Applied Animal Behaviour Science 171 (2015) 226–232

Contents lists available at ScienceDirect

Applied Animal Behaviour Science

journal homepage: www.elsevier.com/locate/applanim

Using self-selection to evaluate the acceptance of a new diet

formulation by farmed fish
Hanna Carlberg a , Ken Cheng b , Torbjörn Lundh c , Eva Brännäs a,∗
a
Department of Wildlife, Fish and Environmental Studies, Swedish University of Agricultural Science, 901 83 Umeå, Sweden
b
Department of Food Science, Swedish University of Agricultural Science, Uppsala BioCenter, PO Box 7051, 750 07 Uppsala, Sweden
c
Department of Department of Animal Nutrition and Management, Swedish University of Agricultural Science, PO Box 7024, 750 07 Uppsala, Sweden

a r t i c l e i n f o a b s t r a c t

Article history: The evaluation of new diet composition is commonly achieved by performing time-consuming growth
Received 11 February 2015 trials, which may negatively impact the welfare of a large number of fish if the feed is not accepted.
Received in revised form 14 July 2015 Instead, the fish’s behavioural responses to a new diet composition can be used as a first step in the
Accepted 10 August 2015
evaluation of new diet composition. The taste acceptance of a new diet by Arctic charr (Salvelinus alpinus)
Available online 22 August 2015
was evaluated over 16 days based on the self-selection of a test diet and a control diet. The test diet
contained ingredients from the nutrient-enriched Baltic Sea, and it is hoped that this diet can contribute
Keywords:
nutrients to the nutrient-poor waters in northern Sweden in which Arctic charr are farmed. The main
Self-selection
Arctic charr
ingredients in the test feed were blue mussel meat (Mytilus edulis), meat from sprat (Sprattus sprattus)
Salvelinus alpinus and herring (Clupea harengus), baker’s yeast (Saccharomyces cerevisiae), fish oil and regionally produced
Betaine rapeseed oil. Individual fish (n = 15) were allowed to choose between abundances of both the test diet
Aquaculture and a fishmeal based control diet that mimicked a standard feed for Arctic charr. After only a few days,
Mussel meal the fish demonstrated a significant preference for the Baltic Sea test feed. These results may depend
Yeast on betaine, which was found at concentrations that were four times higher in the test feed than in the
control feed. Betaine is a known attractant for many fish species and is abundant in marine animals, such
as mussels. Thus, the test diet can be evaluated further without additional taste stimulants before the
new feed formula can be used commercially to ensure fish welfare, the product quality and the economic
feasibility of the new formula.
© 2015 Elsevier B.V. All rights reserved.

1. Introduction (Naylor et al., 2000; FAO, 2014). However, there are drawbacks
to the use of plant materials in fish diets because they commonly
Gustation and olfaction are the most important senses for feed- contain anti-nutritional factors that adversely affect fish, particu-
ing and thus for fish survival (Hara, 2007). Indeed, taste buds are larly carnivorous species like salmonids (Cain and Garling, 1995; Li
more numerous in many fish species than in any other animals and Robinson, 1997; Hughes and Soares, 1998). Additionally, many
(Kasumyan and Doving, 2003). Preferences for tastes in fish are plant materials, such as soybeans, which are commonly used in
highly species-specific and even vary between individuals of the feed for farmed fish, are important protein sources for the growing
same species (Kasumyan and Doving, 2003). The diets of farmed fish human population (Brown, 2012).
must not only meet species-specific nutritional requirements but A future goal is therefore to develop fish feed that is created from
also be appetizing (Raubenheimer et al., 2012). A reduced accep- ingredients that are either unsuitable or unattractive for human
tance of the feed results in poor fish welfare and increased feed consumption and are not limited resources or key species in ecosys-
waste that result in negative environmental effects and increased tems but nevertheless produce healthy and attractive products for
production costs (Jobling, 2001). The fishmeal and fish oil that the table market (Kiessling, 2009). However, new feed composi-
are used to feed carnivorous farmed fish are limited resources; tions, especially those composed of ingredients that are not found in
thus, the use of plant ingredients as substitutes is increasing the natural diets of carnivorous fish, must be thoroughly evaluated
before use.
Feeds need to be designed to attract the specific fish species,
∗ Corresponding author. Tel.: +46 907868592. and the incorporation of feeding stimulants may be necessary
E-mail addresses: [email protected] (H. Carlberg), [email protected] to encourage feeding (Jobling, 2001). Amino acid attractant mix-
(K. Cheng), [email protected] (T. Lundh), [email protected] (E. Brännäs). tures and betaine are examples of substances that improve feed

http://dx.doi.org/10.1016/j.applanim.2015.08.016
0168-1591/© 2015 Elsevier B.V. All rights reserved.
 H. Carlberg et al. / Applied Animal Behaviour Science 171 (2015) 226–232 227

acceptability and growth on plant-based diets (Tandler et al., 1982; source in fish diets. The two fatty fish, i.e., the sprat and herring,
Papatryphon and Soares, 2000). Attractants are of particular impor- are very abundant in the Baltic Sea presumably due to the inten-
tance when designing diet compositions for marine fish larvae (de sive cod (Cadus morhua) fisheries that have resulted in ecosystem
la Higuera, 2001). The evaluations of new diet compositions are imbalances and stocks that need to be decreased by selective fishing
most commonly performed via time-consuming growth trials that (Persson et al., 2014). Because the protein content of baker’s yeast
may negatively impact the welfare of large numbers of fish if the varies between 40 and 65% (Halasz and Lasztity, 1991; Nasseri et al.,
feed is not accepted. 2011), it replicates rapidly and can grow on multiple substrates
The fish’s behavioural approach to a new diet composition could why there is substantial interest in the use of yeast as an alterna-
instead be used as a first step in the evaluation of a new diet tive protein source for fishmeal (Rumsey et al., 1990; Omar et al.,
formula. The fish’s acceptance can be evaluated in short-term stud- 2012; Overland et al., 2013; Vidakovic et al., 2015).
ies based on the self-selection of a test diet and a reference diet This study represents the first step of a thorough evaluation of
by the fish. A fish’s preference for a feed may be determined by the test feed that contains these three different protein fractions
several methods of self-selection. Self-feeders that are activated and began with assessments of the fishes’ behavioural approaches
when the fish actuate a rod are commonly used to measure feed- using the same methods employed by Pettersson et al. (2009).
ing behaviour related to several aspects, such as diel or annual Mussel meat and its extract have been found to be efficient attrac-
variations in appetite and diet preferences (Jobling et al., 2001). tants in the diets of, for example, common sole (Mackie, 1982).
Geurden et al. (2005) used self-feeders to demonstrate that rain- However, very few trials of mussel meal in fish diets have been
bow trout (Oncorhynchus mykiss) can discriminate between diets performed recently. Marine invertebrates, such as mussels, con-
that contain fish oil and those that contain plant oils. Sea bass tain high quantities of betaine (Meyers, 1987), which is a known
(Dicentrarchus labrax) are able to discriminate and prefer diets that attractive substance to some fish species (Mackie et al., 1980;
contain fishmeal over those that contain soybeans and the enzyme Tusche et al., 2013). However, bioassay studies have not identified
phytase, which improves the nutritional efficiency of soy (Sugiura betaine as a stimulant for two salmonid species, i.e., the rain-
et al., 2001). Sea bass also exhibit preferences for fishmeal diets over bow trout (Jones, 1989) and the Chinook salmon (Oncorhynchus
those that contain soybean meal as assessed with the same method tsawytscha) (Hughes, 1993), and its effects on Arctic charr are
(Fortes-Silva et al., 2011). One problem with the use of self-feeders uncertain. Feeds in which fishmeal composes the protein fraction
with groups of fish involves the social interactions between the are preferred to plant-based substitutes by sea bass (Fortes-Silva
individuals. When passive integrated transponders (PIT-tags) were et al., 2011) and are also attractive to Arctic charr (Pettersson et al.,
used to study individual variations in trigger activation in groups of 2009). To our knowledge, there are no reported data regarding
15 Arctic charr, all individuals activated the trigger in the first days the taste responses of fish to yeast. Yeast is often deficient in
after the trial started, but thereafter, only one individual continued certain essential amino acids (Overland et al., 2013), and yeast
to exploit the activations (Brännäs and Alanärä, 1993). Self-feeding was therefore used in the present case as a partial replace-
by trigger actuation has been tested with individual Arctic charr, ment for the fishmeal. The potential adverse or indifferent taste
but these tests were unsuccessful (Brännäs unpublished results). responses to the yeast are likely to be masked by the other
Thus, another method of assaying diet preference was tested on ingredients.
individual fish; this method involved measurements of the given
and rejected feed in aquaria that were divided by one or several
walls that separates two or more diets but allowed the fish to swim
between the compartments (Pettersson et al., 2009). In such set- 2. Material and methods
ups, rainbow trout exhibit a self-selection preference for feeds with
100% fish oil compared with feeds containing 25 to 75% rapeseed 2.1. Experimental feeds
oil replacements and fishmeal as the protein fraction (Pettersson
et al., 2009). The fish were given a choice between two feeds with the
A conceptual model has been created for the small scale Swedish same amounts of protein and energy, i.e., the test feed and a
farming of Arctic charr (Eriksson et al., 2010). This model focuses fishmeal-based control feed with a composition that mirrored that
on Arctic charr that are farmed in northern Sweden in net pens of a commercial Arctic charr feed. Cooked and de-shelled mussels
within extremely oligotrophic water bodies that are affected by were used (Royal Frysk Muscheln GmbH Emmelsbüll-Hornsbül,
hydropower regulations. The fish feed used in this system contains Germany). The fishmeal from the Baltic Sea was produced by
regionally produced ingredients that are unsuitable or unattrac- TripleNine (Esbjerg, Denmark), and the fishmeal in the con-
tive for direct human consumption and are mainly sourced from trol diet originated from the Atlantic Ocean. The Baker’s yeast
the eutrophicated southern Baltic Sea (Ronnberg and Bonsdorff, was cultured on molasses, ammonia, phosphorus, magnesium
2004). Sourcing the feed ingredients from the Baltic Sea transports and vitamins and then dried on a fluidized bed (Jästbolaget® ,
nutrients to the oligotrophic waters in which the Arctic charr are Stockholm, Sweden). The oil fractions of the two feeds both
farmed and is thus beneficial for both ecosystems (Eriksson et al., contained regular commercial fish oil and regionally produced
2010). rapeseed oil.
The protein fraction of the feed that was used in this study The Finnish Game and Fisheries Research Institute manufac-
(i.e., the test feed) was sourced primarily from the Baltic Sea and tured the experimental feeds at the Laukaa Aquaculture station
was composed of one-third of each of the following: Baltic Sea in Finland. Pellets were extruded using a twin-screw extruder
blue mussels (Mytilus edulis), fishmeal from two fatty fish (i.e., (3 mm die, BC-45 model, Clextral, Creusot Loire, France), and lipids
sprat (Sprattus sprattus) and herring (Clupea harengus)), and baker’s were then added with a vacuum coater (Pegasus PG-10VC, Din-
yeast (Saccharomyces cerevisiae). Blue mussels are excellent nutri- nissen, Sevenum, Netherlands). The pellets were 3 mm long (See
ent assimilators, but due to the low salinity of the Baltic Sea, they Tables 1 and 2 for further details about the feeds). The amounts of
never grow large enough to be attractive for the table market betaine (glycine-betaine) in the feeds were analysed because this
(Schütz, 1964). Blue mussels have a favourable amino acid profile compound is a feeding stimulant for salmonids and may be con-
(Langeland et al., 2014) and a fat composition that is suitable for tained in marine invertebrates, such as mussels (Yamashita et al.,
salmonid diets (Berge and Austreng, 1989), but very few studies 2006). The higher betaine content of the test feed was four times
have been published regarding the use of blue mussels as a protein greater than that of the control feed (Table 2)
228 H. Carlberg et al. / Applied Animal Behaviour Science 171 (2015) 226–232

Table 1
Recipes of the test (Baltic Sea-sourced) and control feeds used in the self-selection
test.

Feed composition (kg kg−1 ) Test Control

Regular fishmeal 0.332

Baltic Sea fishmeal 0.216
Fish oil, “regular” 0.071 0.073
Mussel meal 0.212
Yeast 0.253
Rapeseed oil 0.047 0.049
Wheat gluten 0.050
Wheat meal 0.131 0.259
Soy protein concentrate 0.267
Mineral/vitamin mix 0.015 0.015

Total 1.000 1.000

All diets additionally contained 40 mg astaxanthin/kg.

Fig. 1. Aquarium set-up with two compartments for the feed preference trial.
Table 2
Chemical compositions (g kg−1 in DM4 ), energy contents (MJ kg−1 in DM) and betaine
(␮mol g−1 ) contents of the control and test diets for the Arctic charr. compartments that were separated by a grey plastic wall with a
hole in the middle (10 cm in diameter) that allowed the fish to
Test diet Control
swim freely between the two compartments without mixing the
1
Crude protein 474 467
2
two diets. Each compartment was equipped with a water inlet, an
Fat 194 187
3
Ash 78 68
oxygen stone and an artificial sea grass shelter (Fig. 1). The aquaria
4
Gross energy 23 22.9 were supplied with flow-through water at 1 L min−1 at a tempera-
5
Betaine 22.7 5.4 ture of 10 ◦ C (±0.5 ◦ C). The pH (7.0–7.5) and oxygen level (82–85%)
The betaine contents of the feeds were determined in a Bruker 600 MHz 1H NMR were measured at the beginning, when the diets were switched and
spectrometer with a zgesgp pulse sequence (Bruker Spectrospin Ltd., BioSpin, Karl- at the end of the trial. Fish were maintained on a 12:12 h light/dark
sruhe, Germany). cycle. The fish were acclimatized to their aquarium for 10 days,
1
Total nitrogen was determined according to the Kjeldahl method using a 2020 during which time they were fed a commercial feed (Skretting
digester and a 2400 Kjeltec Analyser unit (FOSS Analytical A/S Hilleröd, Denmark).
The crude protein content was calculated as N*6.25 (Nordic Committee on Feed
Optime 1P 2.5 mm) by battery-driven feeders (Fish mate T14) that
Analysis, 1976). were located above each of the two compartments. The two exper-
2
The crude lipid was determined using a hydrolyzation and extraction system imental feeds were tested for durability in the water before the
(1047 Hydrolysing Unit and a Soxtec System HT 1043 Extraction Unit, FOSS Analyt- experiment to ensure that the uneaten pellets would not dissolve
ical A/S, Hilleröd, Denmark).
when left for more than 4 h in the aquaria.
3
Ash was determined after drying at 550 ◦ C for a minimum of 3 h until the ash
was white. The ash was then cooled in a desiccator prior to weighing. The initial weight of each fish was used to determine its feed
4
The gross energies (MJ kg−1 ) of the feeds were determined using an isoperobol ration. The feed rations were set to 1.5% of the body weight (BW)
bomb calorimeter (Parr 6300, Parr Instrument Company, Moline, IL, USA). per day for each of the two feeds, which resulted in a total feed
ration of 3% of the BW per day. This total ration ensured that feed
2.2. Fish and rearing was given in excess. On the sixth day of the trial, the feed rations
were adjusted for seven fish that exhibited greater appetites; these
Arctic charr from the “Arctic superior” selected strain (Nilsson fish ate an average of 1.82% of their BWs per feeding compared with
et al., 2010) were transported from the Aquaculture Centre North the other fish, which ate an average of 1.3% of their BW per feeding.
in Kälarne, Sweden to the Umeå Marine Science Centre in Norrbyn During the trial, the feed was delivered twice daily (8 and 11 a.m.),
in October of 2012. The fish were 0.5 years of age and had an aver- and each feeding lasted approximately 1 h and 40 min. The test feed
age weight of 43.6 ± S.D. 10.1 g. The fish were reared in 1 m3 tanks was delivered into one compartment of the aquaria, and the control
supplied with flow through brackish water (4%) at 5 L min−1 and feed was delivered into the other compartment. Uneaten pellets
fed commercial feeds (Skretting Nutra MP T 1.9 mm and Skretting were carefully collected from each tank daily with a small net and
Optime 1P 2.5 mm) according to a standardize feeding model for counted. This procedure has been used in several previous experi-
Arctic charr until the initiation of the experiment. The light/dark ments (i.e., Pettersson et al., 2009) and has only been shown to elicit
cycle was maintained at L12:D12, and the temperatures followed very short stress responses from the fish. When the pellets were
the natural variations until three weeks before the trials at which collected, the fish typically moved to the opposite compartment.
time the water temperature was slowly increased from the natu- After nine days, the fish exhibited a clear preference for one of the
ral 1 ◦ C to 10 ◦ C. The slope of the temperature increase has been feeds, and the positions of the feeds were reversed in all aquaria to
used several times before, and has not been found to induce any control for left/right biases. The trial continued for another seven
noticeable stress responses in the fish. The study was approved days resulting in a total trial time of 16 days.
by the ethical committee of Northern Sweden (Umeå) no A13-
13. 2.4. Calculations and statistics

2.3. Experimental design The specific growth rate (SGR) was calculated as 100 ×
(ln W1 − ln W0 ) × t−1 , where W0 is the initial weight (g), W1 is the
Sixteen Arctic charr were collected at random and anaesthetized final weight (g), and t is the number of days between W0 and W1 .
(MS-222, 40 mg L−1 ), and their initial weights and fork lengths were The preference results are presented as feed intake % BW day−1 .
recorded in April of 2013. The initial weights ranged from 72.2 to A linear estimation of the increase in weight between the initial
149.2 g (average 118.6 g, S.D. ± 18.1 g). Individual fish were placed and final weights was used to calculate the feed intake as % BW
in aquaria (94 cm long × 40 cm deep × 40 cm high) containing a day−1 . One fish failed to eat during the trials and was excluded
total volume of 70 L of water. The aquaria were divided into two from the analysis, which resulted in a final sample size of n = 15.
 H. Carlberg et al. / Applied Animal Behaviour Science 171 (2015) 226–232 229

0,8

Ratio of eaten test feed/control feed bw-1

0,6

0,4

0,2

0,0

-0,2

-0,4
0 2 4 6 8 10 12 14 16 18

Days

Fig. 2. Daily ratio in feed intake of the test feed from the Baltic Sea region and the fishmeal-based control feed (% BW day−1 ). Positive values indicate greater intakes of the
Baltic Sea test feed, and negative values indicate greater intakes of the control feed. The error bars indicate the standard error. The black dashed line indicates the time at
which the feeds were reversed in the aquaria.

1,4
Ratio of eaten test feed/control feed bw-1

1,2

1,0

0,8

0,6

0,4

0,2

0,0

-0,2
2 4 6 8 10 12 14

Individual fish

Fig. 3. Total ratio (16 days) in feed intake of the test feed from the Baltic Sea region and the fishmeal-based control feed (% BW day−1 ) by 15 individual Arctic charr. The
horizontal dotted line indicates an equal ratio.

The data in the preference trials were analysed with a repeated of the trial, the fish began to exhibit a preference for the test feed.
measures ANOVA in which the feed intakes % BW day−1 on days On the tenth day, when the positions of the feeds were reversed,
1 through 16 were taken as the dependent variables, and the feed there were no differences in the consumptions of the two feeds.
type was the factor. The data were visually checked for normal- From day 11 until the end of the trial, a preference for the test feed
ity by plotting the data in box-plots. Linear regression was used once again became clear (Fig. 2). Examination of the total ratio of
to test the relationships of the growth rate (over 26 days including the eaten pellets revealed that four of the individuals exhibited an
the acclimatization) with the test pellets (g), control pellets and the indifferent response for any of the two feeds, and the remaining
total pellets eaten during the 16-day test period per g of body mass. individuals preferred to consume a higher proportion of the test
The analyses were conducted using Statistica 12. feed (Fig. 3).
All 15 fish ate with good appetites and exhibited a mean growth
3. Results rate of 1.84% per day (S.D. ± 0.56), including the 10-day period
of acclimatization. There were significant positive relationships
The Arctic charr exhibited a significant preference for the test between the growth rate and the total grams of pellets eaten
feed (repeated measures ANOVA, F1,14 = 2.73, p = 0.046). There was per gram body mass over the 16 days of the self-selection trial
no effect of time (F15, 420 = 1.53, p = 0.09). As soon as the second day (t14 = 10.5, p < 0.0001) (Fig. 4) and between the growth rate and the
230 H. Carlberg et al. / Applied Animal Behaviour Science 171 (2015) 226–232

0,7

0,6

Total amount eaten (g) bw (g)-1

0,5

0,4

0,3

0,2

0,1
0,8 1,0 1,2 1,4 1,6 1,8 2,0 2,2 2,4

Gw

Fig. 4. The relationship between the growth rate (% day−1 ) and the total amount of pellets eaten per gram of body mass. The individual Arctic charr (n = 15) were given a
choice of pellets with standard ingredients (control) and Baltic Sea-sourced pellets (test feed) during a 16-day trial. The separate linear relationships of the growth rates with
the two diets are given in the text.

total number of test feed pellets (t14 = 4.2, p < 0.001) and control study. If required, this information could be acquired via several
feed pellets (t14 = 6.9, p < 0.0001) eaten. bioassay methods in which taste preferences are analysed by using
a gel as a carrier for test substances (reviewed in Kasumyan and
4. Discussion Doving, 2003).
Free amino acids are highly efficient attractive substances for
The Arctic charr exhibited a preference for the test feed that was fish, which is not surprising because their prey contain consider-
sourced from the Baltic Sea over the control feed. There are several able amounts of amino acids (reviewed in Carr et al., 1996). The
mechanisms involved in the diet choices of farmed fish, includ- two diets in the present study contained the same amino acid pro-
ing size, colour, taste and nutritional properties (Raubenheimer files (Carlgren et al. submitted), although at somewhat different
et al., 2012). Both feeds were the same size and colour; there- amounts; therefore, the diets should have been equally attractive
fore, these visual cues can be excluded. Some animal studies have in terms of these substances. A possible explanation for the pref-
suggested that the nutritional properties of a diet are generally erence for the Baltic Sea test feed could be the inclusion of the
more important than the taste in terms of feed preferences (Forbes mussel meal, which is known to attract fish (Mackie et al., 1980;
and Kyriazakis, 1995; Simpson and Raubenheimer, 1999). How- Tusche et al., 2013). The test feed contained four times more betaine
ever, some species of fish can achieve species-specific optimal than the control feed, and betaine is a known attractant for several
diets via post-ingestive signals and not by taste from single-source fish species (Kasumyan and Doving, 2003; Yamashita et al., 2006)
macronutrients in gelatine capsules if the capsules are swallowed that is typically found in marine invertebrates such as mussels
and not chewed (Sanchez-Vazquez et al., 1999; Ruohonen and (Meyers, 1987). Although betaine was not found to be a stimulant
Grove, 2001). for rainbow trout (Jones, 1989) or Chinook salmon (Hughes, 1993),
However, in the present study, we argue that taste was the most the highly species-specific taste preferences of fish (Kasumyan and
important cue in terms of the preference for the test feed that was Doving, 2003) indicate that betaine cannot be excluded as an attrac-
sourced from the Baltic Sea region. This argument seems proba- tive substance for Arctic charr. Additionally, due to manufacturing
ble because the fish rapidly demonstrated clear preferences for the difficulties, the Baltic Sea test feed had slightly more fat on the sur-
test feed at the start of the trial and immediately after the feed pos- face that was not completely absorbed by the pellet. This fat may
itions were reversed at the midpoint of the trial. All fish, especially also have had an attractant effect on the taste of the test feed.
salmonids, have highly developed senses of taste (Kasumyan and A possible repellent effect of the baker’s yeast is unlikely, and if
Doving, 2003). We argue that the method of collecting the rejected present, this effect was masked by the other ingredients. Vidakovic
pellets only imposed a temporary stressor on the fish. If the fish et al. (2015) conducted a three-month growth study with Arctic
were frightened too often or subjected to new threats, the feed- charr using the same batches of raw materials that were used in
ing activity might have decreased (Pottinger et al., 1994). All of the the present study. These authors demonstrated that a 40% protein
individuals except one exhibited growth rates between 1.4 and 2.3% substitution with intact baker’s yeast had no effect on the growth
per day, which are within the range that has been reported earlier rate compared with a fishmeal-based control feed or a feed with
for undisturbed individually held Arctic charr from the same ori- 40% mussel meat substitution. In contrast, in a recent study by
gin, of the same size and in the same temperature conditions as Overland et al. (2013), growth reductions were observed when
in the present study (Eriksson et al., 2010). The strong relationship Atlantic salmon (Salmo salar) were given a feed with a 40% pro-
between the amount of consumed pellets and the growth rate indi- tein substitution with baker’s yeast. Whole yeast has been shown
cates that the rejected pellets could be accurately collected, which to be slightly less digestible when used in feed for rainbow trout
validates the present method of assaying self-selection. (Rumsey et al., 1991), salmon (Overland et al., 2013) and lake trout
Identifying the exact nature of the attractive compounds or mix (Salvelinus namaycush) (Rumsey et al., 1990). Rumsey et al. (1991)
of compounds in the test diet is beyond the scope of the present suggested that the cell walls of the yeast should be disrupted to
 H. Carlberg et al. / Applied Animal Behaviour Science 171 (2015) 226–232 231

improve digestibility. However, in a study by Vidakovic et al. (2015), number 071 and the Swedish Research Council for Environment,
substitution with 40% extracted yeast (instead of intact yeast) led Agricultural Sciences and Spatial Planning (Formas), Project “Using
to a significant reduction in the growth of Arctic charr. Although a geometrical framework for tesing if Arctic charr possesses a nutri-
salmonids are similar in many ways, their feed utilization and nutri- tional wisdom”. We would like to thank Odd Lindahl for the pro-
tional requirements are not necessarily the same, which is why duction of the de-shelled mussel meal, Jästbbolaget for contributing
species and strains should be individually studied (Tocher et al., the yeast and Jouni Vielma for the production of the extruded feed.
2001; Azevedo et al., 2004; Krogdahl et al., 2004). Additionally,
the growth medium and drying process of the yeast may affect its
digestibility and possible repellent properties. References
A long-term shift in the preference between the test and control Azevedo, P.A., Leeson, S., Cho, C.Y., Bureau, D.P., 2004. Growth, nitrogen and energy
feeds due to nutritional imbalances seems unlikely. The two diets utilization of juveniles from four salmonid species: diet, species and size
contained equal amounts of energy and protein, but the Baltic Sea effects. Aquaculture 234, 393–414.
Berge, G.M., Austreng, E., 1989. Blue mussel in feed for rainbow trout. Aquaculture
test feed contained greater amounts of marine proteins and fish-
81, 79–90.
meal than the control diet. With respect to the amino acid profiles, Brown, L.R., 2012. Full Planet, Empty Plates. W.W. Norton & Company, New York,
fishmeal is better balanced for carnivorous fish, such as the Arc- NY.
tic charr, compared with the protein sources that originated from Brännäs, E., Alanärä, A., 1993. Monitoring the feeding activity of individual fish
with a demand feeding system. J. Fish Biol. 42, 209–215.
plants (e.g., soy) and were partly used in the control feed. Cain, K.D., Garling, D.L., 1995. Pretreatment of soybean meal with phytase for
There were pronounced individual differences in the prefer- salmonid diets to reduce phosphorus concentrations in hatchery effluents.
ences for the test feed. This finding is in agreement with earlier Prog. Fish Cult. 57, 114–119.
Carr, W.E.S., Netherton, J.C., Gleeson, R.A., Derby, C.D., 1996. Stimulants of feeding
findings in which individual variabilities in taste preferences within behavior in fish: analyses of tissues of diverse marine organisms. Biol. Bull.
the same species have been found in 1 out of 25 rainbow trout 190, 149–160.
(Jones, 1989) and in 3 of 11 individuals of the same species (Mearns de la Higuera, M., 2001. Effects of nutritional factors and feed characteristics on
feed intake. In: Houlihan, D., Boujard, T., Jobling, M. (Eds.), Food Intake in Fish,
et al., 1987). Additionally, common carp (Cyprinus carpio) exhibit vol. 827. Blackwell Science Cost Action, Oxford, pp. 250–268.
individual differences in their acceptance of the single amino acids Eriksson, L.-O., Alanärä, A., Nilsson, J., Brännäs, E., 2010. The Arctic charr story:
l-cysteine (97.3%) and l-glutamic acid (41.8%) (Kasumyan, 2000). development of subarctic freshwater fish farming in Sweden. Hydrobiologia
650, 265–274.
The present conceptual model can most likely support the rather FAO, 2014. The State of the World Fisheries and Aquaculture. Fisheries and
small Arctic charr farming industry in Sweden. Currently, the ingre- Aquaculture Department, Food and Agriculture organization of the United
dients in the test feed are too expensive to be used commercially, Nations, Rome, pp. 223.
Forbes, J.M., Kyriazakis, I., 1995. Food preferences in farm animals—why don’t they
particularly the mussel meat because the mussels had to be de-
always choose wisely? Proc. Nutr. Soc. 54, 429–440.
shelled. However, as mussels remove nutrients, the eco-system Fortes-Silva, R., Sanchez-Vazquez, F.J., Martinez, F.J., 2011. Effects of pretreating a
costs for each unit of mussel produced will need to be discussed plant-based diet with phytase on diet selection and nutrient utilization in
and addressed within the next few years. Yeast has a high repro- European sea bass. Aquaculture 319, 417–422.
Geurden, I., Cuvier, A., Gondouin, E., Olsen, R.E., Ruohonen, K., Kaushik, S., Boujard,
ductive rate (Kiessling, 2009), and the high level of nucleic acid T., 2005. Rainbow trout can discriminate between feeds with different oil
(NA) in baker’s yeast limits its use as a major protein source for sources. Physiol. Behav. 85, 107–114.
farmed mammals and birds. In contrast, fish are able to metabolize Halasz, A., Lasztity, R., 1991. Use of Yeast Biomass in Food Production. CRC Press,
Inc, Boca Raton, FL, USA.
high levels of NA because they exhibit greater levels of liver uricase Hara, T.J., 2007. Gustation. Fish Physiol. 25, 45–96.
activity than terrestrial animals (Kinsella et al., 1985). The dense Hughes, K.P., Soares Jr., J.H., 1998. Efficacy of phytase on phosphorus utilization in
populations of sprat and herring need to be decreased by commer- practical diets fed to striped bass Morone saxatilis. Aquacult. Nutr. 4, 133–140.
Hughes, S.G., 1993. Single feeding response of Chinook salmon fry to potential feed
cial fisheries to restore the balance of the ecosystem of the Baltic sea intake modifiers. Prog. Fish Cult. 55, 40–42.
(Persson et al., 2014). High levels of PCBs and dioxins restrict the Isosaari, P., Hallikainen, A., Kiviranta, H., Vuorinen, P.J., Parmanne, R., Koistinen, J.,
human consumption of these fish (Svensson et al., 1995; Isosaari Vartiainen, T., 2006. Polychlorinated dibenzo-p-dioxins, dibenzofurans,
biphenyls, naphthalenes and polybrominated diphenyl ethers in the edible fish
et al., 2006). However, the removal of dioxins and other harm-
caught from the Baltic Sea and lakes in Finland. Environ. Pollut. 141, 213–225.
ful anthropogenic substances via an inexpensive industrial method Jobling, M., 2001. Feed composition and analysis. In: Houlihan, D., Boujard, T.,
results in a detoxified fishmeal product that is suitable for inclusion Jobling, M. (Eds.), Food Intake in Fish, vol. 827. Blackwell Science Cost Action,
Oxford, pp. 1–24.
in fish feed. In a 10-month growth study that followed the present
Jobling, M., Coves, D., Damsgard, B., Kristiansen, H.R., Koskela, J., Petursdottir, T.E.,
behavioural study, the groups that were fed the test feed exhib- Kadri, S., Gudmundsson, O., 2001. Techniques for measuring feed intake. In:
ited a 11.5% reduction in growth compared to the control group. Houlihan, D., Boujard, T., Jobling, M. (Eds.), Food Intake in Fish, vol. 827.
The lower digestibility of the test feed, likely due to the yeast, is the Blackwell Science Cost Action, Oxford, pp. 49–87.
Jones, K.A., 1989. The palatability of amino acids and related compounds to
most likely reason for this slowed growth rate. Alterations to the rainbow trout, Salmo gairdneri R. J. Fish Biol. 34, 149–160.
test feed recipe, particularly the yeast preparation, will most likely Kasumyan, A.O., 2000. Individual variability of taste preference and of behavioural
improve its digestibility. response to taste stimuli in carp (Cyprinus carpio). J. Ichthyol. 40, 661–669.
Kasumyan, A.O., Doving, K.B., 2003. Taste preferences in fishes. Fish Fish. 4,
In conclusion, a taste-based rejection of a new diet formula- 289–347.
tion that included ingredients not found in the natural diets of the Kiessling, A., 2009. Feed-the key to sustainable fish farming. In: Ackefors, H.,
landlocked Arctic charr, such as mussels and yeast, was excluded Cullberg, M., Wramner, P., Sundberg, P. (Eds.), KSLA- Fish. Trade and
Development KSLA, Stockholm, pp. 303–322.
by beginning the evaluation with a behavioural approach. Self- Kinsella, J.E., German, B., Shetty, J., 1985. Uricase from fish liver—isolation and
selection by the fish is a quick method that does not require many some properties. Comp. Biochem. Physiol. B: Biochem. Mol. Biol. 82, 621–624.
individuals and less time than fish consumption growth studies, but Krogdahl, A., Sundby, A., Olli, J.J., 2004. Atlantic salmon (Salmo salar) and rainbow
trout (Oncorhynchus mykiss) digest and metabolize nutrients differently.
other factors, e.g., differences in digestibility, need to be separated
Effects of water salinity and dietary starch level. Aquaculture 229, 335–360.
from taste preferences. Langeland, M., Vidakovic, A., Vielma, J., Lindberg, J.E., Kiessling, A., Lundh, T., 2014.
Digestibility of microbial and mussel meal for Arctic charr (Salvelinus alpinus)
and Eurasian perch (Perca fluviatilis). Aquacult. nutr., http://dx.doi.org/10.
1111/anu.12268 (In press).
Acknowledgements Li, M.H., Robinson, E.H., 1997. Microbial phytase can replace inorganic phosphorus
supplements in channel catfish Ictalurus punctatus diets. J. World Aquacult.
The study was funded by Innovative Practices and Technologies Soc. 28, 402–406.
Mackie, A.M., 1982. Identification of the gustatory feeding stimulants. In: Hara, T.J.
for Developing Sustainable Aquaculture in the Baltic Sea region, (Ed.), Chemoreception in Fishes. Elsevier Scientific Publishing Co, Amsterdam,
AQUABEST of the Baltic Sea Region Programme 2007–2013, project pp. 275–291.
232 H. Carlberg et al. / Applied Animal Behaviour Science 171 (2015) 226–232

Mackie, A.M., Adron, J.W., Grant, P.T., 1980. Chemical nature of feeding stimulants Rumsey, G.L., Hughes, S.G., Kinsella, J.L., 1990. Use of dietary yeast Saccharomyces
for the juvenile Dover sole, Solea solea (L.). J. Fish Biol. 16, 701–708. cerevisiae nitrogen by lake trout. J. World Aquacult. Soc. 21, 205–209.
Mearns, K.J., Ellingsen, O.F., DÖving, K.B., Helmer, S., 1987. Feeding behaviour in Rumsey, G.L., Hughes, S.G., Smith, R.R., Kinsella, J.E., Shetty, K.J., 1991. Digestibility
adult rainbow trout and atlantic salmon parr, elicited by chemical fractions and and energy values of intact, disrupted and extracts from brewers dies yeast fed
mixtures of compounds identi¢ted in shrimp extract. Aquaculture 64, 47–63. to rainbow trout (Oncorhynchus mykiss). Anim. Feed Sci.Technol. 33,
Meyers, S.P., 1987. Aquaculture feeds and chemo attractants. Infofish Marketing 185–193.
Digest 87 (1), 35–37. Ruohonen, K., Grove, D.J., 2001. In situ study of gastric emptying using
Nasseri, A.T., Rasoul-Amini, S., Morowvat, M.H., Ghasemi, Y., 2011. Single cell X-radiography and macrocapsules. Aquacult. Res. 32, 491–493.
protein: production and process. Am. J. Food Technol. 6, 103–116. Sanchez-Vazquez, F.J., Yamamoto, T., Akiyama, T., Madrid, J.A., Tabata, M., 1999.
Naylor, R.L., Goldburg, R.J., Primavera, J.H., Kautsky, N., Beveridge, M.C.M., Clay, J., Macronutrient self-selection through demand-feeders in rainbow trout.
Folke, C., Lubchenco, J., Mooney, H., Troell, M., 2000. Effect of aquaculture on Physiol. Behav. 66, 45–51.
world fish supplies. Nature 405, 1017–1024. Schütz, L., 1964. Die tierische Besiedlung der Hartböden in der
Nilsson, J., Brännäs, E., Eriksson, L.O., 2010. The Swedish Arctic charr breeding Schwentinemündung. Kieler Meeresforsch 20, 198–217.
programme. Hydrobiologia 650, 275–282. Simpson, S.J., Raubenheimer, D., 1999. Assuaging nutritional complexity: a
Nordic Committee on Feed Analysis, 1976. Determination in Feeds and Faeces geometrical approach. Proc. Nutr. Soc. 58, 779–789.
According to Kjeldahl, No 6, third ed. NMKL, Oslo, Norway. Sugiura, S.H., Gabaudan, J., Dong, F.M., Hardy, R.W., 2001. Dietary microbial
Omar, S.S., Merrifield, D.L., Kuehlwein, H., Williams, P.E.V., Davies, S.J., 2012. phytase supplementation and the utilization of phosphorus, trace minerals
Biofuel derived yeast protein concentrate (YPC) as a novel feed ingredient in and protein by rainbow trout Oncorhynchus mykiss (Walbaum) fed soybean
carp diets. Aquaculture 330, 54–62. meal-based diets. Aquacult. Res. 32, 583–592.
Overland, M., Karlsson, A., Mydland, L.T., Romarheim, O.H., Skrede, A., 2013. Svensson, B.G., Nilsson, A., Jonsson, E., Schutz, A., Akesson, B., Hagmar, L., 1995. Fish
Evaluation of Candida utilis, Kluyveromyces marxianus and Saccharomyces consumption and exposure to persistent organochlorine compounds, mercury,
cerevisiae yeasts as protein sources in diets for Atlantic salmon (Salmo salar). selenuim and methyamines among Swedish fishermen Scand. J. Work Environ.
Aquaculture 402, 1–7. Health 21, 96–105.
Papatryphon, E., Soares, J.H., 2000. The effect of dietary feeding stimulants on Tandler, A., Berg, B.A., Kissil, G.W., Mackie, A.M., 1982. Effect on food attractants on
growth performance of striped bass, Morone saxatilis, fed a plant feedstuff appetite and growth rate of gilthead bream, Sparus aurata L. J. Fish Biol. 20,
based diet. Aquaculture 185, 329–338. 673–681.
Persson, L., Van Leeuwen, A., De Roos, A.M., 2014. The ecological foundation for Tocher, D.R., Bell, J.G., MacGlaughlin, P., McGhee, F., Dick, J.R., 2001. Hepatocyte
ecosystem-based management of fisheries: mechanistic linkages between the fatty acid desaturation and polyunsaturated fatty acid composition of liver in
individual-, population-, and community-level dynamics. ICES J. Mar. Sci. 71, salmonids: effects of dietary vegetable oil. Comp. Biochem. Physiol. B:
2268–2280. Biochem. Mol. Biol. 130, 257–270.
Pettersson, A., Johnsson, L., Brannas, E., Pickova, J., 2009. Effects of rapeseed oil Tusche, K., Nagel, F., Arning, S., Wuertz, S., Susenbeth, A., Schulz, C., 2013. Effect of
replacement in fish feed on lipid composition and self-selection by rainbow different dietary levels of potato protein concentrate supplemented with feed
trout (Oncorhynchus mykiss). Aquacult. Nutr. 15, 577–586. attractants on growth performance of rainbow trout (Oncorhynchus mykiss).
Pottinger, T.G., Knudsen, F.R., Wilson, J., 1994. Stress induced changes in the Anim. Feed Sci. Technol. 183, 202–209.
affinity and abundance of cytosolic corticol binding sites in the liver of rainbow Vidakovic, A., Langeland, M., Sundell, K., Sundh, H., Olstorpe, M., Vielma, J.,
trout, Onchorhynchus mykiss (Walbaum), are not accompanied by changes in Kiessling, A., Lundh, T., 2015. Evaluation of growth performance and intestinal
measurable nuclear binding. Fish Physiol. Biochem. 12, 499–511. barrier function in Arctic Charr (Salvelinus alpinus) fed yeast (Saccharomyces
Raubenheimer, D., Simpson, S., Sanchez-Vazquez, J., Huntingford, F., Kadri, S., cerevisiae), fungi (Rhizopus oryzae) and blue mussel (Mytilus edulis). Aquacult.
Jobling, M., 2012. Nutrition and diet choice. In: Huntingford, F., Jobling, M., Nutr., http://dx.doi.org/10.1111/anu12344.
Kadri, S. (Eds.), Aquaculture and Behaviour. Blackwell Publishing Ltd, Oxford, Yamashita, S., Yamada, T., Hara, T.J., 2006. Gustatory responses to feeding- and
UK, pp. 150–176. non-feeding-stimulant chemicals, with an emphasis on amino acids, in
Ronnberg, C., Bonsdorff, E., 2004. Baltic Sea eutrophication: area-specific ecological rainbow trout. J. Fish Biol. 68, 783–800.
consequences. Hydrobiologia 514, 227–241.