Chemosphere 68 (2007) 15421547

www.elsevier.com/locate/chemosphere

Determination of selenium by GFAAS in slurries of sh feces

to estimate the bioavailability of this micronutrient in feed used
in pisciculture
Fabio A. Silva a, Renato C.F. Neves a, Luis G. Quintero-Pinto a, Cilene C.F. Padilha b,
Sonia M.A. Jorge c, Margarida M. Barros a, Luiz E. Pezzato a, Pedro M. Padilha c,*
a
Faculdade de Medicina Veterinaria e Zootecnia, Programa de Pos-Graduacao em Zootecnia, Departamento de Melhoramento e Nutricao Animal/UNESP,
Caixa Postal 560, 18618-000 Botucatu, SP, Brazil
b
Instituto de Biociencias, Departamento de Fsica e Biofsica/UNESP Caixa Postal 510, 18618-000 Botucatu, SP, Brazil
c
Instituto de Biociencias, Departamento de Qumica e Bioqumica/UNESP Caixa Postal 510, 18618-000 Botucatu, SP, Brazil

Received 20 June 2006; received in revised form 10 January 2007; accepted 4 March 2007
Available online 19 April 2007

Abstract

This paper presents a simple, fast and sensitive method to determine selenium in samples of feces and of sh feed by graphite furnace
atomic absorption spectrometry (GFAAS) through the direct introduction of slurries of the samples into the spectrometers graphite
tube. The limits of detection (LOD) and quantication (LOQ) calculated for 20 readings of the blank of the standard slurries (0.50%
m/v of feces or feed devoid of selenium) were 0.31 lg l1 and 1.03 lg l1, respectively, for the standard feces slurries and 0.35 lg l1
and 1.16 lg l1, respectively, for the standard feed slurries. The proposed method was applied in studies of bioavailability of selenium
in dierent sh feeds and the results proved consistent with that obtained from samples mineralized by acid digestion using the micro-
wave oven.
2007 Elsevier Ltd. All rights reserved.

Keywords: Bioavailability of selenium in sh feed; Selenium in sh feces; Determination of selenium by GFAAS; Toxic limit of selenium in pisciculture

1. Introduction toxic substances. The activity reduction of this enzyme is

The animals tissues possess chemical elements in vari-
able proportions and amounts that are essentials to the
metabolism, contributing to the larger or smaller produc-
tivity of the animal (Mcdowell, 1992). From 2% to 5% of
the animals body are constituted by minerals that vary
according to species, race and own individual (Solomons,
1992).
Selenium is classied as an essential micronutrient for
animals, including sh (Smith, 1987). It is an integral com-
ponent of the enzyme glutathione peroxidase that acts in
the cellular cytosol turning hydrogen peroxide into non-
*
Corresponding author. Tel./fax: +55 14 3811 6255.
E-mail address: padilha@ibb.unesp.br (P.M. Padilha).
the main biochemical indicator of selenium deciency
(Poston et al., 1976; Combs et al., 1984; Combs and
Combs, 1986; Poppe et al., 1986; Lovell, 1998; Oliveira,
1998).
In sh nutrition, the animal may ingest selenium from
both water and diet. High levels of selenium in water are
toxic to sh, resulting in the decrease of the their growth
and even death. Usually the concentration in water is less
than 0.1 lg l1; hence the uptake through gills is very eec-
tive and the mineral is stored in several tissues, except liver,
in its inorganic form. Selenium as selenite is taken up e-
ciently through the gills. For dietary selenium, stored in the
organic form, the margin between the nutritive require-
ments and the toxic limit is very narrow, making dicult
the addition of an exact amount of this mineral in the

0045-6535/$ - see front matter 2007 Elsevier Ltd. All rights reserved.
doi:10.1016/j.chemosphere.2007.03.003
 F.A. Silva et al. / Chemosphere 68 (2007) 15421547
animal feed (Poston et al., 1976). Several factors associated
to the amount of selenium present in foods can aect the
bioavailability of the element in the diet, inuencing the
use of the ingested amount in the phases of digestion/
absorption or metabolism/elimination (Watanabe et al.,
1997). The comparison among diets with fundamentally
dierent compositions showed signicant inuence in the
use of supplementation with sodium selenite (Bell and
Cowey, 1989).
So, the development of new methodologies that allow
the reliable quantication of the metallic nutrients, such
as selenium, present at low concentrations in feeds,
becomes fundamental in sh nutrition studies. In this
context, the determination of metallic analytes in slur-
ries by graphite furnace atomic absorption spectrometry
(GFAAS) is a promising technique (Bendicho and
Loos-Vollebregt, 1991; Millerihli, 1993). It provides several
advantages, such as high sensitivity, low detection limits,
the use of small sample volumes, determination of a wide
variety of trace elements, etc. Considering also that the
atomizer can act as a chemical reactor, the possibility of
making solid sampling presents some advantages over the
conventional digestion procedures (Belchier and Forster,
1970). Besides eliminating the stage of total previous
decomposition of sample, it diminishes the sample prepara-
tion time, decreases the analyte losing for excessive manip-
ulation or retention on insoluble products, reduces the
possibility of sample contamination and, above all,
minimizes the action of dangerous acids on the analyst
(Bendicho and Loos-Vollebregt, 1991; Millerihli, 1993;
Liang et al., 1996).
Thus, this paper describes the development of a method
to determine selenium in slurries of sh feed and feces sam-
ples by GFAAS that eliminates the samples mineralization
step and allows for an estimate of the biavailability of this
micronutrient in samples of feed used in sh nutrition.
2. Experimental
2.1. Reagents, standard solutions and samples
Superpure deionized water (18.2 M X cm1) obtained
with an Elga Ionic system (PURELAB Option, USA),
suprapure nitric acids (Merck), hydrogen peroxide (Merck)
and Triton X-100 (Merck) were used throughout this work.
The solution containing tungsten, employed to coat the
inside of the graphite tube and used as a permanent mod-
ier, was prepared by diluting a stock solution containing
1000 mg l1 of sodium tungstate (Merck) with ultrapure
water. The Pd(II) solution, also employed as a chemical
modier, was made in the same way, utilizing palladium
nitrate (Merck) instead.
Stock solutions of the analytes were prepared from
reagents of spectroscopic purity (Johnson & Matthey,
Royston, Hertfordshire, UK). The remaining solutions,
including the concentrated acid solutions used for mineral-
izing the samples, were analytical grade. All the bottles for
1543
storing samples and standard solutions, the glassware and
the containers of the atomic absorption spectrometers
autosampler were immersed in 10% v/v nitric acid for
24 h, rinsed with ultrapure water and dried before being
used.
The sh feces and feed samples were dried at 50 C in an
oven with forced air circulation for 48 h and then cryogen-
ically ground as described elsewhere (Rosa et al., 2002). A
portion of the samples was also mineralized in a microwave
oven, as follows. Portions of 100 mg of cryogenically
ground samples were transferred directly to the Teon
asks of the microwave oven, and 2.5 ml of suprapure
nitric acid 14 mol l1 plus 0.50 ml of hydrogen peroxide
30% m/m were added. The heating program employed
was the one proposed in the ovens user manual with some
modications (Gallego et al., 1996; Arruda and Fo, 1999).
2.2. Preparation of slurry samples
After cryogenic grinding, 5 mg of samples of biological
material (sh feed or feces) were transferred directly to
the containers of the spectrometers autosampler, to
which were added 5 ll of suprapure nitric acid 14 mol l1,
50 ll of Triton X-100 at 1% v/v, 100 ll of Pd(II) solution
1000 mg l1 and 845 ll of ultrapure water. The slurry sam-
ples of biological material were then sonicated for 40 s
directly in the autosamplers containers.
2.3. Apparatus
A Provecto Analtica model DGT 100 plus microwave
oven (Campinas, SP, Brazil) was used to mineralize the
samples whenever necessary.
For the selenium determinations, a Shimadzu model
AA-6800 atomic absorption spectrometer was used,
equipped with a background absorption corrector with a
deuterium lamp and self-reverse (SR) system, and a pyro-
lytic graphite tube with integrated platform and automatic
ASC-6100 sampler. A Shimadzu hollow cathode selenium
lamp operated with a 10 mA current was also used. The
wavelength applied was 196.0 nm and the spectral resolu-
tion was 0.5 nm. Argon was used as inert gas at a constant
ow of 1 l min1 throughout the heating program, except
during the atomization step, when the gas ow was inter-
rupted. The absorbance signals were measured in the peak
area.
The samples were cryogenically ground in a SpexFree-
zer model Mill 6750 cryogenic mill.
The slurries of sh feces and feeds were shaken in a
Unique ultrasonic cell disruptor.
2.4. Preparation of the graphite tube coated internally with
metallic tungsten
The inner walls of the pyrolytic graphite tubes with inte-
grated platform used for determining selenium were coated
with metallic tungsten. This was done by injecting aliquots
1544 F.A. Silva et al. / Chemosphere 68 (2007) 15421547

of 25 ll of a solution containing 1000 mg l1 of the sodium Table 1

tungstate modier into the atomizer, which was then sub- Graphite tube heating program optimized for determination of Se in
slurries of sh feed and feces
mitted to the stages of the heating program described by
Lima et al. (1999). Such procedure was repeated 20 times. Step Temperature Stage Argon ow
(C) (l min1)
When heated to 500 C, the W(VI) deposited on the graph- Ramp (s) Hold (s)
ite tubes inner wall forms a layer of W 0 that acts as a Drying 90 10 0 1
chemical modier (Fisher and Rademeyer, 1998). In this Drying 150 10 5 1
case, the mass of W 0 deposited was 500 lg. With the treat- Drying 250 10 5 1
Pyrolysis 1400 10 20 1
ment it was possible the use of the graphite tubes for up to Pyrolysis 1400 5 10 1
572 rings. Atomization 2400 1 5 0
Cleanup 2800 5 0 1

2.5. Preparation of the standard slurries

Analytical curves were prepared using sh feed and feces
slurries containing 5, 10, 15, 20 and 25 lg l1 of selenium
(Merck), with the absorbance readings done by GFAAS.
These standard slurries were prepared under the same con-
ditions as those used for preparing the slurries of feed and
feces samples, using, however, 5 mg of standard samples of
biological material devoid of selenium. Thus, to prepare
analytical curves, volumes of 10, 20, 30, 40 and 50 ll of
standard solutions containing 500 lg l1 of selenium were
transferred to the containers of the spectrometers auto-
sampler, so that the selenium concentrations in the stan-
dard slurries were within a range of 525 lg l1. The
standard solutions containing 525 lg l1 of selenium in
10% v/v suprapure HNO3, were also used to make analyt-
ical curves in the selenium determinations in samples of
feces and feeds mineralized by acid digestion using micro-
wave oven.
2.6. Analytical procedures
After the sonication step of the sample in slurries and/or
standard slurries directly in the autosamplers containers, a
volume of 20 ll of standard or sample was injected into the
graphite tube (coated internally with metallic tungsten),
using the autosamplers micropipette. Each measurement
was repeated ve times. Table 1 describes the heating pro-
gram of the graphite tube optimized to determine selenium.
3. Results and discussion
3.1. Optimization of the instrumental conditions
The obtainment of the exact and reproducible analytical
results in the determination of metals by GFAAS using
samples in slurry depends on the optimization of the tem-
peratures of pyrolysis and atomization of the analyte.
Therefore, pyrolysis and atomization curves were drawn
to determine these parameters for the selenium in standard
slurries of sh feed and feces containing 15 lg l1 of Se,
100 mg l1 of Pd(II), using the graphite tube coated inter-
nally with W0 and the sample preparation conditions
described under Section 2.2. Fig. 1 illustrates the inuence
of the pyrolysis and atomization temperatures on the
absorbance signal got for the selenium in the standard slur-
ries of the biological materials. The pyrolysis temperature
of 1400 C was selected because, as Fig. 1A indicates, the
absorbance signals obtained for the selenium remained
constant from 600 C up, declining quickly after reaching
1400 C. As for the atomization temperature (Fig. 1B),
the absorbance signals obtained for the selenium were con-
stant from 2100 C up for both standard slurries, so the
atomization temperature of 2400 C was selected for all
the remaining experiments. Fig. 2 depicts the analyte
absorbance (AA) and background absorbance (BG) signals
for the standard slurries of the biological materials. Both

A 0.060 a - Feces standard slurry

B 0.08 a - Feces standard slurry
b - Feed standard slurry b - Feed standard slurry a
b
0.045
a
0.06
Absorbance

b
Absorbance

0.030

0.04

0.015

0.02

0.000
500 750 1000 1250 1500 1750 1500 1800 2100 2400 2700 3000
o o
Pyrolysis temperature ( C) Atomization temperature ( C)

Fig. 1. Se pyrolysis temperature (A) and Se atomization temperature (B) of the standard slurries of feces (j) and feed (d) containing 10 lg l1 of Se using
W0 as permanent modier with co-injection of Pd(II).
 F.A. Silva et al. / Chemosphere 68 (2007) 15421547 1545

A 0.08
B Feed standard slurry
Feces standard slurry
0.08
AA AA

0.06
0.06

Absorbance
Absorbance

0.04
0.04

0.02
0.02
BG BG

0.00 0.00
0 1 2 3 4 5 6 0 1 2 3 4 5 6
Time (s) Time (s)

Fig. 2. Atomic absorption (AA) and background (BG) signals in the atomization of Se in slurries of feces (A) and slurries of feed (B) containing 10 lg l1
of Se using W0 as permanent modier with co-injection of Pd(II).

(Fig. 2A and B) show a relatively low background absor- 0.09

a - Feces standard slurry
bance, indicating the eciency of W0 as a chemical modi- b - Feed standard slurry
a
er. As the biological materials studied here displayed
magnesium content of about 0.12%, this element in the 0.08

matrix may help in the thermal stabilization of some ana-

Absorbance
b
lytes as, for instance, chromium and selenium (Aleixo
0.07
et al., 2000; Rosa et al., 2002).

3.2. Determination of the optimal sonication time of the 0.06

slurry samples
0 10 20 30 40 50 60 70
Ultrasonic shaking to analyze slurries of solid materials Ultrasonication time (s)
ensures good homogenization of the sample, allowing for
Fig. 3. Inuence of sonication time on the absorption signal of Se in
better reproducibility between measurements (Millerihli,
standard slurries of feces (j) and feed (d) containing 10 lg l1 of analyte
1993). Thus, the samples sonication time was evaluated in using W0 as permanent modier with co-injection of Pd(II).
the interval of 560 s of agitation. Fig. 3 depicts the inu-
ence of the sonication time of samples on the absorbance
signals obtained for Se. An analysis of this gure indicates 0.15
a - Feces standard slurrry a
that the absorbance signals remain constant starting from A = 0.00183 - 0.00521k
b
20 s of sonication. The sonication time of 40 s was consid- 0.12
b - Feed standard slurry
A = 0.00131 - 0.00423k
ered optimal, not only for a good absorbance signal
obtained, but also because the RSD among the measure-
Absorbance

0.09

ments was relatively low (2.7%).

0.06

3.3. Obtainment of the analytical curves

0.03

Based on the parameters of pyrolysis and atomization

0.00
temperature and the prole of the optimized atomic 5 10 15 20 25 30
-1
absorption signal, analytical curves were plotted using Concentration (g l )

standard slurries of sh feces and feed containing Se in
the range concentration of 525 lg l1 (as described before
in Section 2.5). Fig. 4 depicts the analytical curves obtained
and their respective straight-line equations. A comparison
between the analytical curve prepared with sh feces (curve
a) and the one prepared with sh feed (curve b) indicates
that their slopes present close values (k = 0.00521 for feces
curve a and k = 0.00423 for feed curve b). Both analytical
curves show absorbance values approximately 510%
lower than the values of the analytical curve obtained from
the standard solutions prepared in the range of 525 lg l1
of Se in 10% v/v suprapure HNO3 (straight-line equation:
Fig. 4. Analytical curves obtained from standard slurries of feces (j) and
feed (d) of Se using W0 as permanent modier with co-injection of Pd(II).
A = 0.00382 + 0.00172 k). However, the slopes of these
straight-lines obtained for the standard slurries were about
2.5-fold greater, which indicates that the sensitivity of the
proposed method is also greater, attesting to the eciency
of the pyrolysis and atomization temperature stage of the
heating program employed. An accumulation of carbona-
ceous residues inside the graphite tube, causing partial
obstruction of the radiation from the hollow cathode lamp,
can impair absorbance measurements (Aleixo et al., 2000).
1546 F.A. Silva et al. / Chemosphere 68 (2007) 15421547
Table 2
Results (n = 5) obtained for selenium in enriched feed samples using the
developed procedure with slurry sampling and with the microwave-
assisted acid digestion procedure
Samples feed Slurry sampling Microwave-assisted acid
(lg kg1) digestion (lg kg1)
Yeast 252 17 249 15
Corn 524 31 522 34
Soybean bran 612 47 609 42
Fish bran 814 51 811 48
Rice brana 296 26 294 24
a
Certied reference material (300 40 lg kg1) from National Institute
of Standards and Technology.
Nevertheless, the low background (BG) signals obtained in
the selenium optimization stage of the proposed procedure
indicate that the carbon residue left behind by the pyrolysis
state did not impair the absorbance measurements. The
characteristic masses calculated in relation to the standard
slurry of 15 lg l1 Se were 34 and 53 pg for the standard
slurry feces and feed, respectively. The detection limit
(LOD) and the quantication limit (LOQ) were calculated
based on the standard deviation of 20 readings obtained
for the blanks of the standard slurries and the slopes of
the analytical curves (LOD = 3r/slope and LOQ =
10r/slope). Their values were 0.31 lg l1 and 1.03 lg l1
of Se, respectively, that correspond to 62 lg kg1 and
206 lg kg1, for the standard slurry feces and 0.35 lg l1
and 1.16 lg l1 of Se, according to 70 lg kg1 and
232 lg kg1, for the standard slurry feeds (Rosa et al.,
2002). A comparison between the value of the detection
limit obtained by the proposed method with that acquired
by the GFAAS using aqueous standard solutions
(0.70 lg l1 of Se), shows greater sensitivity to the pro-
posed method. Lower detection limits of Se in food using
the same technique have been observed in the literature
(Tuzen et al., 2007). However, the proposed method in this
study, that is concerning speciation of this metallic nutri-
ent, included a preconcentration step. The lifetime of the
graphite tube was equivalent to 572 rings. Considering
the complexity of the biological matrices, the tubes service
life with the proposed method is acceptable when
compared with other methods described in the literature
(Aleixo et al., 2000; Rosa et al., 2002; Minami et al.,
2004). After optimization, the accuracy of the proposed
method for the determination of selenium was investigated
in slurries of ve feed samples enriched with this micronu-
trient and one standard reference material of rice ower
Table 3
Coecient of bioavailability of Se of Nile tilapia juveniles fed with feed containing dierent food supplements
Samples feed
Yeast Corn
Coecient of bioavailability (%) 61 2a 63 2b 60 4a
a
Calculation based on the %Se determined by proposed method.
b
Calculation based on the %Se determined by GFAAS after mineralization of the feed and feces samples in a microwave oven.
that is used in the diet of sh (Table 2). The accuracy of
the results obtained was checked making the mineralization
of the samples in a microwave oven, being observed that it
was not statistically dierent at a 95% condence level
(paired t-test).
3.4. Application of the proposed method
After the procedures of optimization and the LOD and
LOQ determination, the applicability of the newly devel-
oped method was tested in the determination of Se in four
samples of feed containing dierent selenium supplements
used in the diet of Nile tilapia juveniles and in samples of
feces from these sh. Then, based on the values of the per-
centage of Cr2O3 and of selenium determinations of the
feeds, a calculation was made to estimate the coecient
of bioavailability of that micronutrient, using Eq. (1) (Fur-
uya et al., 2001)
   
%Cr2 O3 feed %Sefeces
Da 100  100 1
%Cr2 O3 feces %Sefeed
where Da is the apparent bioavailability, %Cr2O3 feed the
percentage of chromic oxide in the feed, %Cr2O3 feces the
percentage of chromic oxide in the feces, %Sefeed the per-
centage of selenium in the feed, and %Sefeces is the percent-
age of selenium in the feces.
Table 3 lists the values of the coecients of bioavailabil-
ity calculated based on the determinations of selenium by
the proposed method and by GFAAS after mineralization
of the feed samples in a microwave oven. A comparison of
the values of the coecient of bioavailability of selenium
present in the four types of feed used in the diet of Nile tila-
pia juveniles (Table 3) reveals that the values found based
on the determinations using the proposed method are
agreeable with those obtained by the GFAAS method after
mineralization of the feed and feces samples in a micro-
wave oven. The GFAAS method is normally used in sh
nutrition mineral bioavailability studies (Furuya et al.,
2001; Pezzato et al., 2002; Sa et al., 2004), hence, our
results attest to the applicability of the proposed method
in such studies.
4. Conclusions
The proposed method for quantifying Se using samples
of sh feed and feces in the form of suspensions to estimate
the apparent bioavailability of this micronutrient in feeds
used in sh nutrition yielded results comparable to those
Soybean bran Rice bran
63 3b 62 2a 64 2b 63 3a 65 3b
 F.A. Silva et al. / Chemosphere 68 (2007) 15421547
obtained with the GFAAS quantication method, whose
initial step involves the mineralization of samples in a
microwave oven. The main advantage of the proposed
method is that it does not generate toxic residues, which
can be harmful to the analysts health and contaminate
the environment. Moreover, since this new method does
not require mineralizing the samples, it considerably
reduces the time spent on analytical determinations in sh
nutrition analyses. In addition to this, it oers limits of
detection (LOD) and of quantication (LOQ) in the order
of 0.31 lg l1 and 1.16 lg l1, respectively, that correspond
to 62 lg kg1 and 232 lg kg1, using only 20 ll of slurry
samples for each analytical determination. The comparison
of values of the detection limits calculated by the proposed
method with that got by the GFAAS (0.70 lg l1) using
aqueous standard solutions, shows greater sensitivity to
the proposed method.
Acknowledgement
The authors gratefully acknowledge the nancial sup-
port of FAPESP (Brazil) (Process 03/13362-6) and CAPES
for fellowships granted to Fabio Arlindo Silva.
References
Aleixo, P.C., Nobrega, J.A., Santos, D.J., Muller, R.C.S., 2000. Deter-
minacao de selenio em agua de coco e em leite de coco utilizando
espectrometria de absorcao atomica com atomizacao em forno de
grate. Quim. Nova 23, 310312.
Arruda, M.A.Z., Fo, E.R.P., 1999. Mechanised ow system for on-line
microwave digestion of food samples with o-line catalytic spectro-
photometric determination of cobalt at ng/l levels. Analyst 124, 1873
1877.
Belchier, R., Forster, H., 1970. The Destruction of Organic Matter.
Pergamon Press Ltd., New York, pp. 1926.
Bell, J.G., Cowey, C.B., 1989. Digestibility and bioavailability of dietary
selenium from shmeal, selenite, selenomethionine and selenocystine in
Atlantic salmon (Salmo salar). Aquaculture 81, 6168.
Bendicho, C., Loos-Vollebregt, M.T.C., 1991. Solid sampling in electro-
thermal atomic absorption using commercial atomize. J. Anal. Atom.
Spectrom. 6, 353374.
Combs, G.F., Liu, C.H., Lu, Z.H., Su, Q., 1984. Uncomplicated selenium
deciency produced in chicks fed a corn-soy-based diet. J. Nutr. 114,
964976.
Combs, G.F., Combs, S.B., 1986. The Role of Selenium in Nutrition.
Academic Press, London, p. 180.
Fisher, J.L., Rademeyer, C.J., 1998. Kinetics selenium atomization in
electrothermal atomization atomic absorption spectrometry (ETA-
AAS). Part 2: selenium with palladium modiers. Spectrochim. Acta,
Part B 53, 549567.
1547
Furuya, W.M., Pezzato, L.E., Pezzato, A.C., Barros, M.M., Miranda,
E.C., 2001. Coeciente de digestibilidade e valores de aminoacidos
digestveis de alguns ingredientes para tilapia do Nilo (Oreochromis
niloticus). R. Bras. Zootec. 30, 11431149.
Gallego, M., Arruda, M.A.Z., Valcarcel, M., 1996. Semi-on-line micro-
wave-assisted digestion of shellsh tissues for the determination of
selenium by electrothermal atomic absorption spectrometry. J. Anal.
Atom. Spectrom. 11, 169173.
Liang, Y.Z., Li, M., Rao, Z., 1996. Nickel and strontium nitrates as
modier for determination of selenium in urine by zeeman platform
graphite-furnace atomic absorption spectrometry. Anal. Sci. 12, 629
633.
Lima, E.C., Barbosa, F., Krug, F.J., 1999. Tungsten-rhodium permanent
chemical modier for lead determination in sediments slurries by
electrothermal atomic absorption spectrometry. J. Anal. Atom.
Spectrom. 14, 19131918.
Lovell, R.T., 1998. Nutrition and Feeding of Fish. Kluwer Academic
Publishers, Massachusetts, p. 267.
Mcdowell, L.R., 1992. Minerals in Animal and Human Nutrition.
Academic Press, New York, pp. 118.
Millerihli, N.I., 1993. Advances in ultrasonic slurry graphite furnace
absorption atomic spectrometry. Fresen. J. Anal. Chem. 345, 482489.
Minami, H., Yada, M., Yoshida, T., Zhang, T., Zhang, Q., Inoue, S.,
Atsuya, I., 2004. Simultaneous direct determination of aluminum,
calcium and iron in silicon carbide and silicon nitride powders by
slurry-sampling graphite furnace AAS. Anal. Sci. 20, 455459.
Oliveira, J.E.D, 1998. Ciencias Nutricionais, Salvier, Sao Paulo, p. 402.
Pezzato, L.E., Barros, M.M., Miranda, E.C., Pinto, L.G.Q., Furuya,
W.M., Pezzato, A.C., 2002. Digestibilidade aparente de ingredientes
pela tilapia do Nilo (Oreochromis niloticus). R. Bras. Zootec. 31, 1595
1604.
Poppe, T.T., Hastein, T., Froslie, A., Koppang, N., Norheim, G., 1986.
Nutritional aspects of hemorrhagic syndrome (Hitra Disease) in
farmed Atlantic salmon (Salmo salar). Dis. Aquat. Organ. 1, 155
162.
Poston, H.A., Combs, G.F., Leibovitz, L., 1976. Vitamin E and selenium
inter-relations in the diet of Atlantic salmon (Salmo salar) gross,
histological, and biochemical deciency signs. J. Nutr. 106, 892
904.
Rosa, C.R., Moraes, M., Neto, J.A.G., Nobrega, J.A., Nogueira, A.R.A.,
2002. Eect of modiers on thermal behaviour of Se in acid digestates
and slurries of vegetables by graphite furnace atomic absorption
spectrometry. Food Chem. 79, 517523.
Sa, M.V.C., Pezzato, L.E., Barros, M.M., Padilha, P.M., 2004. Optimum
zinc supplementation level in Nile tilapia (Oreochromis niloticus)
juveniles diets. Aquaculture 238, 385401.
Smith, R.M., 1987. Trace Elements in Human and Animal Nutrition.
Academic Press, New York, p. 480.
Solomons, N.W., 1992. Absorption and Malabsorption of Mineral
Nutrients. Alan R. Liss, New York, p. 125.
Tuzen, M., Saygi, K.O., Soylak, M., 2007. Separation and speciation of
selenium in food and water samples by the combination of magnesium
hydroxide coprecipitation-graphite furnace atomic absorption spec-
trometric determination. Talanta 71, 424429.
Watanabe, T., Kiron, V., Satoh, S., 1997. Trace minerals in sh nutrition.
Aquaculture 151, 185207.