Bio-energetic Modeling of Growth and Waste Production of Nile Tilapia
(Oreochromis niloticus L.) in Recirculation Systems
Dr. M.C.J. Verdegem1, Dr. Ir. A.A. van Dam1, M.Sc. A.A. Cabarcas-Nuñez2 and Dr. L.
Oprea3.
1
Fish Culture and Fisheries Group, Department of Animal Sciences, Wageningen University.
P.O.Box 338, 6700 AH Wageningen, The Netherlands.
Mark.verdegem@alg.venv.wau.nl
2
Department of Marine Sciences, University of Puerto Rico, P.O.Box 9013. Mayagüez, PR
00681-5000, USA.
3
Fishing and Aquaculture Department, University of Galati “Dunarea de Jos”, Domneasca
Str. 47, 6200 Galati, Romania.
ABSTRACT
A bio-energetic fish growth simulation (FGS) model was developed for Clarias gariepinus
and subsequently adjusted for the culture of Oreochromis niloticus, Oncorhynchus mykiss
and Colossoma macropomum. The FGS model was extended with a fish waste module
(FWM) to calculate the total waste production due to feeding by tilapias grown in indoor
recirculation systems. Wastes calculated included the amount of uneaten feed, feces and
NH4+ production. The amounts of wastes produced were expressed as g nitrogen (N). The
model was calibrated and validated using 3 independent data sets, together comprising 175
aquarium experiments, monitoring growth in all cases and changes in proximate body
composition of O. niloticus between stocking and harvesting in 51 cases. Fishes were grown
in the individual weight range of 1 – 290 g using 32-54 % protein diets and feeding levels
between 5 and 35 g kg-0.8 d-1. The principal read-out parameters for calibration and
validation of the model were final weight and final body fat level. Because waste production
is the result of the same metabolic processes that lead to protein and fat deposition, it was
assumed that waste production was simulated well when protein and fat deposition are. The
calibrated model was used to review the effect of feeding level and dietary protein level on
N-waste production per kg tilapia produced. Finally, tilapias were grown in 2 different types
of recirculation systems and stagnant water ponds, quantifying N-inputs and the amount of
N-wastes recovered from each system. The latter was defined as the sum of N-waste
discharged (sludge and sludge water drained) and within system accumulation of N-wastes
(organic and inorganic nutrients) during culture. After model calibration, the agreement
between simulated and observed final weight and body fat level for all data sets was
visualized. In recirculation systems different types of N-wastes were estimated well by the
model. Care must be taken when applying the model to pond systems. More insight is
needed on feeding ecology of tilapias in these systems.
INTRODUCTION
It is not possible to grow animals without producing wastes. Only a small part of the nutrient
input is retained in the animals raised (Wit de 1992), and farm management will always be
partially geared towards improving the nutrient retention efficiency. On average, the animals
raised waste 50-85% of the nutrients received. In aquatic systems, these non-retained
nutrients or wastes end up in the same water column where the animals live. The moment
waste is produced within the water column its constitution and concentration change. For
example, some metabolic end products like NH4+ serve directly as nutrients for plants (Syrett
1981, McCarthy 1981), phosphates form insoluble compounds precipitating at the bottom
(Boyd 1995) and wastes like feces or uneaten feeds are mineralized (Mohanti et al. 1994). In
addition, in outdoor systems extra nutrients are added to the water column through
photosynthesis-based processes or N2-fixation (Olah et al. 1994). Consequently, in aquatic
systems it is not possible to measure the exact waste production by farmed animals. Only the
amounts of waste accumulating on-farm and the amounts of effluents discharged during a
production cycle can be measured.
During farm operation, small volatile molecules like CO2, NH3, O2 or N2 are lost from the
systems (Addiscott 1995). In areas with a nutrient surplus, the biomass and nutrients lost in
this way are no longer a waste management problem for the farmer, even though for example
CO2 or NH3 will create environmental problems in farm-dense areas (Kelly et al. 1994).
Contrarily, in nutrient deficient areas, farmers should try to minimize nutrient losses. In both
cases, it is important to estimate the amount of nutrients lost during production.
In this study, N-waste was defined as the N present in uneaten feed, in feces and in ammonia
gill excretion. The objectives were (1) to model accurately waste production by tilapias
grown in recirculation systems based on feed input (2) to analyze the effect of feed
composition and feeding level on growth and N-waste production, and (3) to compare
simulated waste production to the observed waste accumulation and discharge on-farm. The
farming systems used were an indoor recirculation systems, an outdoor recirculation systems
and stagnant water ponds.
MATERIALS AND METHODS
Fish Waste Generator Model
A bio-energetic fish growth simulation model was developed for Clarias gariepinus
(Machiels and Henken, 1986) and subsequently adjusted for the culture of Oreochromis
niloticus and Oncorhynchus mykiss (van Dam and Penning de Vries 1995) and Colossoma
macropomum (van der Meer and van Dam 1998). For a detailed description of parameters
and equations used in the model see van Dam and Penning de Vries (1995). The model was
extended with a fish waste module (FWM) to calculate the waste production as a result of
feeding for O. niloticus grown in an indoor recirculation system. The types of wastes
calculated included uneaten feed, feces and NH4+ production, and were expressed as weight
(grams) of N produced in the system during the entire culture period.
The principal read-out parameters for calibration and validation of the model were protein
and fat deposition. Because waste production is the result of the same metabolic processes
that lead to protein and fat deposition, it was assumed that waste production is simulated well
when protein and fat deposition are.
The original model for O. niloticus, written in Professional Dynamo Plus was translated to
Turbo Pascal. To see whether the two versions of the model work identically, the data set
used by van Dam and Penning de Vries (1995) was used; results were equal.
Digestibility
Commercially formulated diets are composed of quality ingredients with good digestibility.
The fractions of protein, fat or carbohydrate digested were determined for a group of tilapias
raised in recirculation systems and fed at maintenance level (5 g kg-0.8 d-1) (Table 1), and
were different from the original values used by van Dam and Penning de Vries (1995). The
digestibility values obtained for recirculation systems were used in the model.
Table 1. Digestibility of Nutrients in Tilapia Diets. Within Parenthesis, the Proximate
Composition (on a Dry Matter Basis) of the Commercial Diet Used in the
Recirculation System is Given.
Proximate composition feed
Protein (50 %)
Fat (12 %)
Carbohydrate (29 %)
Digestibility used by
van Dam and Penning
de Vries (1995)
0.8
0.5
0.5
Digestibility in
recirculation system
0.9
0.9
0.4
The data sets
Three independent data sets were used for calibration and validation (Table 2, data set 1
through 3). Commercial diets were used, ranging in protein level between 28 and 54%
protein, with feeding levels in the range of 5 – 32 g kg-0.8 d-1. Data set 4 refers to growth and
waste monitoring experiments in ponds (2 ponds), indoor recirculation systems (2) and
outdoor recirculation systems (4). Ponds and outdoor recirculation systems were situated oncampus in Puerto Rico, while the Fish Culture and Fisheries Group in the Netherlands
operated the indoor recirculation systems.
Table 2. Data Sets Used for Validation and Calibration, Giving Minimum and
Maximum Values for Each Set of Feeding/Growth Observations.
Information available
Number of feeding/growth trials
Fish
Initial number of fish
Initial individual weight of fish
(g)
Initial fat %
Final number of fish
Final individual weight of fish (g)
Final fat %
Feed
% dry matter
% protein
% fat
% carbohydrate
Daily amount fed (g)
Number days of feeding
Average water temperature
Data set 1
Calibration
Data set 2
Validation
Data set 3
Validation
34
17
124
Data set 4
Waste
analysis
8
25-40
12-123
10-23
52-180
7-626
1-200
396-800
34-118
11-12
25-40
20-204
8-14
4-11
10-23
60-240
5-10
7-586
5-290
4-11
384-737
99-186
6-10
91
40-50
14
25-35
6-209
25-40
27-28
91-92
38-45
9-13
23-34
0-89
29-68
27-28
91
32-54
7-18
7-43
0-415
5-49
28
89-91
28-47
4-6
29-50
600-2392
30-47
28
Estimating the Amount of Uneaten Feed
The fish growth model developed by van Dam and Pauly (1995) for O. niloticus limited feed
intake by O2 availability, the latter being a function of oxygen pressure in the water column
and gill surface area. The disadvantage of this method is that the oxygen concentration and
temperature in the water column need to be known. Farmers do not record this type of
information. Therefore, maximum feed intake was estimated as a function of body weight.
Groups of tilapias growing in the weight ranges of 10-50, 40-140 and 60-210 g, respectively
were fed different rations in the range of 5 to 35 g kg-0.8 d-1 of diets with a protein level in the
range of 40-50% protein on a dry weight basis. These feeding experiments were part of data
set 1 (Table 1). To estimate the maximum feed intake for different size groups of tilapia a
multiphasic allometric relation was used (Koops and Grossman 1993) (Figure 1):
Y = a1+ b1X –(b1-b2) r ln[1+ e(X-C)/r]
(Equation 1)
-0.8 -1
Where:
X = protein ration (g kg d )
Y = protein growth (g kg-0.8 d-1)
a1, a2 = intercept equation 1 and 2, respectively
b1, b2 = slope equation 1 and 2, respectively
r = smoothness transition between equation 1 and 2 (nearly 0 = abrupt
transition, 2 = smooth transition)
C = central point of transition
Figure 1. Multiphasic Equation, When b2 = 0, the Second Part of the Multiphasic
Equation is a Horizontal Line.
30
Growth (g kg-0.8 d-1)
25
EQUATION 2: Y = a2 + b2X
20
MULTIPHASIC EQUATION:
Y = a1+ b1X –(b1-b2) r ln[1+ e(X-C)/r]
15
EQUATION 1: Y = a1 + b1X
10
5
2
6
C
10
14
18
22
Protein ration (g kg-0.8 d-1)
The parameters a1 and b1 were assumed to be equal for each size group. Parameters were
estimated using the non-linear regression procedure of SAS 6.12. The value of C obtained for
each size group was plotted against the average geometric mean body weight. The resulting
linear relation between body weight and maximum protein intake was incorporated into the
model.
Calibration
The parameters used for calibration were the digestibility of protein, fat and carbohydrate and
the slope and intercept of the linear regression between body weight and maximum protein
intake. Read-out parameters for calibration were final body weight and final body fat content.
Agreement between simulated (Y) and observed (X) values was visualized by plotting
simulated values against observed values. The relative error (RE) for each simulation was
calculated as (Y – X)/((X + Y)/2) (in %) (van der Meer and van Dam 1998). The average
relative error (ARE) is the mean of the RE’s.
Validation
Data set 2 and 3 were used for calibration. Only final body weight was used as read-out
parameter for data set 3 because no data on body fat composition were available.
Estimation of N-waste Production
Using the calibrated model, the amounts of N-waste generated per kg of tilapia produced
were calculated considering two situations. In the first simulation 50-g tilapias were fed a
daily ration of 15-g kg-0.8 d-1 using feeds with protein levels varying between 15 and 50% of
dry matter. In the second simulation, 50-g tilapias were fed a 35% protein diet (% dry matter)
at rations varying between 5 to 40 g kg-0.8 d-1.
Comparison of Waste Production to the Amount of Waste Recovered
The simulated amounts of waste produced by the fish were compared to the amount of waste
accumulated in and discharged from the systems during a production cycle of tilapia (data set
4, Table 2). The following systems were included in the analysis:
indoor recirculation system, 2 growth trials with a 47% protein diet at a feeding level
of 12.5 g kg-0.8 d-1.
stagnant water ponds, 2 growth trials with 28% protein diet at a feeding level of 18.7
kg-0.8 d-1.
outdoor recirculation systems with a 30% protein diet, and a feeding level of 15.7 kg0.8 -1
d .
2 growth trials, with 24 hour solid waste removal.
2 growth trials, with 12 hour solid waste removal.
RESULTS
Estimating the Amount of Uneaten Feed
Figure 2 reviews growth of 3 size classes of tilapia in relation to protein ration. The transition
point above which growth does not increase further is given by the C values (Table 3).
Figure 2. Growth in Relation to Protein Ration for 3 Size Groups of Tilapia.
MULTIPHASIC EQUATION
Y = -0.174+3.259*X-3.259*ln(1+e(X-C))
30
20
15
10
Slope
b = 3.259
GROWTH (g kg
-0.8
-1
d )
25
5
3.3
5.8
7.2
C-values
0
0
-5
2
4
6
8
10
12
14
16
PROTEIN RATION (g kg-0.8 d-1)
10-50 g fish
40-140 g fish
60-210 g fish
18
20
22
24
Table 3. Average Individual Weight of Size Groups with the Corresponding Maximum
Protein Intake Level and Maximum Growth Rate.
Mean body weight
g
22
75
142
Maximum protein
ration (C-value)
g kg-0.8 d-1
7.2
5.8
3.3
Maximum growth rate
g kg-0.8 d-1
23
19
11
The maximum protein ration (Y) was calculated as:
Y = -0.032X + 8.041,
where X = geometric mean body weight (g).
(Equation 2)
Calibration
Adjusted parameter values as a result of calibration are given in Figure 3 and Table 4. Figure
4, gives plots of observed (X-axis) against simulated values (Y-axis) for data set 1. The
uncorrected plots compare observed to simulated values before inclusion of equation 2 into
the model. Before correction, without a limitation to feed intake, simulated weights were
higher than observed weights, especially for the larger fishes. Simulated fat deposition was
higher than observed fat deposition for the higher feeding levels within each size group. The
ARE’s for growth and fat deposition were 19.7 and 17.3, respectively. After calibration the
ARE’s became 0.0 and –3.2, respectively.
Figure 3. Regression of Body Weight Against Maximum Dietary Protein Ration.
7.0
maximum protein ration (g kg
-0.8
-1
d )
8.0
After calibration:
Y = -0.0273X + 8.0409
6.0
5.0
Calculated regression
Before calibration:
Y = -0.0324X + 8.0409
R2 = 0.9905
4.0
3.0
2.0
15.00
45.00
75.00
105.00
mean body weight (gram)
135.00
Table 4. Model Parameters Before and After Calibration.
Parameter
Protein digestibility
Fat digestibility
Carbohydrate digestibility
Slope regression maximum protein
intake against body weight
Before calibration
0.9
0.9
0.4
-0.0324
After calibration
0.9
0.9
0.6
-0.0273
Figure 4. Agreement Between Observed and Simulated Values for Fish Weight and Fat
Content, Using Uncorrected (Before Calibration) and Corrected Parameters
(After Calibration, Table 4), Based on Data Set 1.
Calibration (Data set 1)
250
15
uncorrected
uncorrected
14
200
13
12
150
50
ARE = 19.7
-8.2 <RE < 58.4
0
corrected
200
Simulated fat content (% fw)
Simulated fresh weight (g)
11
100
10
9
ARE = 17.3
-6.2 <RE < 46.0
8
7
corrected
14
13
12
150
11
100
10
9
50
ARE = 0.0
-21.0 <RE < 28.1
0
ARE = -3.2
-22.7 <RE < 13.3
8
7
0
50
100
150
200
250
7
9
10
11
12
13
Experimental fat content (% fw)
Experimental fresh weight (g)
Weight range:
8
10-50 g
40-140 g
60-210 g
14
15
Validation
Validation results are given in Figure 5. For data set 2, for two growth experiments
simulated weights were much higher than observed weights, while the opposite was true for
fat deposition. For the other experiments in data set 2, simulated values for growth and fat
deposition were slightly lower than observed values (ARE –5.9 and –3.9 for weight and fat
deposition, respectively). For data set 3, growth was simulated well for fish up to 200 g. For
fish reaching a final weight of 200-295 g, observed weight were lower than simulated
weights, causing the ARE to be–9.8).
Figure 5. Agreement Between Observed and Simulated Values for Fish Weight and Fat
Content Based on Data Set 2 and 3.
Simulated fat content (% fw)
Simulated fresh weight (g)
Validation (Data set 2)
300
250
200
150
100
50
ARE = -5.9
-16.4 <RE < 16.7
11
10
9
8
7
6
5
4
ARE = -3.9
-45.9 <RE < 16.6
3
2
0
0
50
100
150
200
250
300
Validation (Data set 3)
300
250
200
150
100
50
ARE = -9.8
-61.3 <RE < 17.8
0
0
50
100
150
200
Experimental fresh weight (g)
2
3
4
5
6
7
Experimental fat content (% fw)
Experimental fresh weight (g)
Simulated fresh weight (g)
12
250
300
8
9
10
11
12
Estimation of N-waste Production
The simulated waste production for fish fed feeds with increasing protein levels at 15-g kg-0.8
d-1 is given in Figure 6. The fish reached a maximum weight of 148 g when fed a 45%
protein diet. At still higher dietary protein levels, growth decreased. Ammonia excretion was
the largest fraction of N-waste, and increased with dietary protein level. The same was true
for N in the feces, but given the high digestibility of dietary protein, N-feces always
remained below 6 g N per kg fish produced. Starting from dietary protein levels of 35%
going up, part of the feed remained uneaten. With the 50% protein diet almost 32% of the
total N-waste produced came from uneaten feed.
Figure 6. Final Weight and N-waste Production of 50-g Fishes Fed During 45 Days at
15-g kg-0.8 d-1, Diets with Varying Dietary Protein Levels. Carbohydrates and
Protein Made up 75% of the Diet on a Wet Weight Basis.
60
160
N in faeces
N ammonia
N in waste feed
Fish weight
50
140
40
100
30
80
60
Fish final weight (g)
N waste / kg fish produced
120
20
40
10
20
0
0
15
20
25
30
35
40
45
50
Feed protein content (% dm)
The simulated waste production for fish fed a 35% protein diet at increasing feeding levels is
given in Figure 7. The fish reached a maximum weight of 154 g at a feeding level of 20 g kg 0.8 -1
d . The lowest N-waste production of 30 g N per kg fish produced was obtained with a
feeding level of 10 g kg-0.8 d-1. Below or above that level, N-waste production increased.
Increasing the feeding level from 10 to 15 g kg-0.8 d-1 resulted in a 64 % increase in weight
gain and a 7 % increase in N-waste production. Increasing the feeding level from 15 to 20 g
kg-0.8 d-1 resulted in a 14% in weight gain and a 44% increase in N-waste production. When
more food is given, it remains uneaten, resulting in no extra growth and more N-waste.
Figure 7. Final Weight and N-waste Production of 50-g Fishes Fed During 45 Days a
35% Protein Diet at Feeding Level Varying Between 5 and 40 g kg-0.8 d-1.
120
180
N in faeces
N ammonia
N in waste feed
Fish weight
160
140
80
120
100
60
80
40
60
Fish final weight (g)
N waste / kg fish produced
100
40
20
20
0
0
5
10
15
20
25
30
35
40
Feeding level (g kg-0.8 d-1)
Comparison of Waste Production to the Amount of Waste Recovered
Figure 8 reviews the agreement between simulated and observed final weight of the different
growth trials. The ARE was 4.9%. Except for the pond data, the simulated values are slightly
higher than the observed values.
Figure 9 compares the simulated waste production in each system to the amount of waste that
was discharged from the system or accumulated in the system. Assuming tilapias realize their
growth only based on the administrated feed, the simulated N-waste production is given. In
ponds, in total 60% of the simulated N-waste produced could be measured as accumulated in
the water column and soil, and 40% could not be accounted for. In the indoor recirculation
system N-lost was 51% of the total amount of simulated N-waste produced by the fish. Of
the total amount of N-waste produced only 49 % could be traced in the form of Naccumulated (13%) and N-discharged in the effluent (33%). Similarly, in the outdoor
recirculation system 33 and 53% of the estimated waste production by the fish could be
traced back with the 12 and 24-hour discharge period, respectively. In all cases, the biggest
source of N-waste was N-ammonia.
Figure 8. Agreement Between Observed and Simulated Values for Fish Based on Data
Set 4.
Simulated fresh weight (g)
200
175
150
125
100
ARE = 4.9
-2.3 < RE < 10.8
75
75
100
125
150
175
200
Experimental fresh weight (g)
POND
RECIRC. SYSTEM (outdoor 12-h)
RECIRC. SYSTEM (outdoor24-h)
RECIRC. SYSTEM (indoor cont. disch.)
Figure 9. Comparison Between the Simulated N-waste Production (sim) and the
Amount of N-waste that Could be Traced Back in the Systems (exp). The
Percentage of Sim that Could not be Traced Back is Given in the N-lost Bar.
POND
RECIRCULATION
SYSTEM (outdoor)
12-h discharge
RECIRCULATION
RECIRCULATION
SYSTEM (outdoor)
SYSTEM (indoor)
24-h discharge
continuous discharge
51%
2000
1500
47%
69%
40%
total N-waste for system (g)
2500
1000
500
0
exp
sim
N-ammonium
exp
N-feces
sim
N-wasted feed
exp
sim
N-accumulated
exp
N-discharged
sim
N-lost
DISCUSSION
Estimating the Amount of Uneaten Feed
Within one size group, the multiphasic equation gives a clear prediction of the transition
point where a higher protein ration level does not yield more growth. Unfortunately, this was
done for 3 size groups of tilapias. Therefore, the linear relation relating maximum daily
protein ration to individual body weight was determined with 3 data points only (Figure 3).
For larger fish, the protein rations decrease too fast. Using equation 2, the protein ration
becomes 0 for fishes above 250 g. Through calibration, the slope of equation 2 was raised,
increasing the fish size where protein ration becomes 0 to 295 gram. With more data points,
especially in the larger weight ranges, an exponential function could be fitted with and
asymptotic value for protein ration in large fish.
Using only data for small fishes, equation 2 clearly improved the agreement between
observed and simulated values, indicating that uneaten feed is a problem. Even when fishes
are observed ingesting the food, still a part can be wasted with the water current over the gills
during mastication. Van der Meer et al. (1997) found that Colossoma macropomum fed ad
libitum spilled nearly 30% of the feed apparently consumed. When the feeding level was
reduced to 60% of the ad libitum ration the amount of wasted feed became negligible.
Many factors besides protein feeding level and fish size influence feed intake. Van Dam and
Pauly (1995) based feed intake on O2 availability. This approach is strongly related to the
ability of the fish to take up O2 (gill surface, oxygen pressure in water column). Considering
food availability and food composition this approach allows estimating the energy available
for metabolism and consequently growth. A condition for the use of this approach is the
availability of data of O2 concentration over 24-hour periods. This information is often not
available for outdoor systems. Only with indoor systems that aerate constantly, O2 levels
remain constant during the day. Models exist to predict algae production in ponds (JiménezMontealegre et al. 1995) which can be adapted to predict fluctuations in O2 levels. If these
models could be linked, the lack of data of 24-hour O2 profiles could be avoided.
Estimating of N-waste Production
Nijhof (1994) calculated the effect of feed composition, feed conversion and uneaten feed on
waste discharge in an eel recirculation system, using a time step of 1 day. Although
interesting, this information cannot be used for design purposes. Figure 6 and 7 give insight
on the effect of feed composition and feeding level on waste production in relation to
production. When the fish density and volume of a system are known, the total waste load
can be calculated. The model works with a time step of integration of 3 hours, and changes in
nutrient loading within a 24-hour period are also calculated. Such information is useful to
design water flow, biofilter size and shape, and the sedimentation unit. In addition, the model
allows to select feeding strategies that lead to a minimal waste production. The approach can
also be extended to other types of waste production, like the amounts of organic matter
produced, CO2 production, COD load, etc., and be used as a tool for the development of low
pollution diets (New 1996)
Comparison of Waste Production to the Amount of wWaste Recovered
The highest fraction of the simulated waste production was recovered in ponds, mainly in the
sediment. Ammonia present in the water column is taken up by algae, of which a large
fraction accumulates at the bottom (Brune and Drapcho 1991). This might explain in part
why in ponds a higher percentage of the N-waste produced could be traced back than in
indoor recirculation systems where no N-waste is stored in life or dead algae biomass. Algae
were also present in outdoor recirculation systems, but given the small water volume and
surface area the contribution of algae to N-uptake or storage is negligible.
In all field experiments, including the experiment in Wageningen, feed formulation was
based on ingredients commonly included in commercial tilapia feeds in Puerto Rico. These
pellets had the disadvantage that they felt apart within seconds upon contact with water,
resulting in a higher than normal level of feed spillage. The calibration and validation of the
model was done based on feeding trials in which water-stable sinking pellets were used. This
might explain why for the recirculation systems the simulated weights were slightly higher
than the observed weights. The situation in ponds was different, as besides access to dietary
N, tilapias can also feed on natural feeds. Research by Schroeder (1983) and Anderson
(1987) indicated that in ponds receiving commercial pellets, natural feeds contributed 5080% to the fish or shrimp production. More research is needed, however, if correct, this
model should not be applied to ponds, where fish have easy access to detritus, algae, biofilms
(including periphyton) and benthic organisms as alternative food sources. The fact that
growth was simulated well should then be regarded as a matter of luck.
In recirculation systems, an N recovery level below 50% has been recorded for rainbow trout
(van Weerd et al. 1995) when considering total N-input. In this study it was assumed that no
N got trapped in the biofilms present in the recirculation systems, as these were already fully
developed at the start of each study. It should be checked whether this assumption was
correct. Which processes cause the high levels of N-lost needs further investigation. Some of
the N-loss might be explained through volatilization of gaseous N (Avnimelech et al. 1992).
CONCLUSIONS
The model simulated well growth and fat deposition in tilapia when applying feeds of
varying composition at different daily feeding rations. In situations where fish has limited
access to natural foods, the model can be used to predict N-waste production, including the
amount of N in uneaten feed. Also other types of metabolic wastes can be quantified with the
model. However, the model does not work well for fishes weighing above 250 gram. Data on
feed utilization by tilapias larger than 250 gram are needed to adjust the relationship between
maximum protein intake and body weight. As long as the feeding ecology in ponds is not
well understood, use of the model for pond situations might lead to erroneous conclusions.
REFERENCES
Addiscott T.M., 1995. Entropy and sustainability. European Journal of Soil Science, 46: 161168.
Anderson K.R., P.L. Parker and A. Lawrence, 1987. A 13C/12C tracer study on the utilization
of presented feed by a commercially important shrimp Penaeus vannamei in a pond
growout system. Journal of the World Aquaculture Society 18: 148-155.
Avnimelech Y., N. Mozes and B. Weber, 1992. Effects of aeration and mixing on nitrogen
and organic matter transformations in simulated fish ponds. Aquacultural
Engineering 11: 157-169.
Boyd C.E., 1995. Bottom soils, sediment, and pond aquaculture. Chapman & Hall, New
York, USA, pp. 348.
Brune D.E. and C.M. Drapcho, 1991. Fed pond aquaculture. Pages 15-33 in Aquaculture
Systems Engineering. Proceedings of the World Aquaculture Society and the
American Society of Agricultural Engineers (ASAE): WAS 22nd Annual Meeting,
16-20 June 1991, San Juan, Puerto Rico. Published by the American Society of
Agricultural Engineers, ASAE publication 02-91, 138 pp.
Dam van A.A. and D. Pauly, 1995. Simulation of the effects of oxygen on food consumption
and growth of Nile tilapia, Oreochromis niloticus (L.). Aquaculture Research 26 (4):
427-440.
Dam van A. A. and F.W.T. Penning de Vries, 1995. Parameterization and calibration of a
model to simulate effects of feeding level and feed composition on growth of
Oreochromis niloticus (L.) and Oncorhynchus mykiss (Walbaum). Aquaculture
Research 26 (6): 415-426.
Jiménez-Montealegre R. , J. Verreth, K. Steenbergen, J. Moed and M. Machiels, 1995. A
dynamic simulation model for the blooming of Oscillatoria agardhii in a monomictic
lake. Ecological Modelling 78 (1995) 17-24
Kelly, L.A., A. Bergheim, M.M. Hennessy, 1994. Predicting output of ammonium from fish
farms. Water Research 28(6): 1403-1405.
Koops, W.J. and M. Grossman, 1993. Multiphasic allometry. Growth, Development & Aging,
57: 183-192.
Machiels, M.A.M. and A.M. Henken, 1986. A dynamic simulation model for growth of
African catfish, Clarias gariepinus (Burchell 1822). I. Effect of feeding level on
growth and energy metabolism. Aquaculture 56: 29-52.
McCarthy, J.J., 1981. The kinetics of nutrient utilization. Pages 211-233 in T.J. Platt (ed.),
Physiological basis of phytoplankton ecology. Canadian Fisheries Research Board
Bulletin 210, Ottawa, Canada
Meer van der, M.B., R. Faber, J.E. Zamora and M.C.J. Verdegem, 1997. Effect of feeding
level on feed losses and feed utilization of soya and fish meal diets in Colossoma
macropomum (Cuvier). Aquaculture Research 28, 391-403.
Meer, van der M.B., and A.A. van Dam, 1998. Modeling growth of Colossoma macropomum
(Cuvier): comparison of an empirical and explanatory model. Aquaculture Research
29: 313-332.
Mohanti A. N., D. K. Chatterjee, P. K. Saha and K.C. Pani., 1994. Effect of varying C/N
ratios on the mineralization of organic nitrogen in fish pond soils. Journal of
Aquaculture in the Tropics 9: 9-14
New M.B., 1996. Responsible use of aquaculture feeds. Aquaculture Asia 1: (1) pp. 3-4,612,14-15.
Nijhof M., 1994. Theoretical effects of feed composition, feed conversion and feed spillage
on waste discharge in fish culture. Journal of Applied Ichthyology 10:274-283
Olah J., Pekar F. and P. Szabo, 1994. Nitrogen cycling and retention in fish-cum-livestock
ponds. Journal of Applied Ichthyology 10:341-348.
Schroeder, G.L., 1983. The role of natural foods in tilapia growth: a study based on stable
isotope analysis. In L. Fishelson and Z. Yaron (Eds.), pages 313-322, International
Symposium on Tilapia in Aquaculture, Nazareth, Israel 8-13 May 1983, Israel, pp.
624.
Wit de, C.T., 1992. Resource use efficiency in agriculture. Agricultural Systems 40: 125-151.
Syrett, P.J. 1981. Nitrogen metabolism of microalgae. Pages 182-210 in T.J. Platt (ed.),
Physiological basis of phytoplankton ecology. Canadian Fisheries Research Board
Bulletin 210, Ottawa, Canada.
Weerd van, J.H., A.M. Verastegui and P.A.T. Tijssen, 1995. Nitrogen excretion and
determination of nitrogen and energy budgets in rainbow trout (Oncorhynchus mykiss
R.) under different feeding regimes. Journal of Applied Ichthyology 11:322-328.