JOURNAL OF THE
WORLD AQUACULTURE SOCIETY
Vol. 34, No. 3
A Novel Zero Discharge Intensive Seawater Recirculating
System for the Culture of Marine Fish
ILYA GELFAND, YORAM BARAK, ZIV EVEN-CHEN,
Department o f Animal Sciences, Faculty of Agricultural, Food and Environmentul Quality
Sciences, The Hebrew University of Jerusalem, Rehovot 76100 Israel
MICHAEL D. KROM
School o f Earth Sciences, Leeds University, Leeds LS2 9JT, United Kingdom
Israel Oceanographic and Limnological Research Ltd., National Center for Mariculture,
Eilat 881 12 Israel
Results are presented of a zero-discharge marine recirculating system used for the culture
of gilthead seabream Spurus uurata. Operation of the system without any discharge of water
and sludge was enabled by recirculation of emuent water through two separate treatment
loops, an aerobic trickling filter and a predominantly anoxic sedimentation basin, followed
by a fluidized bed reactor. The fish basin was stocked for the first 6 mo with red tilapia
Oreochrornis niloticus X 0 .
uureus at an initial density of 16 kg/m3. During this period salinity
was raised from 0 to 20 parts per thousand. Then, gilthead seabream, stocked at an initial
density of 21 kg/m3, replaced tilapia at day 167 and were cultured for an additional 225 d.
Non steady-state inorganic nitrogen transformations occurred as a result of these salinity
changes. After day 210, the system operated at all times with those water quality parameters
considered critical for successful operation of mariculture systems, within acceptable limits.
Thus ammonia, nitrite, and nitrate concentrations did not exceed 1.0-mg total ammonia-N/
L, 0.5-mg NO,-N/L and 50-mg NO,-N/L, respectively. Sulfide levels in the fish basin were
below detection limits and oxygen > 6 mg/L after the oxygen generator was added at day
315. Ammonia, produced in the fish basin and to a lesser extent in the sedimentation basin,
was converted to nitrate in the aerobic trickling filter. Nitrate removal took place in the
sedimentation basin and to a lesser extent in the fluidized bed reactor. Sludge, remaining in
the sedimentation basin at the end of the experimental period, accounted for 9.2% of the
total feed dry matter addition to the system. The system was disease-free for the entire year
and fish at harvest were of good quality. Water consumption for production of 1 kg of tilapia
was 93 L and 214 L for production of 1 kg o f gilthead seabream. Additional growth perfor-
mance data of gilthead seabream cultured in a similar but larger system are presented.
During 164 d of operation of the latter system, maximum stocking densities reached 50 kg/
m3 and fish biomass production was 2 7 . 7 kg/m3. Relatively poor fish survival and growth
resulted from occasional technical failures of this pilot system.
Culture of marine fish is conducted al-
most exclusively in cages and in flow-
through ponds and tanks adjacent to the sea.
The large volumes of seawater flowing
through such systems generally do not un-
dergo any effluent treatment. Discharge
from mariculture facilities into adjacent
I Corresponding author.
coastal waters has resulted in eutrophica-
tion, alteration of the natural flora and fauna
by modified nutrient ratios, release of heavy
metals and antibiotics, and pollution of the
wild fish stocks by mariculture “escapees”
(Gowen and Bradbury 1987; Iwama 1991;
Wu 1995; Christensen et al. 2000; Pearson
and Black 2001; Tovar et al. 2001). In
many countries, these environmental prob-
0 Copyright by the World Aquaculture Society 2003
ZERO DISCHARGE MARINE SYSTEM
lems, together with a shortage in suitable
sites, have formed a strong incentive to ex-
plore recirculating mariculture methods, en-
vironmentally sustainable and using less
Water purification in current recirculating
systems consists of devices for: 1) oxygen
supply; 2) removal of particulate organic
matter; 3) removal of ammonia; 4) removal
of CO,; 5) alkalinity control; and 6) disin-
fection. A daily addition of clean water is
required I ) to replace evaporation losses; 2)
to make up for water losses associated with
the discharge of organic matter from the
system; and particularly 3) to control ex-
cessive nitrate accumulation in the system
(Losordo and Westers 1994). A daily water
exchange of 5-20% of the total water vol-
ume, corresponding to water requirements
of 1 m3 or more for each kg of added feed,
is common practice in these systems (In-
golfsson 200 1 ), and requires marine farms
to be close to the sea.
An experimental nitrification-denitrifica-
tion recirculating system that allows zero
water-discharge has been employed for cul-
ture of freshwater fish (van Rijn 1996;
Shnel et al. 2002). Water treatment in this
system was based, in addition to the com-
monly used nitrification stage, on a com-
bination of anoxic degradation of solid and
dissolved organic fish waste and reduction
of nitrate to nitrogen gas by denitrification
(Aboutboul et al. 1995; van Rijn et al.
In the present study, results are presented
on the adaptation of the nitrification-deni-
trification principle for operation with sea-
water. Two pilot-scale, zero-discharge ma-
rine recirculation fishculture systems have
operated, one for a year and one for half a
year. Water quality conditions within the
fish basin were monitored on a daily basis.
The system was designed to convert the
ammonia to nitrate within the nitrifying
trickling filter and to remove the nitrate by
denitrification within the anoxic loop. The
rates of these processes in each module
were determined by inflow-outflow differ-
ences in nutrient concentrations. Where ap-
propriate, additional measurements and ex-
periments were carried out to provide ad-
ditional data on the in-siru processes occur-
ring within the system. Preliminary results
on growth performance of gilthead sea-
bream in the two experimental systems are
Materials and Methods
The pilot plant consisted of a fish culture
tank, a trickling filter, a sedimentation ba-
sin, and a fluidized bed reactor. Tap water
compensated for water losses resulting from
evaporation and leakage. The system was
designed and operated so as to allow quan-
tification of biogeochemical processes, rath-
er than to maximize fish output and biofilter
design efficiency. Red tilapia hybrids Or-
eochromis niloticus X 0. aureus were cul-
tured for the first 167 d of operation and
were replaced by gilthead seabream Sparus
aurata L. The latter marine fish was cul-
tured for an additional 225 d.
Experimental Fish Culture System
A schematic presentation of the closed
recirculating fish culture system, situated at
the Faculty of Agriculture, Rehovot, Israel,
is presented in Fig. 1. The main compo-
nents of the system are: 1) a round polyeth-
ylene, fish culture basin with a sloping bot-
tom (upper diameter: 2.1 m; volume: 2.3
m'; Aquatic Eco-Systems Inc., Apopka,
Florida, USA; catalogue number: TP650);
2) a polyethylene sedimentation basin (di-
mensions: 1.8-m length X 0.8-m width X
0.7-m depth; working volume: 0.4-0.5 m3);
3) a self-constructed trickling filter; and 4)
a self-constructed fluidized bed reactor
(FBR). The trickling filter (dimensions: 1.2-
m length X 0.6-m width X 1.4-m height;
volume: 1 m3) contained PVC cross-flow
medium with a specific surface area of 240
m2/m7 (Jerushalmi Ltd., Ashdod, Israel).
The FBR (total height: 198 cm) consisted
of a Perspex column that was widened at
the top (diameter column: 6.1 cm; diameter
upper 25 cm of the column: 23 cm). The
GELFAND ET AL.
FIGURE 1. Schematic presentation of the zero-discharge mariculture system at Rehovot (not to scale).
total working volume of the reactor was
6.25 L. The reactor was filled with 1.5 kg
of sand (> 97% SO,) serving as bacterial
carrier material with a diameter ranging
from 0.8-1.5 mm (at most 15% of grains >
1.4 mm and 10% of grains < 0.85 mm).
Water from the upper layers in the fish ba-
sin was pumped with a 1/3-hp magnetic
drive pump (model: TE-5.5 MD-HC, Little
Giant Pump Co.,
Oklahoma, USA) over the trickling filter at
a rate of ? 30 Wmin (hydraulic loading:
60.0 m/d). Water was continuously with-
drawn into the sedimentation basin by grav-
itation through a standpipe from the bot-
tom-center of the fish culture basin fitted
with a double drain (Aquatic Eco-Systems
Inc., Apopka, Florida, USA, catalogue
number: D2). Water from the upper layers
of the sedimentation basin was pumped
with a 65-W submersible pump (Model
P5000, Messner Ltd., Kalletal, Germany) at
a rate of 5-7 L/min into the FBR (hydraulic
loadings of FBR: 107-150 m/h) through a
vertical pipe extending from the top to 2 3
cm above the base of the reactor. From the
FBR, water was returned by gravity to the
fish basin after passing a small, rectangular
sedimentation basin (0.96 X 0.5 1 m; Aquat-
ic Eco-Systems Inc., Apopka, Florida,
USA, catalogue number: FFPT) for remov-
al of particulate matter. Once a day, organic
matter, captured in this small sedimentation
basin, was diverted back into the main sed-
Air was added by means of a 0.55-kW
air blower (model: R4110-2, Gast Manu-
facturing Inc., Benton Harbor, Michigan,
USA) and diffused into the fish basin by
means of three airlifts evenly distributed
along the perimeter of the fish basin. At day
315 into the experimental period, pure ox-
ygen generated by a 0.15-kW oxygen gen-
erator (model: 1498, SeQual Technologies,
San Diego, California, USA) was diffused
into the fish basin by means of airstones.
Dissolved and suspended particles were re-
moved from the water in the fish basin by
means of a foam fractionator (model:
TFSAZ, Top Fathom Ltd, Hudsonville,
Michigan, USA) for which inlet water was
pumped from the upper water layers in the
fish basin at a rate of 30 L/min. Effluent
water was returned to the fish basin while
foam produced in the fractionator was di-
verted into the sedimentation basin. Water
velocity in the fish basin (approximately 20
cm/sec in the upper water layer close to the
basin wall) was generated by: 1) diverting
some of the trickling filter inlet stream di-
rectly into the fish basin perpendicularly to
the basin’s radius; and 2) similarly directing
ZERO DISCHARGE MARINE SYSTEM
the outflows from the airlifts and foam frac-
Sulinity o f System
The system was initially filled with tap
water. During the tilapia culture period, sa-
linity was gradually increased by addition
of sea salt (Red Sea pHarm Ltd., Eilat, Is-
rael). Salinity levels of the culture water
were: 0 ppt until day 90, 5 ppt (day 90-
105), 10 ppt (day 105-125), 15 ppt (day
125-135), and 20 ppt (from day 135 on-
ward). Salinity was kept at 20 2 2 ppt dur-
ing culture of gilthead seabream.
Fish and Feeding
Tilapia Oreochromis niloticus X 0. au-
reus were fed daily with pellets (35% pro-
tein, 5% fat; Matmor, Ashkalon, Israel) at a
rate of 1.5-2% of the total fish biomass.
Gilthead seabream Sparus auratu L. were
fed at a rate of 1-1.5% of the total fish bio-
mass with pellets containing 45% protein
and 19% fat (Matmor, Ashkalon, Israel).
Daily nitrogen input to the system was cal-
culated according to equation [l]. Nitrogen
excretion rates (NE) for tilapia and gilthead
seabream were estimated to be 6040%
(Porter et al. 1987; Krom and Neon 1989;
van Rijn and Shilo 1989; Lupatsch and Kis-
NI = FI * PF * 0.16 * NE
Where: NI = Daily nitrogen input; FI =
Daily feed input; PF = Protein content of
feed; 0.16 = Nitrogen content in protein;
NE = Nitrogen excretion (non-assimilated
nitrogen and assimilated excreted nitrogen).
Oxygen and temperature were measured
with a YSI (model 57) temperature/oxygen
probe (Yellow Springs Instrument Compa-
ny, Yellow Springs, Ohio, USA). Salinity
was monitored with a refractometer (model:
S- IOE, Atago, Tokyo, Japan). Total am-
monia (NH, and NH,+), hereafter referred
to as TAN (total ammonia-nitrogen), was
determined by oxidation with phenol-hy-
pochlorite as described by Scheiner (1976).
Nitrite was determined by reaction with sul-
fanilamide according to Strickland and Par-
sons (1968), and nitrate was measured with
the Szechrome Nas reagent (Applied Re-
search Institute, Ben Gurion University of
the Negev, Beer Sheva, Israel). The detec-
tion limit of the nitrate method was 0.5 mgl
L at a precision level of 5%. In order to
prevent salt interference, nitrate analyses
were conducted on water samples that were
diluted up to 100-fold with distilled water.
Total alkalinity was determined by titration
with hydrochloric acid according to Parsons
et al. (1984). Sulfate was measured using a
Quick Chem Ion Analyzer (Lachat Instru-
ments, Milwaukee, Wisconsin, USA) and
sulfide was determined by the methylene
blue method according to Cline (1 969). At
the end of the tilapia and gilthead seabream
culture periods, randomly selected fish were
analyzed by a certified analytical laboratory
(Bactochem, Nes Ziona, Israel) for bacterial
contamination (total bacterial counts as well
as detection for Coliforms, Staphyloccoccus
aureus, Clostridium species, and Salmonel-
la species), heavy metals content (zinc and
copper), and for smell and taste (organolep-
After drainage of the system at the end
of the experimental period, dry weight of
sludge in the sedimentatioddigestion basin
was calculated by determining the volume
of the sludge layer in the basin and the dry
weight of subsamples obtained from this
The larger system (a 10,000-L fish tank,
a 2,000-L sedimentation basin, a 200-L
FBR, and a 4-m' trickling filter), at the Na-
tional Center for Mariculture, Eilat, Israel,
was similar to the small system with the
exception of the following features: up to
one water volume of the system was passed
through an ozonator daily; aeration was en-
riched with pure oxygen from a liquid ox-
ygen source; and in summer a 1.0-hp blow-
er delivered air upward into the trickling
GELFAND ET AL.
O , "
00 0 "
- 9 r
0 $ 0
FIGURE 2. Temperature (open circles) and oxygen
concentrations (closed circles) in the fish basin over
the experimental period. Measurements were car-
ried out between 0800 and lo00 h.
filter for cooling. The salinity of this system
was maintained at 36 ppt during the entire
Water quality parameters were deter-
mined in the Rehovot pilot-plant system
from 29 April 1999 until 28 June 2000, a
total of 394 d. Measurements were made in
the fish basin, as a measure of the water
quality conditions maintained by the system
and seen by the fish. In addition, measure-
ments were made at the inlet and outlet of
each module to provide information on the
net processes occurring within each mod-
ule. Fish growth parameters are presented
for this system as well as the Eilat system
where the reported period lasted from 1
February until 17 July 2001.
Monitoring of Water Quality Parameters
in the Fish Basin
Oxygen concentrations in the fish basin
(measured between 0800 and 1000 h)
ranged from 4.0 to 7.5 mgL (Fig. 2). Fol-
lowing the installation of the oxygen gen-
erator at day 315, oxygen concentrations
did not drop below 6 mg/L (Fig. 2). Tem-
perature ranged from 19 to 30 C (Fig. 2)
FIGURE 3. Concentrations o f ammonia, nitrite, and
nitrate in the Jish basin during the experimenial pe-
riod. Changes in saliniiy are presented in the top
frame. Note the changes in scale between the differ-
ent inorganic nitrogen compounds. Arrow indicates
start of gilthead seabream culture period.
and showed a simple seasonal trend. The
pH of the water in the fish basin was stable
at 7.7 2 0.3 (data not shown).
Concentrations of TAN, nitrite, and ni-
trate in the fish basin are shown in Fig. 3.
During the start-up phase (day 1-210), fluc-
tuations in water quality parameters were
seen. After day 210 until the end of the
sampling period, inorganic nitrogen and
phosphate concentrations remained stable
within rather close limits (Table 1) and were
well within the levels that are considered
acceptable for optimal fish growth in other
fish culture systems. As a result of these
TABLE 1. Average ammonia, nitrite, niirate, and
phosphate concentraiions during seabream culture
period (day 210 onward).
TAN (mg NL)
Nitrite (mg NL)
Nitrate (mg NL)
Phosphate (mg PL)
a Standard deviations are given in parenthesis.
ZERO DISCHARGE MARINE SYSTEM
Fluidized bed reactor
FlGUKE 4. Differences between inlet and outlet values o f ammonia, nitrite and nitrute in the trickling filter.
sedimentation basin, and jhidized bed reactor during the experimental period. Positive vulues indicate re-
moval. The average dail-v removal/addition is obtained by multiplving the concentration differences by the
water flux, which were up to six times higher in the trickling ,filter than in sedimentation basin and ,fluidized
measurements, two periods were recog-
nized for the system-a
day 1-210 and a steady-state phase from
day 2 10-394 (and beyond). High TAN con-
centrations on days 95, 110, and 137 were
associated with imposed salinity changes
(see insert Fig. 3) from 0 to 5, 10 and 20
ppt, respectively. At all other times the
TAN concentration remained at or below
1.0 mg/L. After the system stabilized, the
average TAN value was 0.49 t 0.25 m g L
Nitrite concentration sharply increased up
to 8 mg NO,-N/L after the salinity was
raised from 15 to 20 ppt. and then de-
creased to background levels. Maximum ni-
trate concentrations of approximately 100
'mg N03-N/L were reached at day 120, dur-
ing the start-up phase. Thereafter, when the
system was operated at maximum water sa-
linities, nitrate concentrations decreased to
27.91 -+ 5.25 mg/L N03-N during the
Differences in concentrations of TAN
and nitrate between inlet and outlet of each
start-up phase from
of the modules (trickling filter, sedimenta-
tion basin, and FBR) were determined and
used to provide information on the net bio-
geochemical processes occurring within
each individual module during its routine
operation (Fig. 4). The trickling filter re-
moved TAN on almost all sampling days.
In the sedimentation basin, TAN accumu-
lated during the steady-state phase. TAN re-
moval and addition in the fluidized bed re-
actor were mixed. Nitrate was removed in
the sedimentation basin and FBR on most
sampling days. In all these results there ap-
peared to be systematic changes in TAN
and nitrate flux which occurred over a pe-
riod of several days-weeks. These longer
frequency fluctuations changed the magni-
tude of the net flux though generally did not
change the overall trend of nutrient uptake
or accumulation, with the exception of
TAN in the FBR which went from accu-
mulation to net removal after day 270. Ni-
trate removal/accumulation values in the TF
GELFAND E T AL.
y = 0.1754X + 1.1669
y - -0.0447X + 1.4612
y =. 0.006ZX + 0.4721
4, .- -
k’ - 0.1883
y = 0.1788X + 21.3844
20 20 .
y = 0.1788X + 21.3844
(not shown) fell within the analytical pre-
cision of the method used.
Short-term experiment, in which changes
in TAN and nitrate concentrations were de-
termined while one of the treatment loops
was stopped, allowed estimating the rates
of nutrient conversions specifically by each
treatment component. At days 304, 305,
and 306 (gilthead seabream culture period;
see Fig. 3 for nutrient levels at the time),
feeding was stopped and changes in inor-
ganic nitrogen in the fish basin were ex-
amined, first with a fully operative system
and then after disconnecting either the
trickling filter or the sedimentation basin
nitrate by the trickling filter and sedimentation ba-
sin-fluidized bed reactor (FBR) as measured by un-
coupling these treatment stages at days 305 and
306. Estimated daily nitrogen input to the system
before uncoupling was 25.5 2 3.5 g Nand was cal-
culated according to equation [ I ] where: FI = 500
g and NE = 60-809i.
2. RemovaVaddition (g N/d) of ammonia and
Sedimentation basin and FBR
TAN = T o t a l ammonia-nitrogen.
+ sign stands for accumulation.
and FBR. With both treatment stages con-
nected (Fig. SA), TAN in the fish basin de-
creased at a rate of 0.05 mg/L per h (with
total water volume of 2.8 m3 this equals
3.36-g TAN/d) while it increased at a rate
of 0.18 mgL per h (12.09-g TAN/d) when
the trickling filter was uncoupled (Fig. 5B).
The apparent TAN removal by the trickling
filter (240 mz), therefore, can be estimated
at 15.45-g TAN/d (specific removal rate:
0.064-g TAN/mz per d). Nitrate in the fish
basin decreased at a rate of 0.33- and 0.67-
mg NO3-N/h (22.18- and 45.02-g N03-N/
d) with and without the trickling filter, re-
spectively. It follows that the apparent ni-
trate production by the trickling filter can
be estimated at 22.84-g N03-N/d (specific
production rate: 0.095-g N03-N/mZ per d).
Uncoupling o f the sedimentation basin and
FBR (Fig. 5C) did not affect the TAN con-
centrations in the fish basin. Nitrate de-
creased at a rate of 0.33 mg/L per h (22.18-
g NO3-N/d) before uncoupling and in-
creased at a rate of 0.18-mg N03-N/L per
h (with a fish basin volume of 2.3 m3 this
equals 9.94-g NO,-N/d) after uncoupling
this treatment loop. Therefore, the apparent
nitrate removal in the sedimentation basin
and FBR can be estimated at 32.12-g NO3-
N/d. Feed addition during the period was
500 g/d, corresponding to an estimated dai-
ly nitrogen input of 22-29 g of N (see equa-
tion [l]). Table 2 sums up the main rates
calculated from these “uncoupling” exper-
ZERO DISCHARGE MARINE SYSTEM
over a diurnal cycle at day 382 o f the experimental period.
6. Oxygen concentrations in the fish basin and outlets o f sedimentation basin andjuidized bed reactor
Diurnal Changes within the System
Typical oxygen concentrations in the fish
basin, sedimentation basin, and FBR over a
diurnal cycle are presented in Fig. 6. At this
particular sampling day (day 382), oxygen
concentrations in the fish basin and the aer-
obic loop followed a diurnal pattern with
the morning concentration of over 9 mg/L
decreasing during feeding time ( 1200 and
1500 h) and rebounding at night. Oxygen
concentrations in the water overlying the
sedimentation basin were 4-7 mg/L lower
than in the fish basin, and a further decrease
was observed in the FBR.
The principal pH buffer in seawater is the
FIGURE 7. Alkalinity values at vurious sampling
points in the system measured at different times at
day 382 o f the experimental period.
carbonate-bicarbonate system that was de-
termined as total alkalinity in this study. To-
tal alkalinity was typically lower in the fish
basin and trickling filter than in the sedi-
mentation basin and FBR (Fig. 7).
No organic matter was discharged from
the system during the entire period. The fish
waste and uneaten feed from the fish basin
were drained into the sedimentation basin
for biological degradation. The extent of or-
ganic matter degradation in this basin was
estimated by quantifying the sludge remain-
ing in this basin at the end of the experi-
mental period. Total dry weight of the
sludge was 25.16 kg, accounting for 9.2%
of the 273.8 kg of total feed added to the
system in the examined period.
Fish Pe~ormance-Rehovot System
Performance data of the fish are present-
ed in Table 3. Tilapia growth (during the
first 5 mo of operation) showed two distinct
phases. At low salinities, during the first 2
mo, tilapia showed a specific daily growth
rate of 1.44%, whereas at higher salinities,
during the third, fourth and fifth months,
their daily growth rate decreased to 0.43%
(not shown). Survival of tilapia and gilt-
head seabream in the Rehovot system was
GELFAND ET AL.
TAnLE 3. Growth peformance parameters o f tilapia arid gilthead seabream in the Rehovot system und o f
gilthead seabream in the Eilat system.
seabream Tilapia Rehovot Eilat
Growth period (d)
Initial number of fish
Final number of fish
Initial average weight (g)
Final average weight (g)
I ,067-2.0 19"
Total fish biomass produced (kg)
Specific growth rate (Wd)
Specific yield (kg/m')'
Food conversion coefticient (FCR)
Average food input (g/d)
Freshwater supply (Wd)
Specific water consumption (Wkg of fish)
92.6 214.0 213 (194)
Fish of the same cohort were added gradually in several installments until day 62 into the culture period.
Number in parentheses refers to figures that include also dead fish output (ie., fish biomass that was produced
but not sold).
' Yields were based on fish basin volumes of 2.3 m' and 10 rn' for the Rehovot and Eilat system, respectively.
low, mainly because of oxygen deficiency
in the water following electricity power
failures, which occurred in this pilot plant
on five different occasions. Chronic mor-
tality was experienced during the first 5 mo
of gilthead seabream culture. Oxygen levels
during this period often dropped below 5
mg/L (Fig. 2) and were thought to be the
cause of the observed mortalities, because
this chronic fish mortality ceased when the
fish basin was operated at higher oxygen
levels by addition of an oxygen generator
at day 3 15.
Bacteria and heavy metal content of the
fish were within acceptable ranges. No off-
flavors were detectable in either tilapia or
Specific water consumption (Table 3),
expressed as the amount of water used for
production of 1 kg of fish, was as low as
92.6 L and 214 L during tilapia and gilt-
head seabream culture, respectively. Water
was added mainly to compensate for evap-
oration losses. The difference in water con-
sumption between tilapia and gilthead sea-
bream is partly attributable to the relatively
low gilthead seabream production as com-
pared to tilapia production. In addition, dur-
ing gilthead seabream production, 1,300 L
of water was lost due to a rupture of one
of the water pipes.
The average specific growth rate of the
fish during 164 d in the Eilat system (Table
3) was within the acceptable range, but sur-
vival was poor due to parasitic infections
(monogenean worms and Amyloodinium
ocellutum) and their treatments. Nearly a
third of the fish died because of these in-
fections. Water use for production of 1 kg
of fish was in the same range as in the Re-
hovot system (213 L/kg fish). It should be
noted that in the Eilat system, the evapo-
ration rates during the summer months were
as high as 4% of the total system-volume
per day. The results of the Eilat system with
respect to water quality and transformation
of organic matter and nutrients were largely
similar to those of the Rehovot system
(Neori et al. 2002).
Marine Zero-Discharge System
Most commercial marine recirculating
systems are operated with partial flow-
ZERO DISCHARGE MARINE SYSTEM
through of seawater and with sludge dis-
charge into the sea. Although marine sys-
tems operated with very low water ex-
change rates are mentioned in the “grey lit-
erature,’’ studies on true zero-discharge,
intensive marine recirculating systems are
scarce. Among these studies, some involve
laboratory-scale systems used for experi-
mental rather than production purposes
(Whitson et al. 1993; Thoman et al. 2001),
while others involve systems for low-den-
sity shrimp production (Turk et al. 1997;
Menasveta et al. 2001). In addition, several
public marine aquariums are operated in a
closed mode (Grguric et al. 2000). In other
recent systems, the anoxic conditions nec-
essary for denitrification are attained by de-
oxygenation with nitrogen gas and by add-
ing labile carbon additives such as alcohols
(Grguric et al. 2000; Menavesta et al.
2001). The unique feature of the culture
system developed in this study is that the
organic matter wasted by the fish consumes
oxygen and serves as a sole carbon source
for the denitrification process.
Sa It A duptution
Salinity is easily maintained in a zero-
discharge system. Since the design of the
system was based on a freshwater zero-dis-
charge system (van Rijn and Rivera 1990;
Arbiv and van Rijn 1995; Shnel et al.
2002), we chose to gradually adapt the sys-
tem from fresh to salt water for verification
purposes. A salinity of 20 -+ 2.0 ppt was
chosen as the final salinity as this is close
to the physiological salinity of gilthead sea-
bream (Y. Zohar, University of Maryland,
personal communication). Adaptation of the
system from fresh water to salt water was
accompanied by transient accumulations of
TAN, nitrite, and nitrate in the system. In
particular, nitrite concentrations were dra-
matically influenced by salinity changes.
Based on the presented data, we were un-
able to discern the processes underlying this
nitrite accumulation. It is possible that the
high intermediate nitrite accumulation was
caused by a shift in the nitrite-oxidizing
population as a result of salinity changes or,
alternatively, by a salinity adaptation of the
existing nitrite-oxidizers. In addition, nitrite
might have been produced from incomplete
reduction of nitrate by dissimilatory nitrate
reducers as a result of salinity changes (van
Rijn and Rivera 1990; Park et al. 2001).
The system reached steady state after day
2 10 and water quality fluctuations thereafter
were small. The start-up period for the Eilat
system, which took place in 36 ppt seawa-
ter, lasted about 3 mo.
Water Quality Fluctuations within the
Those water quality parameters that are
considered critical for the successful oper-
ation of a mariculture system, including ox-
ygen (with oxygen enrichment), the differ-
ent N forms and even hydrogen sulfide, re-
mained stable and within acceptable levels
in the fishpond throughout the operation of
the plant, following the start-up phase. Ni-
trate, whose unacceptably high levels re-
quire water exchange in conventional inten-
sive recirculating systems (Losordo and
Westers 1994), remained stable below 30
mg N/L during steady-state phase of this
system. Hydrogen sulfide concentrations
built up to substantial and potentially toxic
levels (> 6 mM) in the lower levels at the
early section of the sedimentation basin.
However the fish were never exposed to
this toxin due to its nearly complete remov-
al in the posterior section of the basin, and
the removal by the FBR of the minute
amounts of sulfide that did exit the sedi-
mentation basin (Fig. 8 from Cytryn et al.,
in press; Neori et al., unpublished). An iron
oxyhydroxide sensor was developed and
tested, which was designed to further iso-
late the anoxic loop from the fish tank in
the unlikely event that any hydrogen sulfide
passed through the FBR and moved to-
wards the fish tank (Poulton et al. 2002).
Understanding the Biogeochemical
Processes Occurring within each Module
By examining the chemical changes
across each module during routine opera-
GELFAND ET AL.
Inlet Center Outlet Inlet Center Outlet
FIGURE 8. Values for ammonia, nitrate, sulfate, and suljde from horizontal (inlet, center, and outlet) and
vertical (T-top, M-middle, and 8-bottom) sampling points in the digestion basin (adup ted from Cytryn et
tion, it is possible to determine the domi-
nant in-situ microbial and other processes
in each module and the relative importance
of each module in controlling the water
quality of the system as a whole. This pro-
cedure has been used by us previously to
understand the operation of algal-based
mariculture systems (Krom and Neori 1989;
Krom et al. 1995; Ellner et al. 1996).
From the results of both the in-situ
changes and the uncoupling experiment, it
is evident that, as in other intensive recir-
culating fishculture systems, TAN was re-
moved and converted quantitatively to ni-
trite and nitrate by the trickling filter in the
aerobic loop of the system. The nitrification
process also consumed the alkalinity of the
water passing the trickling filter.
Unexplained long-term systematic fluc-
tuations in the TAN levels (with minima at
around day 250 and 360 and a maximum
at around day 300) prevent the calculation
of a single net TAN oxidation rate for the
entire steady-state phase in the whole sys-
tem. The specific TAN removal rate by the
trickling filter, as determined during the un-
coupling experiments, was relatively low
(0.064-g TAN/m2 per d). This was expected
since an oversized trickling filter was de-
signed as a precaution. At an average daily
feeding rate of 561 g during the gilthead
seabream period (see Table 3), the expected
daily nitrogen loading of the system was 28
2 4 g (according to equation #1), which
corresponds to a designed specific TAN re-
moval rate by this trickling filter of only
0.12-g TAN/m2 per d. In the Eilat system,
the designed specific nitrification rate was
0.3-g TAN/m2 per d (Neori et al., unpub-
lished). Specific TAN removal rate in trick-
ling filters depends to a large extent on the
hydraulic loading rate of the filter (size of
the trickling filter and water flux) as well as
on the ambient TAN concentrations in the
system. Specific removal rates of between
0.1-0.8-g TAN/m2 per d are commonly
ZERO DISCHARGE MARINE SYSTEM
found in freshwater nitrifying filters (van
Rijn and Rivera 1990; Kamstra et al. 1998;
Lekang and Kleppe 2000). At extremely
high hydraulic loading rates and high am-
bient TAN concentrations, values as high as
3.9-g TAN/m2 per d have been reported
(Greiner and Timmons 1998). Published
data from mariculture are scarce. Reported
area-specific nitrification rates in seawater
were up to 0.28-g N/m2 per d in Nijhof and
Bovendeur (1990) and up to 0.74-g N/m2
per d in Dvir et al. (1999).
The unique design feature that allows the
studied system to operate with zero efflu-
ents is the anoxic loop in which organic
waste is digested and nitrate denitrified. Cy-
tryn et al. (2003) have provided a more de-
tailed picture of the processes location
within this basin in January 2001 (Fig. 8).
The sludge accumulated in the basin was
structured in a classic sequence of biogeo-
chemical zones (Froelich et al. 1979). In the
uppermost layers, still slightly oxic, normal
aerobic respiration occurred. Beneath this
layer, nitrate reduction occurred, the bottom
layer nitrate was used up, and sulfate re-
duction predominated. Genetic DGGE sig-
natures characteristic of nitrate and sulfate
reducing bacteria were found at the relevant
depths in this sludge (Cytryn et al. 2003).
The outflow of sulfide from the posterior
end of the sedimentation basin was rather
small. This can be explained by the fact that
some sulfide, which diffused into the upper
layers of the basin, is removed by microbial
oxidation with either oxygen or nitrate as
electron acceptors. Such a process, which
also occurs in the FBR (Cytryn et al., in
preparation), contributed to the overall de-
nitrification in the system. The processes of
anoxic respiration in general, and sulfate re-
duction in particular, result in the produc-
tion of titration alkalinity (Froelich et al.
1979). The increase in alkalinity within the
anoxic loop was thus expected.
Initially, it was expected that sludge di-
gestion would occur in the sedimentation
basin and most of the nitrate reduction in
the FBR. Yet, in practice, most of the nitrate
was denitrified already in the sedimentation
basin. The major contribution of the FBR
to the operation of the system was the re-
moval of sulfide by microbial oxidation be-
fore it reached the fish basin. This allowed
this pilot system to operate without the use
of the iron oxide cartridge (Poulton et al.
2002), which was part of the original de-
The sludge produced in the system was
highly digestible. Only 9.2% of the feed
solids was recovered as sludge in the sedi-
mentation basin at the end of the experi-
mental period. This was similar to a fresh-
water zero-discharge system, where around
9% of the total added feed was recovered
as sludge after 244 d of operation (van Rijn
et al. 1995). The system was operated in
total for 3 yr including the 394 d considered
in this study without needing to discharge
any sludge externally. Similar to those de-
veloped in freshwater systems (van Rijn et
al. 1995; van Rijn and Nussinovitch 1997),
models are presently being developed that
are aimed at quantifying the long-term
sludge accumulation in this mariculture sys-
tem based on short-term degradation char-
acteristics of the sludge.
Technical failures of the Rehovot system,
related to lack of electricity backup and ini-
tially insufficient oxygen supply, resulted in
low survival rates of fish, especially gilt-
head seabream. It was impossible, there-
fore, to properly evaluate the system in
terms of fish growth performance. Howev-
er, observations in both systems pointed to
the fact that culture conditions allowed for
proper fish growth. Apart from the above-
mentioned mortalities, we found no evi-
dence for water quality related mortalities.
Fish at harvest were of good quality and
were free of microbial contamination. Fur-
thermore, tilapia growth at low salinities
was within the acceptable range for this
species. In the larger Eilat system, survival
GELFAND ET AL.
rate of gilthead seabream was also low.
Here, parasites introduced with newly
stocked fish, and inexperience with chemi-
cal therapy in closed systems, affected the
survival and growth performance of gilt-
head seabream. Growth trials, aimed at de-
termining the true fish growth potential of
this zero-discharge system, are in progress.
Summary and Conclusions
The present study demonstrates that at
gilthead seabream densities of up to 50 kgl
m3, all examined water quality parameters,
including inorganic nitrogen concentra-
tions, were maintained within acceptable
ranges without the need for discharge of
neither water nor sludge. The nitrification
filter converted the entire production of
TAN to nitrate. This nitrate was completely
denitrified and most of the sludge was bro-
ken down in the sedimentation basin, and
some additional nitrate removal occurred in
the FBR. Most of the sulfide that was pro-
duced in the anaerobic zone of the sludge
was microbially oxidized before the water
left the digestion basin. Any remaining sul-
fide was removed by the same process with-
in the FBR.
The zero-discharge mariculture technol-
ogy demonstrated here provides the possi-
bility to culture marine organisms indepen-
dently of distance from the sea and climate.
It avoids the heating and cooling of make-
up water and diminishes the risk of intake
of pathogens and pollutants. A zero dis-
charge technology also avoids environmen-
tal pollution of most kinds to receiving wa-
ter bodies, soil, and ground water. Finally,
a zero discharge system allows the enforce-
ment of strict biosecurity principles for
avoidance of diseases and escapes.
We thank Mr. Israel Snir for his invalu-
able assistance during design and construc-
tion of both pilot-plants. Thanks are due to
Ms. Orit Dvir and Dr. Yossi Tal for assis-
tance in the daily operation of the system.
The svstem in Eilat was ouerated bv Eng.
Vladimir Odinzev, helped by the teams of
nursery, engineering, nutrition, pathology,
and maintenance departments at the Israeli
National Center for Mariculture. This work
was supported by grant FAIR-CT98-4 160
of the Fisheries Directorate-General, Euro-
Aboutboul, Y., R. Arbiv, and J. van Rijn. 1995. An-
aerobic treatment of intensive fish culture efflu-
ents: volatile fatty acid mediated denitrification.
Arbiv, R. and J. van Rijn. 1995. Performance of a
treatment system for inorganic nitrogen removal
from intensive aquaculture systems. Aquacultural
Engineering 14: 189-203.
Christensen, P. B., S. Rysgaard, N. P. Sloth, T.
Dalsgaard, and S. Schwaeter. 2000. Sediment
mineralization, nutrient fluxes, denitrification and
dissimilatory nitrate reduction to ammonia in an
estuarine fjord with sea cage trout farm. Aquatic
Microbial Ecology 2 1 :73-84.
Cline, J. D. 1969. Spectrophotometric determination
of hydrogen sulfide in natural waters. Limnology
and Oceanography 14:454-458.
Cytryn, E., I. Gelfand, Y. Barak, J. van Rijn, and
D. Mintz. 2003. Diversity of microbial commu-
nities correlated to physiochemical parameters in
a digestion basin of a zero-discharge mariculture
system. Environmental Microbiology 555-63.
Dvir, O., J. van Rijn, and A. Neori. 1999. Nitrogen
transformations and factors leading to nitrite ac-
cumulation in a hypertrophic marine fish culture
system. Marine Ecology Progress Series 18 1 :97-
Ellner, S., A. Neori, M. D. Krom, K. Tsai, and M.
R. Easterling. 1996. Simulation model of recir-
culating mariculture ponds with seaweed biofilter:
development and experimental tests of the model.
Aquaculture 143: 167-184.
Froelich, P. N., G. P. Klinkhammer, M. L. Bender,
N. A. Luedtke, G. R. Heath, D. Cullen, P. Dau-
phin, D. Hammond, B. Hartman, and V. May-
nard. 1979. Early oxidation of organic matter in
pelagic sediments of the eastern equatorial Atlan-
tic-suboxic diagenesis. Geochimica et Cosmo-
chimica Acta 43: 1075-1090.
Gowen, R. J. and N. B. Bradbury. 1987. The eco-
logical impact of salmonid farming in coastal wa-
ters: a review. Oceanography and Marine Biology:
An Annual Review 25:463-575.
Greiner, A. D. and M. D. Timmons. 1998. Evalua-
tion of the nitrification rates of microbead and
trickling filters in an intensive recirculating tilapia
production facility. Aquacultural Engineering 18:
ZERO DISCHARGE MARINE SYSTEM
Grguric, G., S. S. Wetmore, and R. W. Fournier.
2000. Biological denitrification in a closed sea-
water system. Chemosphere 40:549-555.
Ingolfsson, G. 0. 200 I. Recirculation systems; from
theory to practice. Pages 113-1 15 in New Spe-
cies, New Technologies. EAS Special Publication
No. 29, European Aquaculture Society, Oostende
Iwama, G. K. 1991. Interactions between aquaculture
and the environment. Critical Reviews in Envi-
ronmental Control 21 : 177-2 16.
Kamstra, A., J. W. van der Heul, and M. Nijhof.
1998. Performances and optimisation of trickling
filters on eel farms. Aquacultural Engineering 17:
Krom, M. D. and A. Neori. 1989. A total nutrient
budget for an experimental intensive fishpond
with circularly moving seawater. Aquaculture 83:
Krom, M. D., S. Ellner, J. van Rijn, and A. Neori.
1995. Nitrogen and phosphorus cycling and trans-
formations in a prototype “non-polluting” inte-
grated mariculture system, Eilat, Israel. Marine
Ecology Progress Series I I8:25-36.
Lekang, 0. I. and H. Kleppe. 2000. Efficiency of
nitrification in trickling filters using different filter
media. Aquacultural Engineering 21:181-199.
Losordo, T. M. and H. Westers. 1994. System car-
rying capacity and flow estimation. Pages 9-60 in
M. B. Timmons and T. M. Losordo. editors. Aqua-
culture water reuse systems: engineering design
and management, developments in aquaculture
and tisheries science, volume 27. Elsevier, Am-
sterdam, the Netherlands.
Lupatsch, I. and G. W. Kissel. 1998. Predicting aqua-
culture waste from gilthead seabream (Spurus uu-
ruru) culture using a nutritional approach. Aquatic
Living Resource I I :265-268.
Menasveta, P., T. Panritdam, P. Sihanonth, S. Pow-
tongsook, B. Chuntapa, and P . Lee. 2001. De-
sign and function of a closed, recirculating sea-
water system with denitrification for the culture of
black tiger shrimp broodstock. Aquacultural En-
Neori, A., M. D. Krom, V. Odintsov, and J. van
Rijn. 2002. Carbon, nitrogen, phosphorus and
budgets in a pilot zero-discharge, super-intensive
mariculture system: biological removal of most
unassimilated nutrients. Proceedings of 4th Inter-
national Conference on Recirculating Aquacul-
ture, 18-2 I July, Roanoke, Virginia, USA.
Nijhof M. and J. Bovendeur. 1990. Fixed film nitri-
fication characteristics in sea-water recirculation
fish culture systems. Aquaculture 87: 133-143.
Park, E. J., J. K. Seo, M. R. Kim, I. H. Jung, J. Y .
Kim, and S. K. Kim. 2001. Salinity acclimation
of immobilized freshwater denitrifiers. Aquacul-
tural Engineering 24: 169-1 80.
Parsons, T. R., Y. Maita, and C. M. Lalli. 1984. A
manual of chemical and biological methods for
seawater analysis. Pergamon Press, Oxford, UK.
Pearson, T. H. and K. D. Black. 2001. The environ-
mental impacts of marine fish cage culture. Pages
1-31 in K. D. Black, editor. Environmental im-
pacts of aquaculture. Sheffield Academic Press
Ltd., Sheffield, UK.
Porter, C. P., M. D. Krom, M. G. Robbins, L.
Bricknell, and A. Davidson. 1987. Ammonia ex-
cretion and total N budget for gilthead seabream
(Spurus aurutu) and its effect on water quality
conditions. Aquaculture 66:287-298.
Poulton, S. W., M. D. Krom, J. van Rijn, and R.
Raiswell. 2002. The use of hydrous iron (111) ox-
ides for the removal of hydrogen sulfide in aque-
ous systems. Water Research 36:825-834.
Scheiner, D. 1976. Determinations of ammonia and
Kjeldahl nitrogen by the indophenol method. Wa-
ter Research 10:31-36.
Shnel, N., Y . Barak, T. Ezer, 2. Dafni, and J. van
Rijn. 2002. Design and performance of a zero-
discharge tilapia recirculating system. Aquacultur-
al Engineering 26: 191-203.
Strickland, J. D. and T. R. Parsons. 1968. A prac-
tical handbook of seawater analysis. Bulletin Fish-
eries Research Board Canada I67:77-80.
Thoman, E. S., E. D. Ingall, D. A. Davis, and C. R.
Arnold. 200 I. A nitrogen budget for a closed, re-
circulating mariculture system. Aquacultural En-
gineering 24: 195-2 10.
Tovar, A., C. Moreno, M. P. Manuel-Vez, and M.
Garcia-Vargas. 2001. Environmental implica-
tions of intensive marine aquaculture in earthen
ponds. Marine Pollution Bulletin 40:981-988.
Turk, P. E., A. L. Lawrence, and P. G. Lee. 1997.
Design and operation of an environmentally iso-
lated marine broodstock culture system using
closed, recirculating water filtration. Advances in
aquacultural engineering. North East Regional
Agricultural Engineering Service, NRAES- 105,
Cornell, New York, USA.
van Rijn, J. 1996. The potential for integrated biolog-
ical treatment systems in recirculating fish cul-
ture-a review. Aquaculture 139: 181-201.
van Rijn, J. and A. Nussinovitch. 1997. An empirical
model for predicting degradation of organic solids
in fish culture systems based on short-term obser-
vations. Aquaculture 154: 173-179.
van Rijn, J. and G. Rivera. 1990. Aerobic and an-
aerobic biofiltration in an aquaculture unit: Nitrite
accumulation as a result of nitrification and deni-
trification. Aquacultural Engineering 9: 1-18.
van Rijn, J. and M. Shilo. 1989. Environmental fac-
tors in fish culture systems. Pages 163-178 in M.
Shilo and S. Sarig, editors. Fish culture in warm
water systems. CRC Press, Boca Raton, Florida.
358 Download full-text
GELFAND ET AL.
van Rijn, J., N. Fonarev, and B. Berkowitz. 1995.
Anaerobic treatment of fishculture effluents: di-
gestion of fish feed and release of volatile fatty
acids. Aquaculture 33:9-20.
Whitson, J., P. Turk, and P. Lee. 1993. Biological
denitrification in a closed recirculating marine cul-
ture system. Pages 458-480 i n J.-K. Wang, editor.
Techniques for modern aquaculture. ASAE, St.
Joseph, Michigan, USA.
Wu, R. S. S. 1995. The environmental impact of ma-
rine fish culture: towards a sustainable future. Ma-
rine Pollution Bulletin 31: 159-166.