquatic ecosystems. Here, we consider
C* to represent the background concentration achievable within the LASAF
wetland given the regions high levels of soil phosphorus (Dadi, 2001). Therefore,
C*TP was chosen as the lower control limit (explanation to follow) of the wetland
process mean, which was nearly identical to the historic TP levels in the local
receiving estuary (Laguna Madre) (Dadi, 2001). C*ISS was chosen in a similar
fashion.
Eq. (2) was used to estimate the wetland area needed to reduce TP and ISS to
different target levels, which were chosen either based on the wetlands current
performance limitations or on typical regulatory wastewater treatment plant per-
mitting standards. The highest target level for ISS, which was based on typical
wastewater permits, represented a stringent, but conceivable regulatory require-
ment (20 mg ISS l 1). Similar permitting guidelines do not exist for TP. The
other target levels were developed based on Statistical Quality Control (SQC)
practices (Montgomery, 1991). Specifically, the methodology for setting control
limits on SQC control charts (i.e. the process mean9 3|) was used. In SQC the
Central Limit Theorem (CLT) provides the basis for designing the upper and
lower control limits on process control charts for variables. Control limits set at
the process mean9 3 standard deviations cover 99.73% of the observations made
for a normally distributed random variable. The appropriateness of the normal
distribution is ensured by the CLT. Therefore, the process mean, the upper
control limit (process mean +3|) and the lower control limit (process mean
3|) were chosen as target levels for TP and ISS outflow concentrations. The
mean and sample standard deviation used in each process control chart was
taken from all sampling stations in the wetland downstream from the inlet (sites
W1 W5). The sample standard deviation was calculated for each set of observa-
tions (n= 5) taken during each sampling date and averaged for all dates during
recirculation (m =10).

3. Results

Since water management goals differ between recirculation and harvest, the
ability of the wetland to effect water quality was evaluated separately for recir-
culation and harvest. Each particular water quality parameter was assessed by
plotting its mean seasonal concentration and standard deviation as a function
wetland treatment area. To note the significance of residence time, the recircula-
tion (24 h HRT) and harvest (1 h) data were plotted on the same chart.
Although the sampling locations for the recirculation and harvest periods were
exactly the same, the harvest data was shifted to the right in the charts so the
confidence intervals could be easily distinguished.
90 D.R. Tilley et al. / Aquacultural Engineering 26 (2002) 81109

3.1. Temporal analysis of water quality

Fig. 2 through Fig. 5 compare the temporal patterns in water quality parameters
for water supply, wetland influent and wetland outlet. The paired t statistical test
was used to detect whether differences between each combination (water supply vs.
wetland influent, wetland influent vs. wetland effluent, water supply vs. wetland
effluent) existed (Table 2). The majority (27 of 33) of P-values were less than 0.10,
indicating that the water quality of each water body was different from the other
over the entire sampling period (recirculation plus harvest).
Fig. 2 shows time series of TDS, pH and DO for the water supply (Hidalgo
Drainage Ditch), wetland influent and wetland effluent during recirculation and
harvest periods. TDS levels in the wetland effluent were always greater than water
supply with a mean difference of 2.01 ppt (P =0.0005). The water supplys pH was
greater than the wetland effluent at all sampling times except for three dates during
harvest when the flow rate was nearly one order of magnitude greater than the
mean. This was also the case for DO, which was 1.87 mg l 1 less at the wetland
outlet than in water supply (P =0.005). The wetlands DO was always less during
recirculation, but was nearly the same during the harvest period (Fig. 2).
Table 2
Paired differences in water quality parameters for water supply, wetland inlet and outlet (mg l1 unless
otherwise noted) with P-values for paired t-test when less than 0.10

Parameter Mean difference (P-value)

Water supplywetland inlet Wetland Water supplywetland outlet


inletoutlet

TDS (ppt) 1.389 0.621 2.011


(0.0005) (0.0082) (0.0005)
pH (units) 0.779 0.672
(0.0004) (0.0009)
DO 0.793 2.661 1.867
(0.0537) (0.0015) (0.0050)
Total ammonia 0.374 0.508 0.882
(0.0177) (0.0074) (0.0018)
NO
3 0.247 0.247
(0.0308) (0.0391)
RP 0.033 0.085
(0.0936) (0.0305)
TP 0.139 0.232
(0.0010) (0.0064)
TSS 42.947 47.368 90.316
(0.0092) (0.0058) (0.0006)
ISS 49.895 35.579 85.474
(0.0028) (0.0085) (0.0006)
VSS 11.789
(0.0101)
BOD 1.745 3.220 1.475
(0.0636) (0.0274) (0.0252)
D.R. Tilley et al. / Aquacultural Engineering 26 (2002) 81109 91

Fig. 2. Temporal pattern of TDS, pH and DO during recirculation and harvest.


92 D.R. Tilley et al. / Aquacultural Engineering 26 (2002) 81109

Fig. 3. Temporal pattern of nitrate and total ammonia during recirculation and harvest.

Time series of nitrate and total ammonia for water supply and wetland effluent
are given in Fig. 3. Nitrate in supply water was greater than in wetland effluent
(0.247 mg l 1, P =0.0391). There was no significant difference in NO 3 between
wetland influent and effluent. Total ammonia, on the other hand, was greater in the
wetland effluent than in water supply (mean difference= 0.882 mg l 1; P= 0.0018)
and wetland influent (mean difference=0.508 mg l 1; P= 0.0074).
RP and TP were greater in supply water than in wetland effluent with mean
differences of 0.085 and 0.232 mg l 1 (P = 0.0305 and 0.0064), respectively (Fig. 4).
Wetland effluent RP and TP were also less than wetland influent (Table 2).
Similarly, TSS and ISS (Fig. 5) were greater in supply water than in wetland
effluent (mean difference=90.3 and 85.5 mg l 1; P =0.0006 and 0.0006, respec-
tively). VSS (Fig. 5), however, showed no significant difference between water
supply and wetland effluent, but was lowered by an average of 11.8 mg l 1 (P
0.0101) from wetland inlet to outlet (Table 2).
D.R. Tilley et al. / Aquacultural Engineering 26 (2002) 81109 93

For all sampling dates except one, BOD5 levels in supply water were greater than
in wetland effluent (Fig. 6) with a mean difference of 1.475 mg l 1 (P=0.0252).
BOD5 levels increased by an average of 1.745 mg l 1 from the water supply to the
pond effluent (P =0.0636), but were reduced across the wetland by an average of
3.22 mg l 1 (P=0.0274).

3.2. Spatial analysis of water quality

Wetland treatment performance is related to the amount of space available for


the biological, chemical and physical processes to interact with the chemistry of the
water. Thus, it is imperative that the spatial dynamics of water quality be
understood, if a constructed wetland is to be employed as an aquacultural filter.

Fig. 4. Temporal pattern of reactive phosphorous and total phosphorous during recirculation and
harvest.
94 D.R. Tilley et al. / Aquacultural Engineering 26 (2002) 81109

Fig. 5. Temporal pattern of TSS, ISS and VSS during recirculation and harvest.
D.R. Tilley et al. / Aquacultural Engineering 26 (2002) 81109 95

Fig. 6. Temporal pattern of BOD during recirculation and harvest.

Fig. 7 shows how ISS generally declined in a logarithmic fashion along the spatial
transect for both recirculation (Fig. 7a) and harvest (Fig. 7b). During recirculation,
the ISS levels were nearly always within the control limits at 13,000 m2, whereas
during harvest when the HRT was about 1 h, 35,000 m2 was required before all ISS
samples were within control limits. Nine of 10 ISS transects were within control
limits by 6000 m2 during recirculation (Fig. 7a) and eight of nine were less than the
upper control limit by 13,000 m2 during harvest. Fig. 8 provides a similar
viewgraph for TP. TP also exhibited a logarithmic decrease with area, but not
nearly as pronounced as ISS. TP was not as well under control as ISS, especially
during harvest (Fig. 7b).
Fig. 9 shows the mean TDS, pH and DO levels 9 1 standard deviation as
functions of wetland area during recirculation and harvest. The average TDS
increased from inlet to the outlet during both recirculation and harvest (Fig. 9a).
During recirculation, the average TDS increased from 2.8 ppt in the inlet to 3.5 ppt
at the outlet. During harvest, the average TDS increased from 2.7 to 3.2 ppt from
inlet to the outlet. TDS remained constant or decreased in the cattail-dominated
region (segments B D in Fig. 1) in the wetland during both operational periods.
The average pH decreased from 8.3 at the inlet to 7.4 at outlet during recirculation
and from 8.6 to 7.9 during harvest (Fig. 9b). The mean DO decreased sharply in the
cattail-dominated region of the wetland (15,00048,000 m2) (Fig. 9c). From the
cattail-dominated sites to the wetland outlet the DO returned to higher levels.
Across the entire wetland during recirculation, the average influent DO was 5.7 mg
l 1 and the average effluent DO was 2.1 mg l 1. During harvest it decreased from
6.2 to 4.6 mg l 1 from inlet to outlet.
Fig. 10a shows that there was no appreciable change in average nitrate concen-
trations during either recirculation or harvest. Fig. 10b shows that average total
ammonia increased asymptotically to 1.1 mg l 1 during recirculation, but reached
96 D.R. Tilley et al. / Aquacultural Engineering 26 (2002) 81109

a slightly higher steady state during harvest (1.9 mg l 1) when input concentration
was 67% greater and HRT was 1/24th of recirculation. Ammonia concentrations
plotted as a logarithmic function of wetland area produced a straight line (R 2 =
0.75 recirculation, 0.62 harvest; figure not shown).
Fig. 11a shows that there was a 20% decrease in average RP across the wetland
during recirculation (0.05 0.04 mg l 1) and a 45% decrease during harvest
(0.110.06 mg l 1). Fig. 11b shows that average TP decreased asymptotically from
0.31 mg l 1 at the inlet to 0.25 mg l 1 at the outlet during recirculation. The trend
and asymptote were similar during harvest when inlet TP levels were higher (0.42
mg l 1) and HRT was shorter. The average TP was reduced by 19 and 48% during
recirculation and harvest, respectively. TP plotted as a logarithmic function of

Fig. 7. Spatial transects of ISS for sampling dates during: (a) recirculation; and (b) harvest.
D.R. Tilley et al. / Aquacultural Engineering 26 (2002) 81109 97

Fig. 8. Spatial transects of TP for sampling dates during: (a) recirculation; and (b) harvest.

wetland area produced a straight line (R 2 =0.81 recirculation, 0.58 harvest; figure
not shown).
Fig. 12a shows that average TSS decreased asymptotically from 47 mg l 1 at the
inlet to 23 mg l 1 at the outlet during recirculation. The trend was similar but the
asymptote was greater (49 mg l 1) during harvest when inlet TSS levels were higher
(120 mg l 1) and HRT was shorter, TSS was reduced by 51% during recirculation
and by 59% during harvest. TSS plotted as a logarithmic function of wetland area
produced a straight line (R 2 =0.66 recirculation, 0.37 harvest; figure not shown).
Fig. 12b shows that average ISS decreased asymptotically from 36 mg l 1 at the
inlet to 13 mg l 1 at the outlet during recirculation. The trend and asymptote were
similar during harvest when inlet ISS levels were higher (70 mg l 1) and HRT was
shorter. These reductions equated to 62 and 70% during recirculation and harvest,
respectively. ISS plotted as a logarithmic function of wetland area produced a
98 D.R. Tilley et al. / Aquacultural Engineering 26 (2002) 81109

Fig. 9. Seasonally averaged TDS (a), pH (b) and DO 9 1 standard deviation as functions of wetland
area during recirculation and harvest.
D.R. Tilley et al. / Aquacultural Engineering 26 (2002) 81109 99

straight line (R 2 =0.79 recirculation, 0.40 harvest; figure not shown). Fig. 12c
shows that the average VSS concentration remained within a narrow range
throughout recirculation (9.5 16.7 mg l 1). At harvest however, influent VSS was
much greater (Fig. 12c), but the wetland was able to reduce average VSS by 48%,
from 49 mg l 1 at inlet to 28 mg l 1 at outlet.
Fig. 13 shows that average BOD5 did not vary by a large amount, but was
reduced from 10.6 mg l 1 at the inlet to 8.8 mg l 1 at the outlet (17%).

3.3. Wetland sizing

Properly sizing a constructed wetland for shrimp aquaculture depends on wetland

Fig. 10. Seasonally averaged nitrate (a) and total ammonia (b) with 9 1 standard deviation as functions
of wetland area during recirculation and harvest.
100 D.R. Tilley et al. / Aquacultural Engineering 26 (2002) 81109

Fig. 11. Seasonally averaged RP (a) and TP (b) 9 1 standard deviation as functions of wetland area
during recirculation and harvest.

area, areal uptake rate, hydraulic loading, inlet concentration and outflow target
levels. Therefore, the wetland design equation (Eq. (2)), which had as parameters,
inlet concentration, hydraulic loading and background concentration, was used to
determine the treatment area required to meet various target effluent concentra-
tions, given probable influent concentrations and typical hydraulic loading rates
(Table 3). The logarithmic decreases in ISS and TP concentrations shown in Figs.
7 and 8 support the use of Eq. (2) to determine area-dependent concentrations.
D.R. Tilley et al. / Aquacultural Engineering 26 (2002) 81109 101

Table 3 provides the estimates of kz for ISS and TP based on the C*s chosen.
Various influent concentrations and effluent targets were chosen (Table 4) to
represent plausible scenarios for a shrimp-wetland agroecosystem and entered into
Eq. (2) to derive the estimates of wetland treatment area needed. The high, low and
average Ci values chosen for ISS and TP were the maximum, the lower control
limit, and the process mean, respectively, observed during recirculation. The low,
high and average C0 values chosen for ISS were the lower control limit, the
regulatory standard, and the process mean, respectively. The low, high and average
C0 values chosen for TP were the lower control limit, the upper control limit
standard, and the process mean, respectively, observed during recirculation. Wet-

Fig. 12. Seasonally averaged TSS (a), ISS (b) and VSS (c) 9 1 standard deviation as functions of
wetland area during recirculation and harvest.
102 D.R. Tilley et al. / Aquacultural Engineering 26 (2002) 81109

Fig. 13. Seasonally averaged biological oxygen demand (c) 91 standard deviation as functions of
wetland area during recirculation and harvest combined.

land treatment area was expressed as the pond production area to wetland surface
area (P:W). This normalization provides a quick estimate of the land resources
required to integrate a constructed wetland into a shrimp production facility. Table
4 presents the P:W ratio for each of the three scenarios. Under the most difficult
operating parameters high loading and low targets the greatest amount of
wetland area was required for both ISS and TP (3.0 units of shrimp pond for each
1.0 unit of wetland treatment area) (Table 4). Similarly, TP required a P:W of 2.4
(Table 4). At the other end of the spectrum the least difficult operating condi-
tions no wetland treatment would be needed due to the high quality of the

Table 3
Approximation of areal rate constants (kz ) for TP and ISS with assumed parameter values for the
kC* model

Parameter Value Units Basis

x 0.078 Fraction First observation point along sampling path


qw 0.18 m per day Mean flow per wetland area
C*TP 0.12 mg l1 Lower control limit of SQC chart
C*ISS 1.0 mg l1 Lower control limit of SQC chart
C*x,TP 0.165 mg l1 First observation point along sampling path
Cx,ISS 7.60 mg l1 First observation point along sampling path
CI,TP 0.256 mg l1 Mean during recirculation
CI,ISS 31.6 mg l1 Mean during recirculation
kz,TP 2.56 m per day Eq. (3); mean observations during recirculation substituted
kz,ISS 3.50 m per day Eq. (3); mean observations during recirculation substituted
D.R. Tilley et al. / Aquacultural Engineering 26 (2002) 81109 103

Table 4
Estimates of pond-to-wetland ratio based on Eq. (2) for various influent and effluent criteria

Possible scenario CI C0 Pond:wetland

TP
High load, low target 0.60 0.12 2.4
Average load and target 0.26 0.16 12.0
Low load, high target 0.17 0.22 No wetland needed
ISS
High load, low target 90.0 1.1 3.0
Average load and target 30 6.4 12.2
Low load, high target 5 20 No wetland needed

influent water. For the expected water quality conditionsmean influent and mean
target the P:W ratio was 12.0 and 12.2 for TP and ISS, respectively (Table 4).

4. Discussion

Sustainable shrimp aquaculture can be achieved by integrating constructed


wetlands into facility operations as water filtration systems. The constructed
wetland that operated under mesohaline conditions at the full-scale LASAF per-
formed satisfactorily as a water treatment filter. The wetland demonstrated the
ability to reduce TP, TSS and ISS to levels significantly lower than either water
supply or pond effluent (Figs. 4 and 5). Ammonia levels were kept below 2.0 mg
l 1 during recirculation, but were slightly higher during harvest (Fig. 10b). Nitrate
levels in the wetland effluent changed little during the entire season and were almost
always below 0.4 mg l 1 (Fig. 10a). BOD5 in the wetland effluent was on average
3.2 mg l 1 less than pond effluent (Fig. 6) and was below 10 mg l 1 except for one
date (10/2/00).

4.1. Wetland sizing

Parameters of the k C* design model (Eq. (2)) presented by Kadlec and Knight
(1996) were determined based on the performance of the LASAF wetland. The
calibrated parameters were then used in Eq. (3) to determine the wetland size need
to treat TP and ISS to probable target levels. Both constituents required that the
ratio of pond surface to wetland surface (P:W) be approximately 12 if pond loading
and target levels were near the mean expectation (Table 4). Thus, a shrimp
aquaculture facility that integrates constructed wetlands as recirculation filters will
require that 7 8% of the production area be reserved for water treatment.
Treatment rate is sensitive to a wetlands HRT (Canning, 1988; Kulzer, 1989;
Stockdale, 1991; Rushton et al., 1995). In many applications of treatment wetlands,
HRTs on the order of 2 5 days have been deemed sufficient to achieve desired
104 D.R. Tilley et al. / Aquacultural Engineering 26 (2002) 81109

treatment. According to our assessment of the LASAF wetland, sufficient treatment


of TP and ISS occurred when residence time was less than or equal to 1 day.
However, there is a trade-off between space and residence time. A longer residence
time allows less wetland to be used to achieve similar treatment rates. However,
greatly extending the HRT beyond 15 20 days may present other water quality
problems such as high salinity, due to more evapotranspiration per unit of wetland
water volume, or extremely low DO concentrations, due to increased time for
wetland soil respiration. Additionally, there will obviously be operational factors of
the shrimp ponds themselves that need to be considered when managing the
hydrodynamics of the wetland.
Wherever shrimp aquaculture may be geographically located, water quality
standards will be different for various reasons (e.g. regulatory, ecological) and the
quality of water supply will vary. The parameter values used in the kC* model
were specific to the LASAF and likely would be appropriate for use in the
surrounding region. However, caution should be taken in using these specific
parameter values elsewhere in the world. Other parameters, such as BOD, DO and
ammonia, may be a part of a facilitys discharge permit. These design parameters
were not determined because: (1) at the LASAF these constituents were consistently
lower than typical regulatory requirements; and (2) at such low concentrations they
were either changed little by the wetland or increased slightly, as in the case of
ammonia (Fig. 10). The size of the LASAF wetland was more than doubled for the
2001 season and continued to be monitored (B. Dyson, personal communication) so
future work is planned to include other constituents such as ammonia, total
nitrogen and BOD5.

4.2. Plants as control agents

Although multiple seeding of salt marsh plants was undertaken in the first year,
the LASAF wetland became dominated by a freshwater plant species (T. latifolia)
tolerant of brackish water (B10 ppt). T. latifolia colonized the middle region of the
wetland (segments B D, see Fig. 1) and became the dominant plant species by the
second year of operation. More technical knowledge on the self-design ability of
ecological systems, such as wetlands, needs to be gained so it can be incorporated
into the design of constructed wetlands (Mitsch et al., 1998; Mitsch and Wilson,
1996). To increase the plant diversity of the constructed wetland and thus its
ecological resilience, provide ample source of seeding and a hydro-period (i.e. water
depth cycle) that mimics a natural wetland such as a nearby coastal marsh or
swamp. In the system studied here, salinity was closer to fresh than the optimum
(1015 ppt) for shrimp aquaculture. This however, enabled freshwater plants to
colonize and dominate the wetland.
The constancy of TDS ( 3.1 ppt) in the cattail-dominated region was an
amazing observation that has ramifications for controlling system salinity. In open
water areas of the wetland and the shrimp ponds themselves, TDS levels increased
appreciablypresumably due to high evaporation rates caused by large water
vapor saturation deficits and strong wind speeds. In effect the wetland vegetation
D.R. Tilley et al. / Aquacultural Engineering 26 (2002) 81109 105

adds a control element to the management of water salinity in shrimp aquaculture,


providing security against large fluctuations in salinity that could easily be observed
in natural water bodies used for supply.

4.3. Water quality

4.3.1. Nitrogen
Average total ammonia showed an increasing but asymptotic trend along the
wetlands hydraulic gradient (Fig. 10b) while average nitrate levels remained within
a narrow range (0.33 and 0.39 mg l 1, Fig. 10a). Paired t-test showed total
ammonia to be greater at the wetland outlet than at either the wetland influent or
water supply (Table 2). Since oxygen levels were pulled down by the high sediment
oxygen demand (SOD) (Badrinarayanan, 2001) there was not sufficient oxygen
available in the LASAF wetland for nitrification to occur at a noticeable rate,
which is often the case with wetlands (Mitsch and Gosselink, 1993). For ammonia
to accumulate the N-mineralization rate must be greater than the nitrification rate
(Martin and Reddy, 1997). The levels of ammonia ( 1 mg l 1) and nitrate (B 0.5
mg l 1) generated by the shrimp ponds were considerably less than levels tradition-
ally treated by constructed wetlands (Kadlec and Knight, 1996). Levels lower than
those observed in LASAF pond effluent may be extremely difficult to reach, unless
auxiliary resources are employed to boost oxygen concentrations and nitrification
rates. However, these low levels of total ammonia observed (1 2 mg N l 1) were
well below the levels reported toxic to L. 6annemei (3566 mg N l 1) by Lin and
Chen (2001).

4.3.2. Phosphorus and suspended solids


During recirculation nearly all of the reduction observed in TP and ISS occurred
within section A, which was sparsely vegetated and had no cattail plants. TP and
ISS were reduced by 31 and 76% within section A. The quiescent conditions
observed for this section of the wetland promoted settling of suspended sediment
particles, which in turn promoted adsorption of phosphorus to suspended particles.
Fig. 14 demonstrates the positive correlation between concentrations of TP and
TSS for both the recirculation (R 2 =0.27) and harvest (R 2 = 0.29) periods. For
every 10 ppm reduction in TSS during recirculation, TP decreased by 0.03 ppm
(Fig. 14a). Therefore, management of TP can be improved by creating a calm water
environment for the settling of suspended solids.

4.3.3. Oxygen
The wetlands demand for oxygen came mainly from the organic sediments
(SOD), not the water column (BOD) (Badrinarayanan, 2001). Support stems from
the fact that BOD5 levels, which were low (B10 mg l 1), did not drop appreciably
(  1.0 mg l 1) across the wetland (Fig. 13), but DO levels did drop ( 4 mg l 1)
from inlet to the cattail-dominated region (Fig. 9c). Badrinarayanan (2001) deter-
mined through modeling efforts that the SOD was of greater importance in
controlling DO than BOD. The wetlands DO levels created little concern for the
106 D.R. Tilley et al. / Aquacultural Engineering 26 (2002) 81109

shrimp production because levels generally increased before leaving the wetland. If
SOD increased to levels that significantly decreased DO, then removal of dead
organic matter might be considered as a management option. Controlled burns or
mechanized harvesting of the standing and dead biomass at the end of the growing
season may help control SOD.
The SOD was likely the greatest influence on the drop in pH. Wetlands are
proficient producers and accumulators of organic matter that creates SOD (Fig. 2c).
High SOD increases the probability for high respiration rates, which add CO2 to
the water column. The generated CO2 dissolves and subsequently adds bicarbonate
and H+ to the water column, lowering pH. Thus, managing SOD may also help
control pH.

5. Conclusions

Two years of operation of the constructed wetland as the recirculation filter for
the full-scale LASAF demonstrated that a constructed wetland could efficiently
reduce TP, TSS and ISS and maintain consistently low levels of BOD, salinity, total

Fig. 14. TP as a function of TSS during recirculation (a) and harvest (b).
D.R. Tilley et al. / Aquacultural Engineering 26 (2002) 81109 107

ammonia, and nitrate. Emergent macrophytes, such as T. latifolia, controlled


salinity fluctuations, which could provide the shrimp farmer a mechanism for
managing salinity. Although pH was reduced through the wetland, it remained
above 7.4 at the outlet. Parameter values for the kC* design model (Kadlec and
Knight, 1996) were determined and used to estimate the wetland treatment area
needed to reduce a given inlet concentration of ISS or TP to a target level. Both TP
and ISS required a pond to wetland ratio of 12 for mean loading rates and average
target levels.
Constructed wetlands will not only improve the quality of discharge water, they
will also increase the retention time of water on the farm, which reduces the
demand for intake water. Non-monetary ecological benefits are also gained from
the creation of a healthy, productive brackish marsh.

Acknowledgements

Gratitude is extended to the H.P. El Sauz Ranch for allowing full access to the
Loma Alta Shrimp Aquaculture Facility and for financially supporting this project
in conjunction with supplementary funds from Texas A&M University-Kingsville,
College of Engineering.

References

APHA, 1992. Standard Methods for the Examination of Water and Wastewater, 18th ed. American
Public Health Association, American Water Works Association, Water Environment Federation,
Washington, p. 560.
Bachand, P.A.M., Horne, A.J., 2000. Denitrification in constructed free-water surface wetlands: effects
of vegetation and temperature. Ecol. Eng. 14, 17 32.
Badrinarayanan, H., 2001. Effect of a mesohaline constructed wetland on water chemistry discharged
from a shrimp aquaculture facility in south Texas. M.S. thesis, Texas A&M University Kingsville,
83 pp.
Canning, D.J., 1988. Urban runoff water quality: effects and management options. Shorelands Technical
Advisory Paper No. 4, 2nd ed. Washington Department of Ecology, Olympia, WA.
Costa-Pierce, B.A., 1998. Preliminary investigation of an integrated aquacultue-wetland ecosystem using
tertiary-treated municipal wastewater in Los Angeles County, California. Ecol. Eng. 10, 341 354.
Dadi, S., 2001. Development of a phosphorus budget for the south Texas hypersaline lagoon, Baffin
Bay. M.S. Thesis, Texas A&M University-Kingsville, p. 70.
Day, J.W. Jr., Hall, C.A.S., Kemp, W.M., Yanez-Arancibia, A., 1989. Estuarine Ecology. John Wiley &
Sons, New York.
Funge-Smith, S.J., Briggs, M.R.P., 1998. Nutrient budgets in intensive shrimp ponds: implications for
sustainability. Aquaculture 164, 117 133.
Gearheart, R.A., 1990. Nitrogen removal at the Arcata constructed wetlands. Presentation at the Water
Pollution Control Federation Meeting, Washington DC.
Gersberg, R.M., Gearheart, R.A., Ives, M., 1989. Pathogen removal in constructed wetlands. In:
Hammer, D.A. (Ed.), Constructed Wetlands for Wastewater Treatment: Municipal, Industrial and
Agriculture. Lewis Publishers, Chelsea, MI.
Greenway, M., Woolley, A., 1999. Constructed wetlands in Queensland: performance efficiency and
nutrient bioaccumulation. Ecol. Eng. 12 (1 2), 39 55.
108 D.R. Tilley et al. / Aquacultural Engineering 26 (2002) 81109

Hach, 1991. HACH DR/2000 Spectrophotometer Procedures Manual. HACH Company, Loveland,
Colorado.
Kadlec, R.H., Knight, R.L., 1996. Treatment Wetlands. Lewis Publishers, New York.
Kivaisi, A.K., 2001. The potential for constructed wetlands for wastewater treatment and reuse in
developing countries: a review. Ecol. Eng. 16 (4), 545 560.
Knight, R.L., Payne, V.W.E. Jr., Borer, R.E., Clarke, R.A. Jr., Prieset, J.H., 2000. Constructed wetlands
for livestock wastewater management. Ecol. Eng. 15, 41 55.
Kulzer, L., 1989. Considerations for the Use of Wet Ponds for Water Quality Enhancement. Municipal-
ity of Metropolitan Seattle, Office of Water Quality, Seattle, WA.
Lin, C., 1995. Progression of intensive marine shrimp culture in Thailand. In: Browdy, C.L., Hopkins,
J.S. (Eds.), Swimming through Troubled Water: Special Session on Shrimp Farming. World
Aquaculture Society, Baton Rouge, pp. 157 163.
Lin, Y.C., Chen, J.C., 2001. Acute toxicity of ammonia on Litopenaeus vannamei Boone juveniles at
different salinity levels. J. Exp. Mar. Biol. Ecol. 259, 109 119.
Longstaff, B.J., Dennison, W.C., 1999. Seagrass survival during pulsed turbidity events: the effects of
light deprivation on the seagrasses Halodule pinifolia and Halophila ovalis. Aquat. Bot. 65 (1 4),
105 121.
Martin, J.F., Reddy, K.R., 1997. Interaction and spatial distribution of wetland nitrogen processes.
Ecol. Model. 105 (1), 1 21.
Mitsch, W.J., Gosselink, G.J., 1993. Wetlands, 2nd ed. Van Nostrand Reinhold, New York, p. 722.
Mitsch, W.J., Wilson, R.F., 1996. Improving the success of wetland creation and restoration with
know-how, time, and self-design. Ecological Applications 6 (1), 77 83.
Mitsch, W.J., Wu, X.Y., Nairn, R.W., Weihe, P.E., Wang, N.M., Deal, R., Boucher, C.E., 1998.
Creating and restoring wetlands a whole-ecosystem experiment in self-design. Bioscience 48 (12),
1019 1030.
Montgomery, D.C., 1991. Introduction to Statistical Quality Control, 2nd ed. John Wiley & Sons, New
York, p. 674.
National Aquaculture Association, 1998. U.S. Aquaculture and Environmental Stewardship. National
Aquaculture Association Environmental Subcommittee, Charles Town, West Virginia. http://
www.natlaquaculture.org/EnvirPaper.htm.
NOAA, 2002. National Oceanic and Atmospheric Administration, National Climatic Data Center.
http://lwf.ncdc.noaa.gov/oa/climate/stationlocator.html.
National Weather Service, 2001. National Oceanic and Atmospheric Administration. http://
www.nws.mbay.net/normrain.html.
Negroni, G., 2000. Management optimization and sustainable technologies for the treatment and
disposal/reuse of fish farm effluent with emphasis on constructed wetlands. World Aquacult. 31 (3),
16 19.
Nerella, S., Weaver, R.W., Lesikar, B.J., Persyn, R.A., 2000. Improvement of domestic wastewater
quality by subsurface flow constructed wetlands. Bioresour. Technol. 75, 19 25.
Paez-Osuna, F., 2001. The environmental impact of shrimp aquaculture: causes, effects, and mitigating
alternatives. Environ. Manag. 28 (1), 131 140.
Perkins, J., Hunter, C., 2000. Removal of enteric bacteria in a surface flow constructed wetland in
Yorkshire, England. Water Resour. 34 (6), 1941 1947.
Rushton, B., Miller, C., Hull, C., 1995. Residence time as a pollutant removal mechanism in stormwater
detention ponds. Proceedings of the 4th biennial stormwater research conference, Clearwater, FL,
Southwest Florida Water Management District, Brooksville, FL, pp. 210 221.
Samocha, T.M., Lawrence, A.L., 1995. Shrimp Farms Effluent Waters Environmental Impact and
Potential Treatment Methods. United States Japan Cooperative Program in Natural Resources
Technical Report No. 24, pp. 33 57.
Sansanayuth, P., Phadungchep, A., Ngammontha, S., Ngdngam, S., Sukasem, P., Hoshino, H.,
Ttabucanon, M.S., 1996. Shrimp pond effluent: pollution problems and treatment by constructed
wetlands. Water Sci. Technol. 34 (11), 93 98.
Serodes, J.B., Normand, D., 1999. Phosphorus removal in agricultural wastewater by a recently
constructed wetland. Can. J. Civ. Eng. 26, 305 311.
D.R. Tilley et al. / Aquacultural Engineering 26 (2002) 81109 109

Stockdale, E.C., 1991. Freshwater Wetlands, Urban Stormwater, and Nonpoint Pollution Control: A
Literature Review and Annotated Bibliography, 2nd ed. Washington State Department of Ecology,
Olympia, WA, p. 275.
TPWD, 2001. Texas Shrimp Fishery. Texas Parks and Wildlife Department, 74th Texas Legislature,
Austin, TX.
UNFAO, 2000. The state of world fisheries and aquaculture: 2000. United Nations Food and
Agricultural Organization, Editorial Group, FAO Information Division, Rome, Italy, http://
www.fao.org/DOCREP/003/X8002E/X8002E00.htm.
USDA, 1998. Census of Aquaculture. U.S. Deptartment of Agriculture, National Agricultural Statistics
Service, Washington DC.
USDA, 2001. Aquaculture outlook: supplement to livestock, dairy, and poultry situation and outlook.
U.S. Deptartment of Agriculture, Economic Research Service (LDP-AQS-13), Washington DC.
Valenti, C.E.M., 2000. Plant Establishment on a Brackish Water Constructed Wetland. MS research
project, Texas A&M University-Kingsville, p. 29.
Verhoeven, T.A., Meuleman, A.F.M., 1999. Wetlands for wastewater treatment: opportunities and
limitations. Ecol. Eng. 12, 5 12.
Vrhovsek, D., Kukanja, V., Bulc, T., 1996. Constructed wetland (CW) for industrial waste water
treatment. Water Resour. 30 (10), 2287 2292.
Zurayk, R., Nimah, M., Geha, Y., Rizk, C., 1997. Phosphorus retention in the soil matrix of constructed
wetlands. Commun. Soil Sci. Plant Anal. 28 (6/8), 521 535.

Potrebbero piacerti anche