Selection for growth in brown trout increases feed intake capacity
without affecting maintenance and growth requirements1
M. Mambrini*2, F. Médale†, M.-P. Sanchez*3, B. Recalde†, B. Chevassus*,
L. Labbé‡, E. Quillet*, and T. Boujard*
*INRA, Laboratoire de Génétique des Poissons, 78350 Jouy-en-Josas, France; †Unité Nutrition,
Aquaculture et Génomique, 64310 Saint Pée sur Nivelle, France; and ‡SEMII, 29450 Sizun, France
ABSTRACT: The correlated responses in feed intake
and G:F ratio with selection for increased growth rate
were evaluated by comparing selected (S) and control
(C) brown trout (Salmo trutta) reared under conditions
known to affect feed efficiency: feed restriction and periods of compensatory growth. Nitrogen and energy requirements for maintenance and growth were also measured. Trout were allotted at comparable BW (3.7 ±
0.06 and 3.8 ± 0.04 g, for C and S respectively) to triplicate groups per treatment. The experiment lasted a
total of 198 d, during which animals were successively
submitted to a 116-d feeding phase and fed 10, 30, 50,
70, 100, and 140% of their usual daily ration (UDR), a
35-d phase of food deprivation, and a 47-d refeeding
phase. The G:F of C and S were comparable in all experimental conditions tested. During the feeding phase, S
grew better than C only when fed 100 and 140% UDR
(P < 0.001). This was explained by a higher feed intake
capacity. The requirements for growth and maintenance were similar among the lines, which is in
agreement with their comparable loss of weight (mean
energy loss of −53 and −55 kJ/(kgⴢd) for C and S, respectively; P > 0.38) observed during the feed deprivation
phase and the lack of differences in carcass composition
(fat, P > 0.35; protein, P > 0.54). During the refeeding
phase, growth performance and G:F were high in all
groups. The daily growth coefficient was higher in S
than in C (P < 0.001) because of a higher feed intake
(P < 0.001). An increase in absolute individual variability in final BW and length was associated with the
level of food restriction in both lines; however, it always
remained lower in S than in C. In conclusion, fish selected for growth under ad libitum conditions will only
exhibit growth superiority when fed diets close to ad
libitum, and there was no evidence that selection was
associated with an improvement in efficiency of maintenance nor in retention of body tissues.
Key Words: Correlated Responses, Energy Requirements, Feed Intake, Growth Rate, Salmo trutta, Selection
2004 American Society of Animal Science. All rights reserved.
Introduction
Selection for growth in fish using familial or individual procedures leads to relatively high genetic gains
(10 to 20% per generation; Gjedrem, 1998; Vandeputte
et al., 2002). Although selection for growth is more recent than in other farmed animals, the selected lines
of fish are now sufficiently divergent to carry out an
analysis of the correlated responses. As in higher verte-
1
The authors thank I. Quéau for the daily care of the fish and the
team at the SEMII fish farm for their help. The technical assistance
of D. Blanc and M. J. Borthaire is also gratefully acknowledged. We
also wish to thank J. B. Denis for his statistical expertise.
2
Correspondence: Domaine de Vilvert (phone: 33 1 34 65 27 05;
fax: 01 34 65 23 90; e-mail: mambrini@jouy.inra.fr).
3
Present address: INRA, Station de Génétique Quantitative et
Appliquée, Domaine de Vilvert, 78352 Jouy-en-Josas Cedex, France.
Received December 17, 2003.
Accepted June 2, 2004.
J. Anim. Sci. 2004. 82:2865–2875
brates (Rauw et al., 1998), the enhanced growth rates
are explained by an increased feed intake associated
(Thodesen et al., 1999) or not associated (Sanchez et
al., 2001) with improved G:F. Response variability may
be explained by the different selection procedures (familial or individual), the selection criteria (BW or
length), the basis of comparisons (wild or domesticated
control lines), and the ways fish are fed and feed intake
is measured. In fish, the group level is generally the
experimental unit. This implies specific experimental
designs, particularly when intake is considered.
Brown trout have been selected by an enhanced individual procedure since 1987 (Chevassus et al., 1992),
and a control line has been maintained. It seems that
selected fish exhibit their growth potential when fed ad
libitum only (Sanchez et al., 2001), and G:F may not be
affected by selection. This suggests that the nutritional
requirements are not different between the lines. The
objective of this study was to verify that G:F is not
affected by selection. Our strategy was to test the ro-
2865
2866
Mambrini et al.
bustness of the response by analyzing the genotype
× environment interactions when the lines are reared
under conditions known to affect feed efficiency: feed
restriction (Cho, 1992) and conditions for compensatory
growth (Dobson and Holmes, 1984). The specific effects
of such rearing conditions are not well known in brown
trout, so they will be described first in the controls.
The specific effect of selection will then be assessed.
In addition, the experiment was designed to compare
maintenance and growth requirements of both lines.
Indeed, the specific effect of selection for growth on
nutritional requirements has never been described in
fish.
after controlling the distribution of BW and L in the
sorted populations (299 C and 302 S). Each line was
divided into nine sets of 100 and nine sets of 150 fish
reared in 180 L flow-through tanks supplied with bore
hole water (temperature 11 to 12°C).
The experiment was conducted in a manner compatible with national legislation on animal care. In application of the French penal and rural code, M. Mambrini
and T. Boujard have personal authorization delivered
by the French Agricultural Ministry for conducting animal experiments (Authorizations 5625 and 5635) and
ensured that all procedures used in this experiment
were ethically acceptable and followed the procedures
stipulated by the French Ministry of Research.
Materials and Methods
Experiment
Animals
A specific strain of brown trout (Salmo trutta) was
selected by an individual selection process for five generations in our experimental fish farm (SEMII, Sizun,
Brittany, France). This strain (population reared in
northern France) was chosen because of its better
growth performance compared with the other brown
trout strains initially tested (French and Danish reared
and wild populations). The criterion of selection was
fork length (L, fish length measured at the clearance
of the caudal fin), mainly because it is easier to measure
than weight (Chevassus et al., 1992). From the same
initial population, a control line (C) was maintained
under the same rearing conditions. With a selection
pressure on L of approximately 5% (proportion of fish
selected), the correlated weight gain at 1 yr of age was
approximately 24% per generation (on average over
four generations) for the selected line (S) compared with
the C (Vandeputte et al., 2002). The S and C groups of
the current study were produced by in vitro fertilization
of 34 S females crossed with 11 S males and 27 C females crossed with 37 C males. To obtain selected and
control offspring reaching a similar weight of approximately 4 g at the beginning of the experiment, we took
into account the value of the growth rate usually observed for such lines (our unpublished data; Sanchez
et al., 2001) and calculated that C had to be fertilized
12 d before S.
After hatching, fingerlings were reared in two tanks
per line supplied with 11°C flow-through water. They
were fed ad libitum a commercial dry feed (Biomar
Ecolife 18 (Nersac, France) containing 51% CP and 18%
lipid (data from the manufacturer) by automatic feeders
delivering food 12 h/d until the beginning of the experiment. The day before allotment, the distribution of
length of the two populations was analyzed to optimize
the constitution of the experimental groups. This was
based on individual L of 580 C (range recorded = 55 to
81 mm) and 782 S (range recorded = 55 to 91 mm) fish.
The individual BW were also recorded. Trout were then
sorted to obtain homogeneous groups with L between
60 and 80 mm. The effect of the sorting was assessed
The experiment was divided into three successive
phases: 1) the feeding phase, where fish were fed six
different feeding levels; 2) the feed deprivation phase;
and 3) the refeeding phase, where all groups of fish
were fed in excess.
During the feeding phase, groups of 150 fish were fed
at 10, 30, or 50%, whereas groups of 100 fish were fed
at 70, 100, or 140% of the usual daily ration (UDR,
calculated using the prediction software “Ecureuil,” developed at the SEMII fish farm, taking into account the
fish species, the water temperature and the weight of
the fish). The group size was chosen so that the biomass
expected at the end of the experiment remained much
lower than 100 kg/m3, a biomass under which growth
performance remains optimal in salmonids (Boujard et
al., 2002). The number of individuals in the more feed
restricted groups was higher compared with the other
groups because of the expected mortality. Moreover,
the induced difference in density does not affect the
behavior of the fish at this biomass level (Boujard et
al., 2002). The fish were fed by automatic feeders in a
2 × 6 × 3 factorial design (line × feeding level × replicates). The uneaten food was collected at the outlet of
each tank three times per week, dried, and weighed,
and this weight was subtracted from the amount of
food distributed to calculate the actual amount of feed
consumed. Fish were fed for 116 d, except those fed 10
and 30% UDR, for which the experiment was stopped
after 82 d. As a higher mortality was expected in those
groups, plans were implemented to stop a feeding level
if mortality was close to 10% of the assigned individuals.
Only the fish fed 50 to 140% UDR were included in the
next two phases of the experiment.
The phase of feed deprivation lasted for 35 d. During
the refeeding phase, fish were all fed 140% UDR for 47
d. As during the feeding phase, fish were fed by automatic feeders and the actual amount of feed consumed
was deducted, taking into account the uneaten food.
The diet used throughout the experiment contained
(DM basis) 49.4% CP, 25% crude fat, and 23.7 kJ of
GE/g. This high-protein, high-energy diet consisted of
extruded pellets of different diameters that were kept
2867
Selection for growth of brown trout
at 4°C until use. Pellet size was chosen according to
the fish size.
Measurements and Calculations
Mortality was recorded daily throughout the experiment and the weight of the dead fish was taken into
account for weight gain calculations. At the end of each
experimental phase (feeding, feed deprived, and refeeding) for the groups initially fed 50 to 140% of their UDR,
and on d 82 only for those fed 10 and 30% UDR, the
fish were counted and weighed by group after 1 d of
fasting. Growth coefficients were calculated according
to Cho (1992), using the daily growth coefficient (DGC),
a value independent of the BW. Gain:feed ratios were
calculated taking into account the actual feed consumed. Individual weights (grams, noneviscerated) and
fork lengths (to the nearest millimeter) were recorded
for all fish. Coefficients of variation of weight, length,
and the conformation coefficient (K-factor) were calculated within each tank.
Twenty fish were initially sampled for whole body
composition. The whole body samples were frozen
(−20°C), ground, and freeze dried. At the end of the
feeding and feed deprivation phases, 10 fish per tank
were sampled and immediately frozen for subsequent
analysis. The following analyses were performed on the
diet and whole carcasses: DM, CP (N × 6.25), GE, and
fat (AOAC, 1990). The water content was calculated by
weight loss after drying at 110°C for 24 h in a forced-air
oven. Nitrogen analysis of all samples was performed by
the Kjeldahl method. The GE content of the diet was
determined with an adiabatic bomb calorimeter (IKA,
Staufen, Germany). Fat content was calculated after
extraction in a Soxhlet apparatus using petroleum
ether as solvent.
Protein and energy gains were plotted against protein
and energy intake expressed as g/(kg BW0.75ⴢd) and kJ/
(kg BW0.75ⴢd), respectively, to estimate maintenance
and growth requirements. Data were adjusted using
linear model and saturation kinetics model (Mercer,
1982), which both led to similar estimates. Maintenance
requirements were interpolated using the linear model,
as the amount of protein and energy intake resulting
in protein and energy gain equal to zero (x-intercept),
and requirements for growth were estimated as the
inverse of the slope. In addition, for feeding levels
higher than 30% UDR, protein and energy use for body
accretion were calculated as the intake of protein and
energy per kilogram of wet weight gain during the feeding phase. Nutrient losses during the food deprivation
phase were deducted from the body composition data.
Statistical Analyses
Statistical analyses were based on a completely random experimental design. Group data were compared
using an analysis of covariance, including the effects
of line, feeding level, and interaction between these
two factors, which were tested using the tanks as the
experimental unit (three replicates per treatment). Percentage data were transformed (arcsine square root)
before being subjected to the analysis.
For the group growth data, the covariate included in
the model was the initial BW of each experimental
phase. For the feeding phase, the initial BW was taken
into account because of a slight difference between the
two groups (see results). For the feed deprivation and
the refeeding phases, the mean BW at the end of the
previous phases were used as a covariate. For the body
composition data, the covariate included in the model
was the final BW because it is accepted that body composition greatly influences BW (Shearer, 1994).
Individual data were compared using an ANOVA including the effects of the line, the feeding level, the
interaction between these two factors, and the effect
of the tank nested into this interaction (100 to 150
individuals per tank, with three replicates per
treatment).
Probabilities of differences between treatments and
residual standard deviation were generated after ANOVA was performed using the GLM procedure of SAS
(SAS Inst., Inc., Cary, NC). When the interaction was
significant, the respective effects of the line and the
feeding level were tested separately with one-factor ANOVA. The means were subsequently compared using
the test of Newman and Keuls (with P < 0.05).
To compare the extent of the absolute variation for
each individual variable (BW, length, K-factor), they
were transformed to logarithms, and the equality of
the absolute deviates in each class was controlled. The
absolute deviate was calculated as (|lnYij − lnỸj|), where
Yij is the jth term in the ith sample, and lnỸi is the
mean logarithm of the ith sample. The scatter of the
absolute deviates was markedly unequal, indicating
that a nonparametric test has to be applied (Sokal and
Braumann, 1980). The CV were thus compared with
the nonparametric test of Kruskal-Wallis using the
NPAR1WAY procedure of SAS. The line and feeding
level effects were tested separately (three replicates
per treatment).
Results
Allotment
Before the allotment, and although S were fertilized
after C, S were unexpectedly heavier and longer than
C (BW = 3.8 vs. 4.4 g, L = 67 vs. 70 mm for C and S,
respectively; P < 0.001). Conversely, the K-factor was
not different between the two lines (1.27 and 1.26 for
C and S respectively, P > 0.87). The mean L was not
affected by the sorting out, and C remained shorter
than S (6.8 and 7.0 cm for C and S, respectively, P <
0.001). The sorting out tended to lead to a decrease in
the mean BW, although C remained lighter than S (3.7
and 3.8 g for C and S, respectively; P < 0.001). As a
consequence, the sorting induced a decrease in the K-
2868
Mambrini et al.
Table 1. Growth and final body composition of brown trout selected for growth (S) or controls (C) fed at 10 and 30%
of the usual intake (as-fed basis) for 82 da
Feeding level and line
10%
Measurements
P-valueb
30%
C
S
C
S
SEM
L
F
L×F
Tank
3.7
1.3
0.11
0.53
3.8
1.3
0.08
0.39
3.6
3.1
0.51
1.27
3.8
3.1
0.48
1.22
—
0.01
0.008
0.031
—
0.87
0.96
0.67
—
0.001
0.001
0.001
—
0.75
0.90
0.53
—
—
—
—
0.28
0.66
1.18
0.53
0.70
0.20
0.32
0.32
0.68
0.72
0.08
0.11
0.99
0.52
0.11
0.48
—
—
—
—
0.11
0.04
0.006
0.31
0.006
0.006
0.001
0.001
0.001
0.90
0.63
0.61
0.001
0.001
0.001
—
—
—
0.45
0.81
0.21
0.001
0.001
0.46
—
—
—
—
—
—
c
Tank records
Initial mean BW, g
Cumulative intake, g/fish
Daily growth coefficient, %/dd
G:F
Final body composition, % or kJ/g BWe
Water
Protein
Fat
Energy
Individual records nested by tanksf
Final BW, g
Final length, cm
K-factor, g/cm3g
CV, %h
Final BW
Final length
K-factor
74.5
15.3
5.0
6.1
4.1
7.3
1.04
31.7
8.5
15.7
74.2
16.1
4.3
6.1
74.2
14.6
6.7
6.4
4.4
7.6*
1.00*
28.3
8.3
15.3
74.0
14.8
6.8
6.6
7.1
8.3
1.12
49.9
14.7
14.5
7.2
8.5*
1.10*
45.5
13.8
12.9
a
Data were obtained on three tanks per treatment, with 150 fish in each tank. Tank records are means of measurements performed on
each tank. Individual records are means and CV of measurements performed on each individual within each tank. When the line × feeding
method interaction was significant (P < 0.05), separate variance analyses testing the effect of the line within a feeding level were performed.
*Indicates when S was different (P < 0.05) from C at the corresponding feeding level.
b
Probability for the effects of line (L), feeding level (F), the line × feeding level (L × F) interaction and the tank effect.
c
Analysis of covariance.
d
100(Final BW ¹⁄₃ − Initial BW ¹⁄₃)/d.
e
Whole body initial composition: (for S) 76.8% water, 13.9% protein, 5.3% fat, 5.8% energy; (for C) 77.4% water, 13.6% protein, 5.1% fat,
5.5 kJ/g energy.
f
ANOVA.
g
100(BW/fork length3).
h
Kruskal-Wallis test – separate analysis.
factor, which was more important for S than for C (1.12
and 1.17 respectively, P < 0.001). Thus, there was a
slight but significant difference of BW, L, and K-factor
between S and C at the beginning of the experimental
period, which was taken into account in further
analyses.
Survival rate at the end of the experiment was higher
than 95% in groups fed at least 50% UDR. For groups
fed less than 50% UDR, the experiment lasted only for
82 d, and mortality ranged between 88 and 90%.
Feeding Phase
During the feeding phase, the effect of the line on
daily growth coefficient and feed intake depended on
feeding level (Tables 1 and 2). Daily growth coefficient
was significantly different (P < 0.001) between lines
only when they were fed 100 and 140% UDR. For groups
fed 10 to 70% UDR, no uneaten feed was detected, so
it was considered that all the feed distributed was
eaten. At these feeding levels, cumulative feed intake
was not different between C and S (Tables 1 and 2). At
100% UDR, uneaten feed was observed for C only (8.2
± 2.3% of the distributed feed). At 140% UDR, the proportion of distributed feed that was uneaten was higher
in C (14.9 ± 3.2%) than in S (9.4 ± 0.9%). Thus, the
main differences between lines in fish fed 100 and 140%
UDR was in their respective cumulative feed intake
(Table 2). Gain:feed ratio was affected by the feeding
level, being lower at the lowest feeding levels (10, 30,
and 50% UDR) and maximal at 100% UDR. Whatever
the feeding level, G:F was never different between lines.
Regarding the final body composition, protein content
values were comparable among the lines and the feeding levels. Fat content values increased with the feeding
level, but this was only because of differences in BW.
Indeed there was no significant effect of the treatments
when final BW was included as a covariate (Tables 1
and 2).
The protein and energy gains increased linearly with
the protein and energy intake, respectively (Figure 1),
without any differences between lines. The estimated
maintenance requirements were similar among the
lines: 8.1 and 8.5 g of CP/(kg BW0.75ⴢd) and 446 and 433
kJ of GE/(kg BW0.75ⴢd) for S and C lines, respectively.
The growth requirements were also comparable between the lines: 1.53 and 1.44 g of CP/g of protein gain
and 1.37 and 1.35 kJ of GE/kJ of energy gain for S
and C lines, respectively. The amounts of CP and GE
utilized to produce 1 kg of wet weight gain were also
similar for both lines at each feeding level tested (Table
2869
Selection for growth of brown trout
Table 2. Growth and final body composition of brown trout selected for growth (S) or controls (C) fed at 50, 70, 100,
and 140% of the usual intake (as-fed basis) for 116 da
Feeding level and line
50%
Measurements
C
70%
S
C
100%
S
C
P-valueb
140%
S
C
S
SEM
L
F
L×F
Tank
c
Tank records
Initial mean BW, g
3.6
3.8
3.7
3.8
3.7
3.8
3.7
3.8
Cumulative intake, g/fish
10.3
10.5
16.6
17.0
23.4
30.5*
26.3
41.4*
Daily growth coefficient, %/dd
0.88
0.87
1.26
1.25
1.55
1.77*
1.59
2.03*
G:F
1.31
1.32
1.44
1.39
1.46
1.47
1.36
1.40
Final body composition, % or kJ/g BWe
Water
72.3
72.4
70.7
70.6
69.5
69.2
69.7
69.3
Protein
15.4
15.6
16.1
15.8
16.4
16.3
15.9
15.9
Fat
7.8
7.8
9.1
9.1
9.9
10.3
9.9
10.9
Energy
7.1
7.0
7.7
7.8
8.1
8.2
8.0
8.3
Protein and energy utilization for body accretion
Crude protein, g/kg
354
369
324
335
317
318
340
333
Gross energy, kJ/kg
16.9
17.7
15.6
16.1
15.3
15.3
16.4
16.0
Individual records nested by tanksf
Final BW, g
16.8
17.2
27.5
27.1
37.6
48.1*
39.4
60.9*
Final length, cm
11.2
11.5*
13.0
13.1*
14.2
15.6*
14.4
16.6*
K-factor, g/cm3g
1.15
1.10*
1.24
1.17*
1.29
1.26*
1.30
1.31*
CV, %h
Final BW
29.2
30.5
22.6
22.9
21.5
18.3
22.0
16.7
Final length
9.7
10.3
7.4
7.5
7.0
6.2
7.0
5.5
K-factor
7.7
9.2
6.1
7.6
7.0
6.3
7.8
6.5
—
—
—
—
0.20
0.001 0.001 0.001
0.008 0.001 0.001 0.001
0.007 0.55
0.001 0.32
—
—
—
—
0.17
0.07
0.05
0.04
0.76
0.54
0.35
0.35
0.31
0.22
0.13
0.25
—
—
—
—
1.79
0.08
0.15
0.18
0.001 0.14
0.001 0.14
0.98
0.74
0.82
0.71
0.02
0.001 0.001 0.001
0.14
0.001 0.001 0.001
0.002 0.001 0.001 0.001
—
—
—
0.001 0.002
0.007 0.001
0.36
0.17
—
—
—
0.001
0.001
0.001
—
—
—
a
Data were obtained on three tanks per treatment, with 150 individuals in each tank for the 50% feeding level, or 100 individuals for the
feeding levels of 70, 100, and 140%. Tank records are means of measurements performed on each tank. Individual records are means and
CV of measurements performed on each individual within each tank. When the line × feeding method interaction was significant (P < 0.05),
separate variance analyses testing the effect of the line within a feeding level were performed. *Indicates when S was different (P < 0.05)
from C at the corresponding feeding level.
b
Probability for the effects of line (L), feeding level (F), the line × feeding level (L × F) interaction and the tank effect.
c
Analysis of covariance.
d
100(Final BW ¹⁄₃ − Initial BW ¹⁄₃)/d.
e
Whole body initial composition: (for S) 76.8% water, 13.9% protein, 5.3% fat, 5.8% energy; (for C) 77.4% water, 13.6% protein, 5.1% fat,
5.5 kJ/g energy.
f
ANOVA.
e
100(BW/fork length3).
h
Kruskal-Wallis test – separate analysis.
2). The values were the lowest at 100% UDR, and the
highest at 50 and 140% UDR for both lines.
Individual final BW were different (P < 0.001) between lines when they were fed 100 and 140% UDR
only (Tables 1 and 2). Regardless of the feeding level,
S was always longer (P < 0.001) than C. Except for fish
fed 140% UDR, the K-factor was always lower (P <
0.001) in S than in C. The apparent discrepancy at
140% UDR could be attributed to the large difference
(P < 0.001) in feed intake and final BW between the C
and S.
Variability of the final BW and L was higher in fish
fed 30% than in those fed 10% UDR (Table 1). For
higher feeding levels (Table 2), the variability decreased
with the level of feed intake. In addition, at the highest
feeding levels (100 and 140% UDR), BW and L variability were lower in S than in C. Because maximal intake
was higher for S than for C, the CV in weight were
plotted against the observed level of feed restriction
(cumulative intake at a given feeding level/cumulative
intake measured at 140% UDR; Figure 2). This shows
that the variability in the final weight of S is always
lower than that of C at the same restriction level.
Feed Deprivation Phase
During the feed deprivation phase (Table 3), the relative loss of BW was not significantly affected (P > 0.07)
by either the line or the previous feeding level (P >
0.18). The relative daily loss of BW was only slightly
higher in S than in C for a previous feeding level of
140% UDR. Body composition was not affected by the
previous feeding level or by the line. Relative losses of
nutrients (protein loss, −0.97 gⴢkg−1ⴢd−1 for C and S, P
> 0.57; lipid loss, −0.58 and −0.67 gⴢkg−1ⴢd−1 for C and
S, respectively, P > 0.22; energy loss, −53 and −55 kJⴢkg−
1 −1
ⴢd for C and S, respectively, P > 0.38) also were unaffected by feeding level and line. The K factor values
were much lower at the end of the feed deprivation
phase than at the end of the feeding phase. The line and
the previous feeding level affected body conformation in
a way similar to that at the end of the first experimental
period. This indicates that there was no interaction
between the effect of 35 d of feed deprivation and the
line or the previous feeding level. Briefly, S fish remained slightly leaner than C fish when previously
fed restricted rations, whereas they were fatter than C
2870
Mambrini et al.
Figure 2. Coefficients of variation in individual weight
of each group of fish related to feed intake (expressed
relatively to the maximal intake) in selected (S) and control (C) brown trout lines at the end of the growth phase.
The curves correspond to the fit to polynomial secondorder models.
Figure 1. Whole body protein (a) and energy (b) increments related to intake in selected (S) and control (C)
brown trout lines at the end of the growth phase. Data
at zero intake were obtained at the end of the feed deprivation phase. Maintenance requirements correspond to
the x-intercept of the slope.
mal voluntary intake. Growth performance was high
in all groups and significantly better for S than for C.
This was largely explained by the higher cumulative
intake (P < 0.001) of S compared with C. The lower
the previous feeding level, the higher the growth rate
during the refeeding period. There was no clear effect
of the line on G:F. At the end of the refeeding period,
S was still longer than C (P < 0.001). The K-factor was
much higher compared with the values observed at the
end of the feed deprivation phase, but no clear differences appeared between the lines. With the exception
of the fish previously fed 50% UDR, the CV in BW and
L were lower for S than for C.
Discussion
Control Fish
when previously fed 140% UDR. The tendency for a
higher heterogeneity in BW (P < 0.001) and L (P <
0.007) in C than in S remained after this period of
feed deprivation.
Refeeding Phase
During the refeeding phase (Table 4), fish were fed
140% UDR, and uneaten feed was recovered in each
tank. This indicates that all the groups were fed in
excess; thus, the cumulative feed intake reflects maxi-
In C fish, growth rate and G:F increased with the
feeding level until 100% UDR. Cumulative feed intake
was similar between 100 and 140% UDR because of an
increase in uneaten feed. This suggests that, as in other
fish species studied (de la Higuera, 2002), maximal feed
intake of brown trout is tightly regulated. A slight decrease in G:F was seen at the 140% UDR (i.e., in overfed
fish). This negative effect of a high feeding level on
G:F has already been observed in other fish species
(Storebakken and Austreng, 1987a,b; Ballestrazzi et
al., 1998; Valente et al., 2001). It has been assumed to
2871
Selection for growth of brown trout
Table 3. Mean body weight and weight loss of brown trout selected for growth (S) or controls (C) deprived of food
for 35 d and previously fed at 50, 70, 100, and 140% of the usual intake for 116 da
Feeding level and line
50%
Measurements
70%
100%
P-valueb
140%
C
S
C
S
C
S
C
S
SEM
L
F
L×F
Tank
16.9
−0.49
17.0
−0.49
27.6
−0.51
27.4
−0.48
37.9
−0.49
48.3
−0.48
39.4
−0.52
61.4
−0.56
—
0.009
—
0.07
—
0.12
—
0.39
—
—
72.3
15.9
8.5
7.3
73.6
15.7
7.5
6.7
71.2
16.2
9.3
7.6
72.5
16.1
8.3
7.2
70.5
16.4
9.7
7.9
70.1
16.1
10.3
8.1
70.9
16.4
9.5
7.8
70.5
15.6
10.7
8.1
0.16
0.65
0.03
0.06
0.48
0.45
0.27
0.37
0.44
0.64
0.45
0.63
0.84
0.90
0.73
0.73
—
—
—
—
13.8
11.2
0.94
14.0
11.4*
0.90*
23.1
13.0
1.03
23.2
13.2*
0.98*
32.5
14.4
1.06
42.1*
15.8*
1.05
33.6
14.5
1.08
53.0*
16.9*
1.09*
0.02
0.14
0.002
0.001
0.001
0.001
0.001
0.001
0.001
0.001
0.001
0.001
0.001
0.001
0.002
32.5
9.9
1.9
33.7
1.9
1.9
25.8
8.2
7.7
25.2
8.0
7.8
22.6
7.2
5.9
19.0
6.6
6.2
22.5
7.5
5.8
18.1
6.0
6.2
—
—
—
0.39
0.40
0.64
0.001
0.001
0.002
—
—
—
—
—
—
c
Tank records
Initial mean BW, gd
Daily growth coefficient, %/de
Final body composition, % or kJ/g BW
Water
Protein
Fat
Energy
Individual records nested by tanksf
Final BW, g
Final length, cm
K-factor, g/cm3g
CV, %h
Final BW
Final length
K-factor
a
Data were obtained on three tanks per treatment, with 150 individuals in each tank for the 50% feeding level, or 100 individuals for the
feeding levels of 70, 100, and 140%. Tank records are means of measurements performed on each tank. Individual records are means and
CV of measurements performed on each individual within each tank. When the line × feeding method interaction was significant (P < 0.05),
separate variance analyses testing the effect of the line within a feeding level were performed. *Indicates when S was different (P < 0.05)
from C at the corresponding feeding level.
b
Probability for the effects of line (L), previous feeding level (F), the line × previous feeding level (L × F) interaction and the tank effect.
c
Analysis of covariance.
d
Initial BW differed from the final BW indicated at the end of the growth phase (see Table 1) because 10 fish were slaughtered between
the two experimental phases.
e
100(Final BW¹⁄₃ − Initial BW¹⁄₃)/d.
f
ANOVA.
g
100(BW/fork length3).
h
Kruskal-Wallis test – separate analysis.
result from increased energy expenditure, namely a
higher heat increment of feeding to sustain protein and
fat deposition (Storebakken et al., 1991). Another explanation is that it is difficult to assess with accuracy the
amount of uneaten feed in water running out of the
tank when feed is distributed in excess. As discussed
further, the amounts of uneaten feed were much lower
for selected fish fed 100 and 140% UDR than for the
controls, whereas the G:F were similar among the two
lines, which is in favor of a negative effect of high feeding levels on feed efficiency. Indeed, dietary protein and
energy use efficiencies were negatively affected when
fish were fed in excess.
A negative relationship between feed intake and BW
heterogeneity was observed. It is well known that in
salmonids, when the access to food is qualitatively or
quantitatively restricted at the group level, the interindividual variability of feed intake, and thus variability
in individual growth performance, is increased (McCarthy et al., 1992; Jobling and Koskela, 1996; Gélineau
et al., 1998; Kristiansen, 1999). The observed weight
heterogeneity in feed-restricted groups may be explained by individual intake heterogeneity. Other results obtained in the current study support this assumption. At the end of the feed deprivation phase, BW
and length variabilities were similar to those observed
at the end of the feeding period. When the fish were
refed, the heterogeneity of the groups previously restricted was largely reduced, whereas it remained similar for the groups previously unrestricted. One might
argue that the variability in individual intake is linked,
at least in salmonids, to increased competition for food
when fish are restricted (Davis and Olla, 1987). However, the share of food may depend on the strength of
social interactions. They seem to play a larger role in
brown trout than in rainbow trout (Kristiansen, 1999).
Increased feed intake led to a carcass fat concentration increase, and this is also reflected by a K-factor
increase. At the end of the feed-deprivation phase, the
K-factor was largely diminished and fish have lost
about the same relative BW (approximately 16%), regardless of the previous feeding level. In addition, there
was no effect of the previous feeding level on the extent
of lipid and energy losses. This is in line with the fact
that in our experiment, there was no link between the
carcass fat concentration at the beginning of the feeddeprivation phase and the extent of weight loss. These
results are surprising because during fasting, brown
trout are known to mobilize in priority their perivisceral
fat (Navarro et al., 1992). One might suggest that in the
present experiment, the length of the feed deprivation
phase was too short to observe any effect of fattening
2872
Mambrini et al.
Table 4. Growth and feed intake of brown trout selected for growth (S) or controls (C) fed at 140% of the usual intake
(as-fed basis) for 47 d after an experimental phase of food deprivation of 35 d (previous feeding level refers to the
situation before feed deprivation)a
Feeding level and line
50%
Measurements
70%
100%
P-valueb
140%
C
S
C
S
C
S
C
S
SEM
L
F
L×F
Tank
13.8
21.7
2.40
1.51
14.0
29.5*
2.85*
1.43
23.1
27.2
2.16
1.37
23.2
39.5*
2.82*
1.37
32.5
28.6
2.11
1.53
42.1
44.2*
2.60
1.51
33.6
27.4
2.00
1.52
53.0
47.5*
2.52
1.54
—
0.17
0.014
0.013
—
0.001
0.001
0.04
—
0.001
0.04
0.005
—
0.003
0.017
0.28
—
—
—
—
46.0
14.8
1.38
55.5*
15.8*
1.37*
60.4
16.4
1.35
76.6*
17.7*
1.38*
76.0
17.7
1.35
108.4*
20.0*
1.34
75.1
17.7
1.32
125.8*
21.0*
1.35*
0.03
0.41
0.002
0.001
0.001
0.001
0.001
0.001
0.03
0.001
0.001
0.001
0.001
0.001
0.82
26.4
8.9
6.0
28.1
9.8
6.4
27.2
8.9
5.3
22.9
7.5
6.0
25.9
8.6
6.3
20.9
7.2
5.7
24.7
8.3
5.9
19.2
6.5
6.1
—
—
—
0.015
0.08
0.25
0.05
0.02
0.51
—
—
—
—
—
—
c
Tank records
Initial mean BW, gd
Cumulative intake, g/fish
Daily growth coefficient, %/de
G:F
Individual records nested by tanksf
Final BW, g
Final length, cm
K-factor, g/cm3g
CV, %h
Final BW
Final length
K-factor
a
Data were obtained on three tanks per treatment, with 150 individuals in each tank for the 50% feeding level, or 100 individuals for the
feeding levels of 70, 100, and 140%. Tank records are means of measurements performed on each tank. Individual records are means and
CV of measurements performed on each individual within each tank. When the line × feeding method interaction was significant (P < 0.05),
separate variance analyses testing the effect of the line within a feeding level were performed. *Indicates when S was different (P < 0.05)
from C at the corresponding feeding level.
b
Probability for the effects of line (L), previous feeding level (F), the line × previous feeding level (L × F) interaction and the tank effect.
c
Analysis of covariance.
d
Initial BW differed from the final BW indicated at the end of the growth phase (see Table 1) because 10 fish were slaughtered between
the two experimental phases.
e
100(Final BW¹⁄₃ − Initial BW¹⁄₃)/d.
f
ANOVA.
g
100(BW/fork length3).
h
Kruskal-Wallis test – separate analysis.
on weight loss. Indeed, brown trout are still able to
mobilize the perivisceral fat after 50 d of feed deprivation (Navarro et al., 1992). These losses were in the
same range as values reported for other salmonid species, as were the maintenance requirements (Cho and
Bureau, 1995). Concerning the requirements for
growth, we were unable to compare our results with
published data because they were obtained on fish of
different sizes and reared at different temperatures (Arzel et al., 1992). They seem lower than the recommendations given for rainbow trout (NRC, 1993). There are
slight differences in the way brown trout and rainbow
trout utilize nutrients for growth (Arzel et al., 1992,
1994). Our results indicate that brown trout may be
more efficient than rainbow trout.
Growth rates exhibited during the refeeding phase
were much higher than those observed during the feeding phase. Compensatory growth capacities have been
demonstrated in several salmonid species (Dobson and
Holmes, 1984; Jobling et al., 1994; Jobling and Koskela,
1996), and our results indicate that it is also the case for
brown trout. In our experiment, as is usually observed,
compensatory growth was achieved by a burst in both
feed intake and G:F. In addition, the extent to which
feed intake relative to BW increased was correlated
to the strength of the feed restriction before the feed
deprivation phase. Voluntary feed intake of fish may
be affected by their past nutritional history, in particu-
lar the previous level of feed restriction (Boujard et al.,
2000) and the duration of feed deprivation (Hayward
et al., 1997). Taken together, these data show that in
brown trout, feed intake is a key factor for the compensatory growth phenomenon.
Selected Fish
This study confirms that the main factor affected by
our selection procedure for increased body length is
feed intake. In selected lines, feed refusals were only
observed at 140% UDR. At this feeding level, fish exhibited their best growth potential and reached a BW 55%
higher than that of the control at the end of the feeding
phase. This is in line with the values of the genetic gain
expected at this age considering the average genetic
gain obtained with this selection procedure (Sanchez
et al., 2001; Vandeputte et al., 2002). This is also in
agreement with the results obtained with other selection protocols for growth in salmonids (Gjedrem, 1998).
At 100% UDR, selected fish were slightly restricted (no
uneaten feed recovered, lower growth rates than when
fed 140% UDR), showing that the level of voluntary
feed intake was higher than in the control line. During
the refeeding phase, the feed intake capacities of S were
always higher than for C, and as observed in controls,
intake relative to BW was all the higher as the previous
restriction level was high, even though the feed intake
2873
Selection for growth of brown trout
capacities were always higher for S than for C. This
high intake capacity appears to be a feature of the S
line, remaining evident even after a period of feed restriction.
When feed intake was restricted, the S lines exhibited
the same growth rate as the C line. The G:F remained
comparable among the lines regardless of the feeding
level, confirming on a larger scale our previous results
(Sanchez et al., 2001). The main response correlated to
this selection procedure was an increased intake capacity. In higher vertebrates, most of the heritable variation in BW is associated with differences in feed intake
(Rauw et al., 1998). This also seems to be the case
in fish taking into account estimates of heritabilities,
which are not significantly different from zero for feed
efficiency (Kinghorn, 1983) and range between 0.15 and
0.40 for feed intake (Kinghorn, 1983; Gjoen et al., 1991).
Salmon selected with a familial procedure when compared with wild fish exhibit increases in both daily feed
consumption and feed efficiency (Smith et al., 1988;
Thodesen et al., 1999), but in these studies, the results
merge the gains due to both genetics and domestication.
The lack of variability in feed efficiency is in line
with the constancy of the body composition and nutrient
retention among the lines. In addition, both lines seem
to have the same needs for protein and energy gains,
at least for the range of weights tested in the current
study. Moreover, the maintenance needs were similar
among the lines. This is also illustrated by the fact that
selected and control fish lost the same relative weight
and the same amount of protein, lipid, and energy during the feed-deprivation phase. Our selection procedure
affected neither the nutrient requirements nor nutrient
use efficiency. The increase in feed intake of the selected
line in the current study is therefore disconnected from
modifications of maintenance and growth requirements. In broilers, selection for increased BW was
shown to damage the hypothalamic satiety mechanism,
not to diminish hunger drive, and to lead to overeating
(Burkhart et al., 1983; Denbow et al., 1986). Broilers
are able to consume more than required until reaching
the highest limit of their gastrointestinal capacity (Barbato et al., 1984) because of a higher rate of food gastrointestinal transit or because of increased digestion processes (Dunnington and Siegel, 1995). Such parameters
could be specifically studied in our lines; however, their
influence on feed intake is still largely unknown because in fish, the mechanisms governing feed intake
regulation are marginally understood. Lines involved in
this study might be good models for such investigations.
Intake in fish is measured on groups and, as exemplified in the current study, the food sharing and individual intake are affected by feeding level, whereas social
interactions may influence individual intake. It may be
that selection has affected the overall behavior. The
interindividual variability of BW was lower in S than
in C, regardless of the restriction level. This points to
less competition for food in S fish. However, the decrease in the phenotypic variability of the selected trait
has been observed in poultry after several generations
of selection (Kawahara et al., 1974; Metodiev and
Drbohlav, 1998). In our case, it does not seem to be
linked to the genetic variability, which, based on enzymatic and microsatellite markers, was not affected by
the selection procedure (B. Chevassus, unpublished results). Rather, the lower variability of individual BW
of the selected line seems to be the result of decreased
sensitivity to environmental factors. This may have positive implications in production because fish have to be
re-sorted regularly due to their high interindividual
BW heterogeneity.
The rearing management during selection may partly
explain why selection for growth mainly increased intake capacities. During the selection process, fish were
always fed in excess. Mice selected for growth under ad
libitum feeding conditions exhibited increased intake
capacities, whereas under feed restriction conditions,
selection for growth decreased their appetite (Hetzel
and Nicholas, 1978). The other important factor that
may have influenced the results is that fish were selected based on length, and not weight. These two variables are closely linked, as indicated by the relative
constancy of the K-factor in salmonids, at least when
they are in the fast growing phase. Heavy but fatty
fish might not have been retained during the selection
procedure, which may explain the relative constancy of
body composition. It has been demonstrated in pigs that
a greater growth rate is achieved when animals are
selected for lean growth efficiency (Chen et al., 2003;
Fabian et al., 2003). Our results show that it may be
difficult to improve metabolic efficiency with a selection
for growth in fish. In higher vertebrates, the strategy
was to select on residual feed intake (individual deviations from the regression of feed intake on BW gains).
This selection procedure implies both that individual
intake is accurately measured and that growth models
have been correctly established (Kennedy et al., 1993).
These conditions are still not fulfilled in fish. Individual
feed intake cannot be measured in a continuous manner
(Jobling et al., 2001). Growth models are only available
for a few species. They are not commonly used and must
integrate the constant improvements of the rearing performance observed in recently domesticated species.
Then, a specific selection procedure is needed and indirect criteria, such as lean deposition, may be considered.
Implications
The main correlated response to our selection procedure for increased body length was an increase in feed
intake capacity. Requirements for growth and maintenance were identical between selected and control lines.
Thus, selected fish will only exhibit their full growth
potential when their appetite is satisfied. Fish selected
for growth require a higher feeding allowance. Whereas
this will not reduce feed costs, it will shorten rearing
duration, with no increase in body fat content, at least
2874
Mambrini et al.
within the range of weight tested. Lower body weight
variability in the selected line may decrease the frequency with which fish need to be re-sorted into similarsize groups. The effect of the selection procedure on
feed intake capacity may result from a more even share
of food due to behavioral differences at the group level
and to a higher appetite at the individual level. These
lines constitute good models for gaining better insight
into the physiological basis, as well as some genetic
determinants, of feed intake regulation in fish.
Literature Cited
AOAC. 1990. Official Methods of Analysis. 15th ed. Assoc. Offic. Anal.
Chem., Arlington, VA.
Arzel, J., R. Metailler, C. Huelvan, A. Fauré, and J. Guillaume. 1992.
The specific nutritional requirements of brown trout (Salmo
trutta). Icel. Agric. Sci. 6:77–92.
Arzel, J., F. X. Martinez Lopez, R. Metailler, G. Stéphan, M. Viau,
G. Gandemer, and J. Guillaume. 1994. Effect of dietary lipid on
growth performance and body composition of brown trout (Salmo
trutta) reared in seawater. Aquaculture 123:361–375.
Ballestrazzi, R., D. Lanari, and E. D’Agaro. 1998. Performance, nutrient retention efficiency, total ammonia and reactive phosphorus
excretion of growing European sea-bass (Dicentrarchus labrax
L.) as affected by diet processing and feeding level. Aquaculture
161:55–65.
Barbato, G. F., P. B. Siegel, J. A. Cherry, and I. Nir. 1984. Selection
for body weight at eight weeks of age. 17. Overfeeding. Poult.
Sci. 63:11–18.
Boujard, T., C. Burel, F. Médale, G. Haylor, and A. Moisan. 2000.
Effect of past nutritional history and fasting on feed intake and
growth in rainbow trout Oncorhynchus mykiss. Aquat. Living
Resour. 13:129–137.
Boujard, T., L. Labbé, and B. Aupérin. 2002. Feeding behaviour,
energy expenditure and growth of rainbow trout in relation to
stocking density and food accessibility. Aquac. Res. 33:1233–
1242.
Burkhart, L. A., S. A. Cherry, H. P. Van Krey, and P. B. Siegel.
1983. Genetic selection for growth alters hypothalamic satiety
mechanisms in chickens. Behav. Genet. 13:295-300.
Chen, P., T. J. Baas, J. C. M. Dekkers, K. J. Koehler, and J. W. Mabry.
2003. Evaluation of strategies for selection for lean growth rate
in pigs. J. Anim. Sci. 81:1150–1157.
Chevassus, B., F. Krieg, R. Guyomard, J. M. Blanc, and E. Quillet.
1992. The genetics of the brown trout: twenty years of French
research. Icel. Agric. Sci. 6:109–124.
Cho, C. Y. 1992. Feeding systems for rainbow trout and other salmonids with reference to current estimates of energy and protein
requirements. Aquaculture 100:107–123.
Cho, C. Y., and D. P. Bureau. 1995. Determination of the energy
requirements of fish with particular reference to salmonids. J.
Appl. Ichthyol. 11:141–163.
Davis, M. V., and B. L. Olla. 1987. Aggression and variation in growth
of chum salmon (Oncorhynchus keta) juveniles in sea water:
effects of limited rations. Can. J. Fish. Aquat. Sci. 44:192–197.
De la Higuera, M. 2002. Effects of nutritional factors and feed characteristics on feed intake. Pages 250–268 in Food Intake in Fish.
D. Houlihan, T. Boujard, and M. Jobling, ed. Blackwell Science,
Oxford, U.K.
Denbow, D. M., H. P. van Krey, and P. B. Siegel. 1986. Selection for
growth alters the feeding response to injections of biogenic
amins. Pharmacol. Biochem. Behav. 24:39–42.
Dobson, S. H., and R. M. Holmes. 1984. Compensatory growth in
rainbow trout, Salmo gairdneri Richardson. J. Fish Biol.
25:649–656.
Dunnington, E. A., and P. B. Siegel. 1995. Enzyme activity and organ
development in newly hatched chicks selected for high or low
eight-week body weight. Poult. Sci. 74:761–770.
Fabian, J., L. I. Chiba, D. L. Kuhlers, L. T. Frobish, K. Nadarajah,
and W. H. McElhenney. 2003. Growth performance, dry matter
and nitrogen digestibilities, serum profile, and carcass and meat
quality of pigs with distinct genotypes. J. Anim. Sci. 81:1142–
1149.
Gélineau, A., G. Corraze, and T. Boujard. 1998. Effects of restricted
ration, time-restricted access and reward level on voluntary food
intake, growth and growth heterogeneity of rainbow trout (Oncorhynchus mykiss) fed on demand with self-feeders. Aquaculture 167:247–258.
Gjedrem, T. 1998. Developments in fish breeding and genetics. Acta
Agric. Scand. Suppl. 28:19–26.
Gjoen, H. M., T. Storebakken, E. Austreng, and T. Refstie. 1991.
Genotypes and nutrient utilization. Pages 19–26 in Fish Nutrition in Practice. S. J. Kaushik and P. Luquet, ed. INRA,
Paris, France.
Hayward, R. S., D. B. Noltie, and N. Wang. 1997. Use of compensatory
growth to double hybrid sunfish growth rates. Trans. Am. Fish.
Soc. 126:316–322.
Hetzel, D. J. S., and F. W. Nicholas. 1978. Growth and body composition of mice selected for growth rate under ad libitum or restricted feeding. Proc. Aust. Soc. Anim. Prod. 12:194.
Jobling, M., and J. Koskela. 1996. Interindividual variations in feeding and growth in rainbow trout during restricted feeding and
in a subsequent period of compensatory growth. J. Fish Biol.
49:658–667.
Jobling, M., O. H. Mely, J. dos Santos, and B. Christiansen. 1994.
The compensatory growth response of the Atlantic cod: effects
of nutritional history. Aquac. Int. 2:75–90.
Jobling, M., D. Covès, B. Damsgard, H. R. Kristiansen, J. Koskela,
T. E. Peturdottir, S. Kadri, and O. Gudmundsson. 2001. Techniques for measuring feed intake. Pages 49–87 in Food Intake
in Fish. D. Houlihan, T. Boujard, and M. Jobling, ed. Blackwell
Science, Oxford, U.K.
Kawahara, T., A. Mita, M. Saito, and N. Sugimoto. 1974. Genetic
changes in Japanese wild quails under domestic conditions.
Annu. Rep. Natl. Inst. Genet. 24:72.
Kennedy, B. W., J. H. J. van der Werf, and T. H. E. Meuwissen.
1993. Genetic and statistical properties of residual feed intake.
J. Anim. Sci. 71:3239–3250.
Kinghorn, B. 1983. Genetic variation in food conversion efficiency
and growth in rainbow trout. Aquaculture 32:141–155.
Kristiansen, H. R. 1999. Discrete and multiple meal approaches applied in a radiographic study of feeding hierarchy formation in
juvenile salmonids. Aquac. Res. 30:519–527.
McCarthy, I. D., C. G. Carter, and D. F. Houlihan. 1992. The effect
of hierarchy on individual variability in daily feeding rainbow
trout, Oncorhynchus mykiss (Walbom). J. Fish Biol. 41:257–263.
Mercer, L. P. 1982. The quantitative nutrient-response relationship.
J. Nutr. 122:560–566.
Metodiev, S., and V. Drbohlav. 1998. Short-term divergent selection
on five-week body weight in Japanese quail. Selection differentials. Bulg. J. Agric. Sci. 4:491–497.
Navarro, I., J. Gutiérrez, and J. Planas. 1992. Changes in plasma
glucagon, insulin and tissue metabolites associated with prolonged fasting in brown trout (Salmo trutta fario) during two
different seasons of the year. Comp Biochem Physiol Part A
Physiol. 102:401–407.
NRC. 1993. Nutrient Requirements in Fish. Natl. Acad. Press, Washington, DC.
Rauw, W. M., E. Kanis, E. N. Noordhuizen-Stassen, and F. J. Grommers. 1998. Undesirable side effects of selection for high production efficiency in farm animals: A review. Livest. Prod. Sci.
56:15–33.
Sanchez, M. P., B. Chevassus, L. Labbé, E. Quillet, and M. Mambrini.
2001. Selection for growth of brown trout (Salmo trutta) affects
feed intake but not feed efficiency. Aquat. Living Resour.
14:41–48.
Shearer, K. D. 1994. Factors affecting the proximate composition
of cultured fishes with emphasis on Salmonids. Aquaculture
119:63–88.
Smith, R. R., H. L. Kincaid, J. M. Regenstein, and G. L. Rumsey.
1988. Growth, carcass composition, and taste of rainbow trout
Selection for growth of brown trout
of different strains fed diets containing primarily plant or animal
protein. Aquaculture 70:309–321.
Sokal, R. R., and C. A. Braumann. 1980. Significance tests for coefficients of variation and variability profiles. Syst. Zool. 29:50–62.
Storebakken, T., and E. Austreng. 1987a. Ration level for salmonids
I. Growth, survival, body composition and feed conversion in
Atlantic Salmon fry and fingerlings. Aquaculture 60:189–206.
Storebakken, T., and E. Austreng. 1987b. Ration level for salmonids:
II. Growth, feed intake, protein digestibility, body composition
and feed conversion in rainbow trout weighing 0.5–1 kg. Aquaculture 60:207–221.
Storebakken, T., S. S. O. Hung, C. C. Calvert, and E. M. Plisetskaya.
1991. Nutrient partitioning in rainbow trout at different feeding
rates. Aquaculture 96:191–203.
2875
Thodesen, J., B. Grisdale-Helland, S. J. Helland, and B. Gjerde. 1999.
Feed intake, growth and feed utilization of offspring from wild
and selected Atlantic salmon (Salmo salar). Aquaculture
180:237–246.
Valente, L. M. P., B. Fauconneau, E. F. S. Gomes, and T. Boujard.
2001. Feed intake and growth of fast and slow growing strains
of rainbow trout (Oncorhynchus mykiss) fed by automatic feeders
or by self-feeders. Aquaculture 195:121–131.
Vandeputte, M., E. Quillet, F. Krieg, M. G. Hollebecq, A. Fauré, L.
Labbé, J. P. Hiseux, and B. Chevassus. 2002. The “PROSPER”
methodology on brown trout (Salmo trutta fario): Four generations of improved individual selection on growth rate. CD-ROM,
communication 0604, Proc. 7th World Cong. Genet. Appl. Livest.
Prod., Montpellier, France.