Environmental Pollution, Vol. 95, No. 3, pp.

379-387, 1997

PII: S0269-7491(96)00117-0

1997 Elsevier Science Ltd

All rights reserved. Printed in Great Britain
0269-7491/97 $17.00+0.00

ELSEVIER

TROUT F A R M EFFLUENTS: CHARACTERIZATION A N D

IMPACT ON THE RECEIVING STREAMS

R u i B o a v e n t u r a , a* A n a M . P e d r o , b J o ~ o C o i m b r a b & E d u a r d o L e n c a s t r e c
aDepartamento de Engenharia Qulrnica, Faculdade de Engenharia, 4099, Porto Codex, Portugal
blnstituto de Ci~ncias Biomddicas Abel Salazar, 4000, Porto, Portugal
cCentro Aquicola do Rio Ave, 4480 Vila do Conde, Portugal

(Received 9 January 1996; accepted 17 September 1996)

Abstract
Effluents from three rainbow trout (Oncorhyncus
mykiss) farms located in Northern Portugal were characterized and their impact on the receiving streams was
evaluated. Mean fish productions in the studied fish farms
were 15, 55 and 500 t of trout per year, respectively. The
feeding water was abstracted from Fornelo, Inha and
Coura Rivers, at flow rates ranging from 1.2 (15 t year -1
fish farm) to 4.8 litre s -1 per ton annual fish production
(500 t year -1 fish farm).
As the water flows through the farms, net variations in
the chemical characteristics were observed." a mean
reduction in the dissolved oxygen (DO) concentration
between 0.7 and 2.4 mg litre-l; mean increases between
1.9 and 3.2 mg CaC03 litre -I for total alkalinity,
between 0.9 and 14 mg litre -1 for BODs, between 0.27
and 1.46 mg litre -1 for ammonia nitrogen (NH4-N),
between 0.060 and 0.579 mg litre -1 for soluble phosphorus (PO4-P) and less than 16 mg litre -~ for suspended solids; variations in the pH value and nitrate nitrogen
concentration were not statistically significant (p
< 0.05). At the 500 t year-l fish farm it was alsopossible
to detect net increases of total hardness (3.2 mg
CaCOs litre-1), electric conductivity (19 mS cm -1) and
permanganate value (3.6 mg02 litre-1). At the other
farms net variations in these parameters were not
significant.
Net mass flow variations reported to the annual fish
production are presented. The DO mass flow decreased,
on average, between 255 and 549 g t -1 offish per day.
The mean daily BODs increase ranged from 353 to
1510 g t -1 offish. The corresponding ranges for the other
parameters were 105-157 g t -1 for NH4-N, 24-62 g t -1
for PO4-P, 348-1035 g CaCOs t -1 for total alkalinity
and 224 x 106-506 x 106 t-1 for mesophilic bacteria. Daily
net variations of suspended solids, total hardness, electric
conductivity and permanganate value were below
1753 g t -t, 342 g CaCOz t -1, 2081 mS c m t - 1 and
392 gO2 t -1, respectively.
Longitudinal concentration profiles for the most relevant parameters show the impact of the effluent discharges on the physico-chemical and bacteriological river

water quality downstream from the trout farms. Analyzing the situations from a purely chemical point of view,
the polluted stretches were 3, 5 and 12 km long downstream from the effluent discharges, respectively. The
microbiological contamination extended over longer distances. 1997 Elsevier Science Ltd
Keywords: Fish farm, trout farm, effluent, water pollution, environmental impact, pollution load.

INTRODUCTION
The quantification of environmental impacts of intensive aquaculture provide information for decisionmakers in licensing new fish farms. The expansion of
aquaculture in a fiver basin must be limited by the
necessity of minimizing environmental degradation.
Intensive aquaculture affects water quality in many
ways, including (Gowen et al., 1990) hypernutrification,
benthic enrichment, organic matter increase and bacterial changes. The effects are sometimes quite severe
and include (Warrer-Hansen, 1982) an increase in BODs,
a decrease in DO concentration, a N and P enrichment
leading to an acceleration of algae and plants growth,
some changes of the bottom fauna and of the amount
and composition of sediments, the presence of chemicals
and drugs used against parasites and pathogenic bacteria and, eventually, the production of foul smell.
A monitoring programme was developed in 1989 in
Denmark to quantify and reduce N and P loadings to
the aquatic environment (Kronvang et al., 1993; Iversen,
1995). Discharges from fish farms have been identified
as a major pollution source to inland waters draining to
the North Sea. Nitrogen has been considered as the limiting nutrient except in lakes, where P limits primary production. Land use affects the P concentration of the surface
waters which increases on agricultural areas. Nevertheless, discharges from sewage treatment plants and fish
farm effluents produce much higher concentrations.
Total N and P loadings measured over a 12-month
period from a Northern Ireland trout farm were reported as 124.2 kg and 25.6 kg, respectively (Foy & Rosell,
1991a). Day-to-day variations were observed and a

*To whom correspondence should be addressed.

379

R. Boaventura et al.

380

relationship between temperature and N and P loss

rates was investigated. No evidence of such relationship
was detected for P and only a weak one was apparent
for N.
The distribution of the total N and P contents
between soluble and particulate fractions was also
studied by the same authors (Foy & Rosell, 1991b).
Soluble P has been then separated in the reactive and
unreactive species and soluble N in the ammoniacal,
nitrite+nitrate and organic fractions. Particulate P
represents about 30% of the total P, whereas particulate
N only represents about 8.9%. Loss rates were
determined for all species and it was concluded that
temperature had no effect on particulate P and N loss
rates.
The impact of an intensive trout culture on a river
depends on the size of the farm, the fanning practices,
the nature and volume of the wastes produced, the
dilution and self-purification capacities of the water
body and the physico-chemical and bacteriological
characteristics of the river water upstream from the
discharge.
Impacts of a particular farm may be classified as
internal, local or regional (Silvert, 1992). Internal
impacts are those of a fish farm on itself and the
immediate environment. Local impacts extend generally
to a kilometer downstream from the discharge and can
affect nearby farms. The effects on the entire water
body, with space scales of many kilometers, are considered as regional impacts.
This study characterizes intensive rainbow trout
(Oncorhyncus mykiss) farming effluents, both from the
point of view of their physico-chemical and bacteriological composition and flow rate, and investigates the
magnitude and extent of river water quality changes
associated to the metabolites, faeces and non-ingested

food derived from particular trout farms. Three farms

of different capacity and located in Northern Portugal
were selected for that purpose (Fig. 1). Different small
rivers supply the feeding water to the farms and receive
their effluents.

MATERIALS AND METHODS

Description of farms
The smallest farm (A) is located at Bustelo, Amarante,
on the right bank of the Fornelo River, a small tributary
of the Ovelha River. Part of the river water is diverted
through farm, a situation which is common to the three
study farms. Fornelo River is a mountain stream running in a forested narrow valley with an average slope of
about 2.5% at the study reach. There are no domestic,
agricultural or industrial pollution sources other than the
trout farm discharge. Fornelo River has a dry period
flow of 0.1-0.2 m 3 s-l and farm A has a production
yield of about 15 t year -1 and a dry feed consumption
between 50 and 100 kg day -1. The feed protein content
was between 38% (in Winter) and 42% (in Summer).
For water quality monitoring purposes four sampling
stations were considered along the river: station A1,
immediately upstream from the trout farm discharge,
and stations A2, A3 and A4, 200, 2200 and 4000 m
below the farm, respectively.
The medium capacity farm (B) is located at Rebordelo, with an annual trout production of 55 tons and a
feed consumption of 400-500 kg day -1, with a protein
content as in farm A. The Inha River, with a dry period
flow of 0.2-0.3 m 3 s-l, constitutes the water source and
the effluent receiving stream. Like the Fornelo River,
this tributary of the Douro River also crosses a forested
area and runs through a narrow, steep valley (about 2%

VIANA
CAST[

USl

VER BASIN

kin

Fig. 1. Location of the study trout farms (A, B and C) in Northern Portugal.

Trout farm effluents

slope at the study reach). Water flows with high velocity
and there are no anthropogenic pollution sources
downstream and upstream from the trout farm. Water
quality was determined at five sampling stations along
the river: station B1, above the farm effluent discharge,
and stations B2, B3, B4 and B5, 200, 700, 3000 and
4000 m downstream from the discharge, respectively.
The largest farm (C) is on the left bank of the Coura
River, at Paredes de Coura, and produces about 500
tons of rainbow trout annually with a maximum feed
consumption between 3500 and 4000 kg day -l, also
with the above mentioned protein content, since all
farms were supplied by the same company. The Coura
River has a dry period flow of 1 m 3 s-l, about five fold
higher than those of the Fornelo and Inha Rivers. Most
of the drainage area downstream from the trout farm is
agricultural land and the river flows through a wide,
slightly sloping (about 1.3% at the study reach) valley.
There are no industrial effluents or domestic sewage
directly discharged into the stream but feedlot wastes
downstream the farm may contaminate the water. Six
sampling stations were located along the river: station
C1, just upstream from the farm effluent discharge, and
stations C2, C3, C4, C5 and C6, 250, 1000, 6250, 8000
and 12 000 m below the discharge, respectively.
The physical, chemical and microbiological characterizations of the feeding water and trout farm effluents
were carried out between April 1988 and February
1989, through seven campaigns for farms A and B and
eight campaigns for Farm C. Generally, samples were
taken between 10 and 12 a.m. The impact on the water
quality of the receiving streams was evaluated between
August 1988 and February 1989, after an initial effluent
characterization period to estimate their relative
importance. During this period, the flow rates ranged
from 0.1 to 0.4 m 3 s -1 in the Fornelo River, 0.2 to
0.4 m 3 s -l in the Inha River and 1.0 to 2.0 m 3 s-l in the
Coura River. Minimum flow rates occurred between
August to October.
River flow rates
River flow rates were calculated from the cross-sectional
area and longitudinal velocity data at one or two sampiing sites in each campaign. The cross-sectional area
was first determined by depth measurements and then
divided into vertical sections where river vel,ocity was
measured using a current meter. Total flow was computed summing the flow increments for all the vertical
sections. Along the river stretches where the aquaculture
impacts were evaluated there are no significant tributaries or water withdrawals, so the calculated flow rates
may be assumed as constant at all the sampling sites.
Analytical methods
Measured parameters in feed waters and effluents
included temperature, total suspended solids, turbidity,
pH, total alkalinity, total hardness, electric conductivity, DO, BOD5, permanganate value, ammonia (as
NH4-N), nitrates (as NO3-N), phosphates (as POa-P)
and mesophilic aerobic bacteria. Temperature, conduc-

381

tivity, pH and DO were measured in situ using a Horiba

U-7 water quality checker. The other analyses were
performed according to Standard Methods (1985),
taking two replicates per sample. The water and the
effluents were analyzed during the same period and the
means and 95% confidence limits of the results were
calculated.
The mean variations in the chemical and bacteriological river water composition induced by fish farms
were evaluated by analyzing water samples taken
upstream (reference levels) and downstream from the
effluent discharges. The selected sampling stations in
each river and the corresponding distances were those
reported before. Characterization included determination of electric conductivity, turbidity, BODs, DO,
NH4-N, soluble P and mesophilic bacteria. The adopted analytical methods were the same as for the inlet
water and effluents.

RESULTS AND DISCUSSION

Feeding water and effluents
The mean water flow rates through the farms and
the associated 95% confidence limits, during the
period of the field observations, are 7222 litre s-1,
25031 litre s-1 and 62433 litre s-1 at farms A, B
and C, respectively. Taking into account the annual
productions, the specific mean water flow rate is minimum at farm C (1.2 litre s-l per ton annual fish production) and maximum at farm A (4.8 litre s-1 per ton
annual fish production). The corresponding value for
farm B is 4.5 litre s-1 per ton annual production. These
values are small when compared with about
5 litre s-it -1 in Denmark, 10 litre s-it -1 in Finland and
35 litre s-it -1 in Italy (Alabaster, 1982). Also in a
Northern Ireland trout farm (Foy & Rosell, 1991a) the
average water consumption over a 12-month period was
6.8 litre s-1 per ton annual production. For fish farms
supplied with river water it has been proposed (Alabaster, 1982) that a direct relationship exists between
the maximum annual production and the dry weather
flow. Whereas for UK and Denmark average productions are 40 and 200 ton per m 3 s-I dry weather flow,
respectively, in farm C the production is about 500 ton
per m 3 s-1 dry weather flow (Coura River).
The physico-chemical and bacteriological composition of the river water, determined at each farm inlet,
is presented in Table 1.
Effluent flow rates at each farm outlet are assumed to
be the same as those of the feeding water. Table 2 shows
the means and the 95% confidence limits of the analytical results. The relationship between the inlet and
outlet data sets for the different parameters was investigated. Then the correlation coefficient was calculated to
determine whether the two data sets from all samples
are related, that is, whether larger values of one set are
associated with large or small values of the other.
A paired two-sample comparison-of-means t-test was
used (Table 3) for comparing the means of the input and

382

R.

et al.

Boaventura

output data sets. At the 95% significance level, the null

hypothesis (equal means) is accepted for the following
parameters: pH, total hardness (farm A), turbidity and
total suspended solids (farm A), temperature and electric conductivity (farms A and B), permanganate value
(farms A and B), nitrates and mesophilic bacteria. In
farm C the mean values for all parameters but p H and
nitrates are statistically different from those of the feeding water. Table 4 shows the mean calculated variations
between the inlet and outlet of each farm (concentrations and mass flows) for those parameters presenting
statistically significant concentration changes. Mass
flow variations in constituents of water, reported to the
annual production of fish in each farm, are presented in
Table 5.
The mesophilic bacteria density in the effluents fluctuates greatly throughout the year, which is indicated by

large confidence limits when compared with the means.

The highest concentrations are observed at farm C, and
farms A and B present smaller values of the same order
of magnitude (Table 2).
Except for total hardness and electric conductivity
which are much higher at farm B, the composition of
the feeding water is similar at the three fish farms.
Although some seasonal variations were detected, the
95% confidence intervals usually keep small values for
most parameters (Table 1).
Taking into account the data from the three fish
farms, no significant correlations were systematically
observed between the inlet and outlet data sets. In
spite of some high correlation coefficients sporadically obtained in a particular fish farm, data sets are
only fairly correlated (r<0.8) or uncorrelated when
considering all fish farms. The physico-chemical and

Table 1. River water composition (mean and 95% confidence limits) at each farm inlet

Parameter

Farm A

Temperature, C
pH b
Total alkalinity, mg CaCO3 litre -l
Total hardness, mg CaCO3 litre-1
Turbidity, NTU
Total suspended solids, mg litre-i
Conductivity (20C), mS cm -1
Dissolved oxygen, mg litre -l
BODs, mg litre -I
Permanganate value, mg 02 litre -I
NH4-N, mg litreNOa-N, mg litreNO2-N, mg litre-I
PO4-P, mg litreMesophilic bacteria, colonies ml -I

Farm B

Farm C

na

Mean confidence limits

na

Mean confidence limits

na

Mean confidence limits

7
7
7
7
7
6
6
6
7
6
7
5
7
6
6

13.22.5
6.50.5
5.7~1.2
8.0~1.3
1.30.1
1.3~1.4
4313
10.80.8
1.1~0.3
0.90.2
0.040.02
1.10.3
<0.2
0.0120.~6
175176

7
7
6
5
7
5
6
6
6
6
6
5
7
5
6

12.12.2
6.10.5
5.11.0
18.32.1
1.70.5
0.60.2
7615
11.11.1
1.30.2
0.70.1
0.060.04
2.20.4
<0.2
0.0050.000
7918

8
8
7
7
8
8
7
8
7
6
8
7
8
8
4

12.42.5
6.10.3
4.41.0
8.31.6
2.40.9
1.50.9
392
10.80.6
1.60.4
0.90.1
0.070.03
1.00.3
<0.2
0.0120.013
19179

Number of observations (it was not possible to analyze some parameters in all samples).
b Mean of [H+].
Table 2. Effluent composition (mean and 95% confidence limits) at each farm outlet

Parameter

Temperature, C
pH b
Total alkalinity, mg CaCO3 litreTotal hardness, mg CaCO3 litre -I
Turbidity, NTU
Suspended solids, mg litre-i
Conductivity (20C), mS crn -1
Dissolved oxygen, mg litre-I
BODs, mg litre -I
Permanganate value, mg 02 litre-I
NH4-N, mg litre -1
NO3-N, mg litre-I
NO2-N, mg litre -~
PO4-P, mg litre -1
Mesophilic bacteria, colonies ml -~

Farm A

Farm B

Farm C

na

Mean conf.
limits

na

Mean 4- conf.
limits

n~

Mean conf.
limits

7
7
7
7
7
6
6
6
7
6
7
5
7
6
6

13.52.8
6.30.3
7.61.6
8.11.7
1.70.5
1.80.3
5011
9.50.3
2.00.6
1.30.6
0.420.17
1.00.3
<0.2
0.0980.034
718543

7
7
6
5
7
5
6
6
6
6
6
5
7
5
6

12.62.5
6.20.3
7.71.3
20.02.3
3.20.5
2.41.6
836
10.50.9
3.11.2
1.20.6
0.320.12
2.10.4
<0.2
0.0650.030
842750

8
8
7
7
8
8
7
8
7
6
8
7
8
8
4

13.12.9
6.10.2
7.61.6
11.51.0
9.61.8
17.84.9
583
8.40.5
15.68.4
4.53.2
1.520.25
1.00.2
<0.2
0.5910.111
1657623242

Number of observations (it was not possible to analyze some parameters in all samples).
b Mean of [H+].

Trout farm

bacteriological c o m p o s i t i o n of the effluents is m a i n l y

d e t e r m i n e d by f a r m i n g practices. T h e net p r o d u c t i o n o f
N O 3 - N close to zero that was observed in the three
farms has been already reported in other studies
(Bergheim e t a l . , 1984; Solb6, 1982). The D O conc e n t r a t i o n decreased from the inlet to the outlet, b u t

383

eff?uents

the r e m a i n i n g p a r a m e t e r s present higher c o n c e n t r a t i o n s

at the farm outlets. A d d i t i o n a l l y , the recorded differences at farm C were always higher t h a n those at farms
A a n d B. C o m p a r i n g the p e r f o r m a n c e at these last two
farms a discrepancy seems to occur c o n c e r n i n g D O ,
N H 4 - N a n d P O a - P c o n c e n t r a t i o n s . Higher v a r i a t i o n s

Table 3. Comparison of the inlet and outlet parameter means using a paired two-sample t-test (5% significance level)
Parameter

Farm A

Temperature, C
pH
Total alkalinity, mg CaCO3 litre -1
Total hardness, mg CaCO3 litre- I
Turbidity, NTU
Total suspended solids, mg litre-~
Conductivity (20C), mS cm -1
Dissolved oxygen, mg litre 1
BODs, mg litre -1
Permanganate value, mg 02 litre -1
NH4-N, mg litre-1
NOa-N, mg litre- 1
PO4-P, mg litre -~
Mesophilic bacteria, colonies ml i

Farm B

Farm C

Ia

tc b

ta

tc b

la

tc b

1.656
0.980
5.857
0.111
1.882
0.696
1.704
5.121
4.054
1.784
5.020
1.633
5.357
1.977

1.943
1.943
1.943
1.943
1.943
2.015
2.015
2.015
1.943
2.015
1.943
2.132
2.015
2.015

1.329
1.549
3.850
2.163
5.716
2.295
1.684
3.722
3.243
1.464
5.153
1.089
3.961
1.980

1.943
1.943
2.015
2.132
1.943
2.132
2.015
2.015
2.015
2.015
2.015
2.132
2.132
2.015

2.341
0.719
4.482
5.852
9.410
6.099
13.684
8.761
3.456
2.296
11.007
0.132
9.620
4.168

1.895
1.895
1.943
1.943
1.895
1.895
1.943
1.895
1.943
2.015
1.895
1.943
1.895
2.353

a Calculated t.
b Critical one-tail t.

Table 4. Net variation (mean 95% confidence limits) in constituents of water passing through trout-farms
Parameter
Total alkalinity, mg CaCO3 litre- 1
kg CaCO3 day -~
Total hardness, mg CaCO3 litre -~
kg CaCO3 day -~
Suspended solids, mg litre -1
kg day -1
Conductivity (20C), mS cm - l
S cm-lday -I
Dissolved oxygen, mg iitre -la
kg day -~a
BOD 5, mg litre- ~
kg day-~
Permanganate value, mg 0 2 litre- i
kg 02 day J
NH4-N, mg litre-~
kg day -~
PO4-P, mg litre- 1
g day-J

Farm A

Farm B

Farm C

1.9 + 0.6
11.7 + 3.9
------1.3 + 0.5
8.2+3.2
0.9 + 0.4
5.3 + 2.6
__
-0.38 + 0.15
2.360.92
0.086 0.032
533.6 195.2

2.6 + 1.3
56.9 + 29.0
--1.8 + 1.5
38.9-4- 33.2
--0.7 + 0.3
14.1 +7.4
1.8 + 1.1
37.84- 22.9
--0.27 + 0.10
5.762.19
0.060 0.030
1297.0 641.8

3.2 + 1.4
174 76
3.2 + 1.1
171 + 57
16 + 5
877 282
19 4-3
1040+ 149
2.4 + 0.5
128+29
14 -4-8
755 -4-428
3.6 3.1
196+ 167
1.46 0.26
78.5 14.0
0.579 O.118
31227 + 6362

a _ decrease.

Table 5. Daily net mass flow variations in constituents of water per ton of fish (annual production)
Parameter
Total alkalinity, g CaCO3 t -~ day -1
Total hardness, g CaCO 3 t -~ day -~
Total suspended solids, g t -1 day -1
Conductivity (20C), mS cm -1 t -~ day -1
Dissolved oxygen, g t-1 day-i
BODs, g t -1 day-1
Permanganate value, g 02 t-~ dayNH4-N, g t -1 day - l
PO4-P, g t - i day- 1
Mesophilic bacteria, 106 colonies t -1 day -1

Farm A

Farm B

Farm C

783
---549
353
-157
36
224

1035
-707
-255
688
-105
24
300

348
342
1753
2081
256
1510
392
157
62
506

384

R. Boaventura et al.

are observed at farm A, where the annual fish production is lower. Nevertheless, variations of DO and
N H 4 - N mass flows are, as would be expected, higher at
farm B.
Total alkalinity increased from the inlet to the outlet
of each farm and variations range between 1.9+0.6
(farm A) and 3.2+-1.4mg CaCO3 litre -1 (farm C).
These values correspond to mean daily productions of
783 and 348 g CaCO3 t -1 of annual production of fish,
respectively. Farm B daily production is three times
higher: 1035 g CaCO3 t -1 offish.
Net increase of total hardness is insignificant at farm
A. At farms B and C, however, hardness increased, on
average, 1.7+1.5 and 3 . 2 a : l . l m g CaCOslitre -1,
which equates to a net daily production of 342 and
559 g CaCO3 t -l, respectively. Similarly, the electric
conductivity increase is only significant in farm C. The
reported mean increment is 19+-3 mS cm -~, corresponding to a daily increase of 2081 mS cm-lt-~ of fish.
The DO concentration decrease between the
water inlet and outlet averaged 2.4+-0.5 mg litre -1 at
farm C, but the reduction was lower at farms B
(0.7+0.3 mg litre -l) and A (1.3+0.5 mg litre-l).
Even at farm C, however, the effluent DO concentration
maintained a high value, varying from 7.5 to
8.5 mg litre -~. Corresponding results are reported elsewhere (Alabaster, 1982; Bergheim & Selmer-Olsen,
1978; Solbr, 1982). The DO daily mass flow decrease
reported to the annual fish production is quite similar at
farms B and C but about twice as high at farm A. The
anomaly is probably related to differences in the tank
configuration and a lower reaeration rate when water
flows through successive tanks at farm A.
Water from the Inha and Fornelo rivers had BOD5
ranging between 0.4 and 1.6 mg litre-l; in the Coura
River it was somewhat higher, not exceeding, however,
the value of 2.7 mg litre-t The highest BOD5 values
were found at farm C effluent (between 6.8 and
39.4 mg litre-l), with an increase of 14+8 mg litre -~
between water inlet and outlet. At farms A and B
the increases were, on average, 0.9+0.4 and
1.8+1.1 mg litre -l, respectively. The daily output of
organic matter during the observation period, expressed
in terms of BODs, was about 353 g t -~ annual production of fish at farm A, 6 8 8 g t -1 at farm B and
1510 g t -1 at farm C.
The maximum increase in the NH4-N concentration
was observed at farm C (1.46+0.26 mg litre-~), corresponding to a daily increment of 157 g t-1 annual trout
production. At farms A and B daily increments were
157 and 105 g t -1 annual production, respectively.
These values are relatively low compared with the usual
reported range of 100-1500 g t -1 (Alabaster, 1982).
The highest increase of PO4-P was also at farm C
(0.579+0.118 mg litre -1) and the lower one at farm B
(0.060 + 0.030 mg litre-1). The calculated daily net mass
flow for PO4-P varied from 24 (farm B) to 62 g t -1
annual production (farm C). These values are near the
lower limit of the range (30-300 g t-lday -1) usually
found in trout farm effluents (Alabaster, 1982). The

annual outputs (8.8-22.6 kg t-lyear -l) are comparable

to the mean P release (16 kg t-lyear -1) from UK farms
producing mainly rainbow trout (Solbr, 1982). Annual
loading rate from a Northern Ireland farm (Foy &
Rosell, 1991a) was 25.6 kg total P per ton of rainbow
trout produced. Assuming that soluble reactive P,
according to the same authors, is 60.0% of total P, the
annual output is 15.4 kg t -l, a value close to that
obtained in UK farms.
Total suspended solids mean concentrations ranged between 1.8+l.5mglitre -~ at farm B and
16+5 mg litre -~ at farm C. These levels are considerably low when compared with typical situations in fish
farms (Sumari, 1982). The daily contribution to the
suspended solids production was about 707 and
1753 g t -~ of annual production at farms B and C,
respectively. As previously reported, the suspended
solids net difference at farm A was not significant. The
total annual amount produced at farm B is below
the range (1.3-11 kg t -1) mentioned in the literature
(Alabaster, 1982).
In terms of daily mean increment, mesophilic bacteria
concentration varied between 224106 (farm A) and
506 106 (farm C) colonies per ton of annual production
of fish. The corresponding value at farm B was
300 106 t -1.
Impact on river water quality

The mean values of the physical, chemical and bacteriological parameters in the sampling stations upstream
and downstream from trout farms are shown in Table 6.
The longitudinal concentration profiles (mean+95%
confidence limits) for BODs, DO, NH4-N, PO4-P
and aerobic mesophilic bacteria are presented in Figs 2
and 3. The electric conductivity increases, at a statistically significant level, only at farm C. Also turbidity
seems to increase only in stations C2 and C3, but the
high dispersion of the individual results makes impossible a safe conclusion. So, the corresponding longitudinal profiles were drawn only for the Coura River
(Fig. 2(c) and Fig. 2(d)). The first point corresponds to
the feeding water in all the figures.
Effluent BOD5 was 2.1+0.6, 3.2+-1.1 and 15.7+
10.0 mg litre -1 at farms A, B and C, respectively. After
dilution by river water, maximum BOD5 amounted
to 1.7+0.2mglitre -l in the Fornelo River, 2.5+
0.6 mg litre -1 in the Inha River and 5.6+2.1 mg litre -1
in the Coura River (Fig. 2(a)). The running water selfpurification capacity rapidly reduced the BOD level. In
the Fornelo and Inha Rivers, BOD5 returned to the
feeding water values (1.1 +0.3 and 1.3+-0.3 mg litre -1,
respectively), about 2 and 3 km downstream from the
effluent discharges (0.94-0.4 and 1.3+0.3 mg litre -~,
respectively). In the Fornelo River, however, there
was a slight increase between stations A3 and A4, but
even at station A4 the 95% confidence range is lower
than at station A2, where the maximum values were
recorded. A more marked effect was observed in
the Coura River (maximum of 5.6+2.1 mg litre -~ at
station C2) and BOD5 only recovers at station C6,

Trout farm effluents

a b o u t 12 k m below the f a r m discharge ( 1 . 9 + 0 . 6 m g
litre - ] , n o t statiscally different f r o m the feeding water
BODs, 1 . 6 + 0 . 4 m g litre-]). This p a r a m e t e r exceeded
the m a x i m u m r e c o m m e n d e d level (EEC, 1978) for

385

s a l m o n i d waters (3 m g litre -~) in the 1-km reach d o w n stream f r o m the effluent discharge.
Effluents f r o m the fish farms were well o x y g e n a t e d
( 9 . 6 + 0 . 3 , 1 0 . 5 + 0 . 9 a n d 8 . 5 + 0 . 6 mgO2 litre -1 at farms

T a b l e 6. M e a n v a r i a t i o n s in fiver w a t e r c o m p o s i t i o n

Station

BOD5
mg litre -~

DO mg litre -1

NH4-N
mg litre -~

PO4-P
mg litre -]

Farm A
A1
A2
A3
A4

1.1
1.7
0.9
1.0

10.8
10.7
11.1
10.3

0.04
0.27
0.11
0.04

0.011
0.096
0.026
0.025

Farm B
BI
B2
B3
B4
B5

1.3
2.5
2.2
1.3
1.2

11.1
9.7
10.7
11.2
10.8

0.04
0.20
0.16
0.09
0.10

Farm C
C1
C2
C3
C4
C5
C6

1.6
5.6
5.6
2.5
2.0
1.9

10.7
9.6
10.0
10.8
11.3
11.0

0.06
0.89
0.95
0.32
0.18
0.11

Mes. bacteria
colonies m1-1

Conductivity
/zS crn-1

Turbiditya
NTU

174
496
440
356

43
47
50
60

1.3
1.8
-1.3

0.005
0.090
0.099
0.078
0.075

79
3069
1725
395
154

85
87
87
84
83

2.4
2.7
2.8
2.8
2.7

0.005
0.468
0.459
0.217
0.192
0.183

191
13 360
52 064
16 517
2777
5509

40
52
52
49
47
45

2.6
4.7
4.7
2.5
2.4
2.1

a Individual values, mean of 2 values and mean of 3 values for Farms A, B and C, respectively.

10~

(a)

(b)

---.o--- FarmA
FarmB
FarmC

9
8:

"-4 5
o 4

---a.-- FarinA
*
FarmB

FarmC

-1000
6o

1000

3000

5000
7000
Distance(m)

9000

-1000

I1000 13000

1000

3000

5~
7~
Distance(m)

9000

11000 130OO

(d)

(c)

55

FannC

so

'~ 45
u 40
"5

m 35
3 0

a FarmC
,

-1000

1000

3000

5000
7000
Distance(m)

9000

11000 13000

-1000

1000

3000

5000
7000
Distance(m)

9000

11000 13000

Fig. 2. BOD5 (a) and DO (b) mean variations (mean + 95% confidence limits) in the Fornelo (Farm A - 15 t year-l), Inha (Farm
B - - 55 t year -1) and Coura (Farm C - - 500 t year -I) Rivers, upstream and downstream from effluent discharges; electric
conductivity (c) and turbidity (d) variations (mean 4- 95% confidence limits) in the Coura river.

386

R. Boaventura et al.

A, B and C, respectively). Nevertheless, the organic

matter biodegradation caused slight decreases immediately downstream from the effluent discharges (Fig. 2(b)).
The lowest concentrations were observed in the Coura
River (station C2) but they decreased, on average, from
10.74-0.8 mg litre -1 upstream from the discharge to
9.6 4- 1.1 mg litre-I immediately downstream from the
discharge. No injurious effects to aquatic life would be
anticipated at these concentrations.
The NH4-N concentration in the effluents from farms
A, B and C was 0.464-0.17, 0.27+0.08 and 1.624-0.30
mg litre -1, respectively. In spite of dilution with uncontaminated river water (0.044).06 mgNH4-N litre-1),
1.5 I

(a)
+

FarmA
*

Farm B

Farm C

1.0

0.5

0.0

'

-1000

1000

'

3000

'

5000

"

" '

'

7000

'

'

'

'

9000

. . . . . . .

11000

13000

Distance (m)
0.8

Co)
n
,

0.7

Farm A
Farm B
Farm C

0.6

0.5

t~

0.4
0.3

0.21

0.1

0.0

.-.

1000

3000

-..

5000

9000

7000

11000

13000

Distance (m)

10000~

(c)
+

FarmA

,ooooo

Farm B

,,

Farm C

e.

I0000

,ooo

tO0

10 "

JlO00

1000

3000

5000

7000

9000

I I 0 0 0

13000

Distance (m)

Fig. 3. NH4-N (a), PO4-P (b) and mesophilic bacteria (c)

variations (mean95% confidence limits) in the Fornelo
(Farm A - - 15 t year-l), Inha (Farm B -- 55 t year-1) and
Coura (Farm C - - 500 t year-1) Rivers, upstream and downstream from effluent discharges.

high NH4-N levels occurred downstream from the three

trout farms (Fig. 3(a)). Maximum concentrations of
0.27+0.09, 0.20+0.07 and 0.95+0.11 mg litre -1 were
observed in the Fornelo, Inha and Coura Rivers,
respectively. Even though the NH4-N content decreased
downstream, moderately high concentrations (twice the
feeding water concentration) were still present at the
end of the Inha and Coura River study reaches
(0.10+0.04 and 0.11+0.05mglitre -1, respectively).
The maximum concentrations exceeded the recommended limit for salmonid waters (0.04 mgNH4 litre -1) but
they remain below the maximum allowable level
(1 mg litre -1) (EEC, 1978) even in the Coura River.
However, NH4-N was already above the recommended
level upstream from farms B and C.
The longitudinal concentration profiles for POa-P
(Fig. 3(b)) show a pattern similar to that observed for
ammoniacal nitrogen. The PO4-P content markedly
increased downstream from the trout farms, which has
been reported by other authors (Carr & Goulder, 1990),
with maximum values immediately downstream from
the farms of 0.099 + 0.050 mg litre-1 in the Fornelo and
Inha Rivers and 0.468 +0.172 mg litre -1 in the Coura
River. The POa-P content in the Fornelo River rapidly
nearly reached the feeding water level but only a slight
decrease was observed in the Inha river and concentrations stayed above 0.075 mg litre-l; in the Coura River
concentrations almost decreased to 0.2 mg litre -l along
a 6 km reach but remained constant downstream. Concentrations up to 1 mg litre-1 can be considered as not
impairing aquatic life (EEC, 1978).
A characteristic effect of trout farm effluents on river
water is an increase in the number of microorganisms,
especially bacteria (Sumari, 1982). The effluent aerobic
mesophilic bacteria densities during the study period
showed great variation at each farm, making the 95%
confidence levels very large. The calculated values were
662651, 8424-751 and 4723+ 1691 m1-1 at farms A,
B and C, respectively. Figure 3(c) illustrates changes in
the mesophilic aerobic bacteria abundance downstream
from the discharges. Maximum measured densities were
4964-455 m1-1 in the Fornelo River (station A2),
30694-3493 ml -I in the Inha River (station B2) and
52 064+88 171 m1-1 in the Coura River (station C3).
Mean bacterial levels in the Coura River decreased to
2780 bacteria m1-1 at station C5, even though the mean
value (5500 bacteria m1-1) increased again at station
C6, 12 km downstream from the farm. The mean value
at this station is the result of an abnormally high value
(10 000 bacteria ml-1) obtained in one trip. Eliminating
this value, the bacterial density decreased downstream
from station C3. Farms A and B induced a more limited
effect and the mean bacterial densities returned to the
feeding water mean values 4 km downstream from the
discharge.
The effect of effluent discharges on water conductivity
and turbidity was evaluated in all rivers but only detected in the Coura River (Fig. 2(c) and Fig. 2(d)). The
effluent conductivity at farm C was 58 4- 3/zS cm -1 and
the maximum registered in the river (station C2) was

Trout farm effluents

5 2 + 2 / z S cm -l, slightly higher than the feeding water

conductivity (39 + 2/xS cm-1). The corresponding effluent turbidity was 11 3 NTU and the higher values in
the river occurred at stations C2 and C3 (4.7 1.6 and
4.7 2.2 NTU, respectively).
The pollution load per unit of dry weather flow may
be a useful parameter for the impact assessment of discharges on the receiving streams when water temperature, species, feeding and farming practices do not vary
much as in Northern Portugal. Assuming that the
pollution load in the critical dry weather conditions is
proportional to the annual fish production, the ratio
P : D W F (annual fish production, t:dry weather flow,
m 3 s -I) should be an approximate parameter for calculating the effected downstream distance from similar
trout farms, in a given region. Considering only the
analytical chemical data, it is possible to estimate (from
Figs 2 and 3) the effected distances of about 3 km for
P : D W F = 75 (farm A), of about 5 km for P : D W F = 183
(farm B) and of about 12 km for P:D W F = 500 (farm
C). So, the relationship between the length of polluted
stretch (Lp, km) and the ratio P : D W F is given by:
Lp = 1.264 + 0.0214 (P : D W F ) r2 ~_ 1

CONCLUSIONS
In the present study, water consumption in trout farms
was found to be below typical values in other countries
(maximum of 4.8 litre s -1 t -l annual trout production,
against about 5 litre s-~t -I in Denmark, 7 litre s-It -l
in Northern Ireland, 10 litre s-it -~ in Finland and
35 litre s-It -1 in Italy). On the other hand, annual fish
productions per dry weather river flows are higher.
Loadings of total alkalinity, total hardness, BODs,
NH4-N, PO4-P, total suspended solids and mesophilic
bacteria increased in general as water passed through
the farms. DO concentration showed a slight decrease
(maximum decrease of 2.4 mg litre -1 at farm C, but
always kept above 7.5 mg litre -1) and nitrates were
unchanged. NH4-N daily increments were between 105
and 157 g t -l annual production, at the low end of
most farms (100-1500 g t-l). PO4-P was between 24
and 62 g t -1, close to the lower limit of the typical tange
in trout farms: 30-300 g t -l. Nevertheless, the impact of
these two constituents on the receiving streams is still
detectable at the end of the study reaches of farms B
and C, although the concentrations are below the maximum recommended level for salmonid waters. Organic
matter has a slight impact on rivers, either in terms of
BOD5 or DO. Maximum BOD5 was observed in the
Coura river, the receiving stream of the largest fish farm
effluent, but concentration only increased by a factor of
4 regarding the upstream value. The associated DO
depletion doesn't exceed 1.1 mg litre -1. Bacterial contamination downstream from the trout farms is an
important feature but only for the largest trout farm the
impact extends over all the study reach. Considering
only the effect on the chemical composition of the

387

receiving stream, a linear relationship between the ratio

P : D W F (annual fish production, t:dry weather flow,
m 3 s -l) and the extension of the polluted stretch was
obtained.

REFERENCES

Alabaster, J. S. (1982). Survey on Fish-Farm Effluents in some

EIFAC Countries. In: Report of the EIFAC Workshop on
Fish-Farm Effluents, ed. J. S. Alabaster. Silkeborg, Denmark, 26-28 May 1981, European Inland Fisheries, Advisory Commission, Technical Paper No. 41, Food and
Agriculture Organization of the United Nations, Rome,
pp. 5-15.
Bergheim, A., Hustveit, H., Kittelsen, A. & Selmer-Olsen, A.
R. (1984). Estimating pollution loadings from Norwegian
fish farms. II. Investigations 1980-1981. Aquaculture, 36,
157-168.
Bergheim, A. & Selmer-Olsen, A. R. (1978). River pollution
from a large trout farm in Norway. Aquaculture, 14, 267270.
Carr, O. J. & Goulder, R. (1990). Fish-farm effluents in rivers.
II. Effects on inorganic nutrients, algae and the macrophyte
Ranunculus penicillatus. War. Res., 24, 639-647.
European Economic Community (1978). Council Directive of
18 July on the quality of fresh waters needing protection or
improvement in order to support fish life, 78/659/EEC,
Brussels.
Foy, R. H. & Rosell, R. (1991). Loadings of nitrogen and
phosphorus from a Northern Ireland fish farm. Aquaculture,
96, 17-30.
Foy, R. H. & Rosell, R. (1991). Fractionation of phosphorus
and nitrogen loadings from a Northern Ireland fish farm.
Aquaculture, 96, 31-42.
Gowen, R. J., Rosenthal, H., M/ikinen, T. & Ezzi, I. (1990).
Environmental impacts of aquaculture activities. In: Aquaculture Europe'89 - Business Joins Science, eds N. de Pauw
& R. Billard. European Aquaculture Society, Spec. Publ.
No. 12, Bredene, Belgium, pp. 257-283.
Iversen, T. M. 0995). Fish farming in Denmark: environmental impact of regulative legislation. Wat. Sci. Tech.,
31(10), 73-84.
Kronvang, B.,/Ertebjerg, G., Grant, R., Kristensen, P., Hovmand, M. & Kirkegaard, J. (1993). Nationwide monitoring
of nutrients and their ecological effects: state of the Danish
aquatic environment. Ambio, 22, 176-187.
Silvert, W. (1992). Assessing environmental impacts on finfish
aquaculture in marine waters. Aquaculture, 107, 67-79.
Solbr, J. F. de L. G. (1982). Fish-Farm Effluents: A United
Kingdom Survey. In: Report of the EIFAC Workshop on
Fish-Farm Effluents, ed. J. S. Alabaster. Silkeborg, Denmark,
26-28 May 1981, European Inland Fisheries, Advisory
Commission, Technical Paper No. 41, Food and Agriculture
Organization of the United Nations, Rome, pp. 29-55.
Standard Methods for the Examination of Water and Wastewater (1985), 16th ed., APWA, New York.
Sumari, O. (1982). A Report on Fish-Farm Effluents in Finland. In: Report of the EIFAC Workshop on Fish-Farm
Effluents, ed. J. S. Alabaster. Silkeborg, Denmark, 26-28
May 1981, European Inland Fisheries, Advisory Commission, Technical Paper No. 41, Food and Agriculture
Organization of the United Nations, Rome, pp. 21-27.
Warrer-Hansen, I. (1982). Evaluation of matter discharged
from trout farming in Denmark. In: Report of the EIFAC
Workshop on Fish-Farm Effluents, ed. J. S. Alabaster.
Silkeborg, Denmark, 26-28 May 1981, European Inland
Fisheries, Advisory Commission, Technical Paper No. 41,
Food and Agriculture Organization of the United Nations,
Rome, pp. 57-63.