Journal of Fish Biology (2004) 65, 1233–1252
doi:10.1111/j.1095-8649.2004.00499.x, available online at http://www.blackwell-synergy.com
A non-invasive stress assay based upon measurement of
free cortisol released into the water by rainbow trout
T. E L L I S *†, J. D. J A M E S *, C. S T E W A R T ‡
AND
A. P. S C O T T *
*CEFAS Weymouth Laboratory, Barrack Road, The Nothe, Weymouth, Dorset,
DT4 8UB, U.K. and ‡CEFAS Lowestoft Laboratory, Pakefield Road,
Lowestoft, Suffolk, NR33 0HT, U.K.
(Received 14 May 2003, Accepted 16 June 2004)
A procedure previously used for sex steroids was adapted to extract free cortisol and cortisone
from water samples taken from rainbow trout Oncorhynchus mykiss tanks. Both corticosteroids
could be readily detected by radioimmunoassay (RIA), with cortisol being predominant. All
stages of the sampling, extraction and RIA procedure were validated for cortisol. An intermittent problem with poor replication was traced to the use of diethyl ether during the extraction
procedure, and was overcome by the use of ethyl acetate. Other modifications were also
introduced to speed up the procedure. The concentration and time course of release of both
corticosteroids were shown to be related to the degree of stress that the fish had been subjected
to. It was confirmed that cortisol concentrations in water and estimated cortisol release
rates increased in response to handling stress, and that both were correlated with plasma
cortisol concentrations. The potential for using water cortisol concentration and release rates
to assess the primary stress response of fishes as a non-invasive alternative to blood sampling is
# 2004 Crown copyright
discussed.
Key words: cortisol; non-invasive assay; rainbow trout; stress.
INTRODUCTION
There is much interest in assessing the stress level of fishes. The welfare of farmed
fishes is of increasing concern, and the ‘level of stress’ is an important welfare
indicator, as stress has detrimental effects on growth, reproduction, immunological
function and survival (Pickering, 1992; Ellis et al., 2002; C. Adams, V. Braithwaite,
F. Huntingford, S. Kadri, T. Pottinger & J. Turnbull, pers. comm.). In response to
acute and chronic stress, the fish interrenal synthesises corticosteroids, principally
cortisol, the concentration of which in blood plasma is commonly used to indicate
stress level (Jeney et al., 1992; Pickering, 1992; Mommsen et al., 1999). The required
blood sampling procedure, however, is problematic. Firstly the very act of sampling
is a potential stressor and hence elevates cortisol concentrations in handled fishes;
a problem that can be partially overcome by bleeding the fishes rapidly and
using anaesthetic (Pottinger et al., 1992). Obtaining true basal concentrations,
however, can still be difficult because a plasma cortisol response has been detected
†Author to whom correspondence should be addressed. Tel.: þ44 (0) 1305 206600; fax: þ44 (0) 1305 206601;
email: [email protected]
1233
#
2004 Crown copyright
1234
T. ELLIS ET AL.
within 30 s of applying an acute stress (Gerwick et al., 1999). Secondly the
disturbance associated with removal of a fish can elevate cortisol concentrations
in the remaining fishes (Pickering et al., 1982; Laidley & Leatherland, 1988). This
means that none of these fishes are of any further use in an experiment (unless
given several days to recover). This in turn means that any experiment involving
the measurement of cortisol at several time points is inherently complex, requiring
as it does a separate tank of fishes for each sampling point (Barton, 2000). Thirdly,
blood sampling can be lethal, due to the dose of anaesthetic used to eliminate
a cortisol stress response (Laidley & Leatherland, 1988), or the need to sacrifice
animals to obtain a blood sample (Oliveira et al., 1999).
To avoid the problems inherent in blood sampling, avian and mammalian
researchers have developed methods to measure corticosteroid concentrations
either less invasively (in hair and saliva samples) or non-invasively (in urine and
faecal samples) (Parrot & Misson, 1989; Wasser et al., 2000; Koren et al., 2002).
Previous research has shown that fishes excrete metabolized steroids via the
urine and bile (Vermeirssen & Scott, 1996), and measurement of cortisol metabolites in faeces has been used as a measure of stress in fishes (Oliveira et al.,
1999; Turner et al., 2003). Research that has been carried out on the excretion
of sex steroids in fishes, however, suggests an alternative non-invasive way of
monitoring cortisol status in fishes. In rainbow trout Oncorhynchus mykiss
(Walbaum) and goldfish Carassius auratus (L.), it has been shown that certain
free (i.e. unconjugated) steroids are released into the water via the gills
(Vermeirssen & Scott, 1996; Sorensen et al., 2000). It has also been shown
that: the release of these sex steroids is closely tied to specific physiological (in
this case reproductive) events (van der Kraak et al., 1989; Stacey et al., 1989;
Scott & Sorensen, 1994); the amounts of sex steroid released are substantially
increased by injection of gonadotrophin or gonadotrophin-releasing hormone
(Sorensen & Scott, 1994; Greenwood et al., 2001); and the pattern of release of
the steroids matches the pattern of secretion in the plasma (Stacey et al., 1989;
Greenwood et al., 2001). Also, in one of these studies (Sorensen & Scott, 1994),
it was shown that goldfish released cortisol into the water and that spawning
males and females released substantially more free (and sulphated) cortisol into
the water than non-spawning females. On the basis of these findings, the present
study set out to establish whether the major free corticosteroids in rainbow
trout, cortisol and cortisone (Pottinger & Moran, 1993), are also released into
the water and, if so, develop a robust procedure for their sampling, extraction and
assay. The principle of this work has already been outlined (Scott et al., 2001).
MATERIALS AND METHODS
RADIOIMMUNOASSAYS
Cortisol antibody (sheep anti-cortisol) was obtained from Diagnostics Scotland
(Carluke, U.K.). Cortisone antibody was obtained from Chemicon Europe (The Science
Centre, Eagle Close, Chandlers Ford, SO53 4NF, U.K.; product code: AB1299).
Tritiated cortisol was purchased from Amersham Biosciences and tritiated cortisone
was synthesized by chromic acid oxidation of cortisol as described by Truscott (1981).
Standard cortisol and cortisone were obtained from Sigma Aldrich, U.K. Assay buffer
#
2004 Crown copyright, Journal of Fish Biology 2004, 65, 1233–1252
A NON-INVASIVE STRESS ASSAY FOR FISHES
1235
comprised 8 g l 1 NaCl, 58 g l 1 Na2HPO4, 2 g l 1 BSA, 13 g l 1 NaH2PO4H2O, 03 g l 1
EDTA and 01 g l 1 sodium azide.
For radioimmunoassay (RIA) measurement of cortisol or cortisone, 100 ml of extract
or dilution of extract was added to duplicate glass tubes. Nine cortisol or cortisone
standards ranging from 2 to 500 pg 100 ml 1 were made up by serial dilution. To all
tubes was then added 100 ml of assay buffer containing c. 5500 DPM of radioactive
steroid and sufficient antibody to bind c. 40% of the radiolabelled corticosteroid in the
absence of radioinert steroid. Tubes were left to equilibrate overnight (16 h, 4 C), and
unbound steroid was then removed by addition of dextran-coated charcoal (Scott et al.,
1982; on ice; 30 min incubation; 12 min centrifugation at 1000g).
The supplier of the cortisol antibody provided cross-reactivity data indicating it
to be highly specific (<06% cross-reactivity with 20 steroids including cortisone, 11deoxycortisol, deoxycorticosterone, corticosterone). The supplier of the cortisone antibody also supplied cross-reactivity data indicating a 5% cross-reaction with cortisol. The
cortisol and cortisone antisera specificities were checked by thin layer chromatographic
(TLC) separation of rainbow trout water extracts, with the mobile phase consisting of
chloroform : ethanol 50 : 3 (v/v). The reproducibility (interassay coefficient of variation,
CV) of the cortisol RIA was assessed from repeated assays of two ‘line standards’ derived
from extracted rainbow trout water samples. The precision (intra-assay CV) of the
cortisol RIA was examined for three samples differing in cortisol concentration. Parallelism of the cortisol RIA was examined by assay of extracted rainbow trout water samples
serially diluted in a manner analogous to the dilution of the standards.
W A T E R S A M P L I N G A N D P R O C E S SI N G
Tank water samples were collected in plastic bottles via a siphon tube fitted with a tap,
to avoid disturbing the fish. Gloves were worn at all times to prevent contamination.
Corticosteroids were extracted from water samples as described by Greenwood et al.
(2001). Briefly, tank water samples were peristaltically pumped at c. 25 ml min 1 through
a pre-filter (045 mm pore-size: Pall Life Sciences, U.K.) and then through an activated
solid phase extraction cartridge (Sep-pak1 Plus C18, Waters Ltd., U.K.). Although this
pumping speed is higher than that recommended by the manufacturer and that used in
other studies (Greenwood et al., 2001), the recovery was found to be satisfactory.
Approximately 500 ml of water was typically processed with the actual volume determined gravimetrically. After pumping, the cartridges were washed with 5 ml de-ionized
water (DI) and either stored on ice temporarily (<2 h), or frozen prior to elution.
For the original elution procedure, as described by Sorensen & Scott (1994) and
Greenwood et al. (2001), all corticosteroids (free and conjugated) were retrieved from
the cartridge by elution with 4 ml methanol, evaporated at 45 C under nitrogen,
re-dissolved in 200 ml DI and the free steroid fraction extracted from the DI with 4 ml
diethyl ether. The diethyl ether was transferred to a fresh glass tube and allowed to
evaporate in a water-bath (45 C) in a fume cupboard. The residue was then reconstituted
in 1 ml RIA buffer and stored frozen until assayed.
Although diethyl ether has been widely used in the assay of cortisol and other steroids
(Sorensen & Scott, 1994; Oliveira et al., 1999; Greenwood et al., 2001), occasional
anomalous results occurred with the above method that were subsequently attributed
to the quality of the diethyl ether. The elution procedure was therefore revised whereby,
after the cartridge was purged of excess DI, free corticosteroids were eluted directly with
4 ml ethyl acetate. The ethyl acetate was evaporated under a stream of nitrogen gas at
45 C, and the residue was dissolved in 1 ml RIA buffer and stored frozen until assayed. It
was confirmed that the ethyl acetate did not elute conjugated corticosteroids (unpubl.
data), and that the revised procedure had the additional benefit of removing the need for
a phase separation step.
To determine whether cortisol concentration was affected by the volume of water that
was processed, 30 replicate water samples were taken from a stock tank of rainbow trout.
Three different volumes (250, 500 and 1000 ml) of water were pumped, with 10 replicates
#
2004 Crown copyright, Journal of Fish Biology 2004, 65, 1233–1252
1236
T. ELLIS ET AL.
per volume. The cortisol was retrieved from the extraction cartridges using the original
(methanol/ether) elution method and assayed.
The recovery of cortisol from water was assessed by adding radioinert cortisol to water
samples from a rainbow trout stock tank to give a predicted cortisol addition of 5 ng l1.
Spiked (n ¼ 4) and unspiked (n ¼ 4) water samples (500 ml) were then pumped through
extraction cartridges. The cortisol was retrieved from the cartridges using the original
(methanol/ether) elution method, and the amount was quantified by RIA.
The effect of freeze storage, of either the water sample or the extraction cartridge, was
examined at the same time as comparing the original (methanol/ether) elution with the
revised (ethyl acetate) elution procedure. The experiment was therefore a 3 2 design
with three storage treatments (no storage, freeze storage of water sample and freeze
storage of extraction cartridge) and two elution procedures. Replicate 500 ml water
samples (36) were taken from a rainbow trout stock tank, providing six replicates per
treatment. Frozen extraction cartridges and water samples were thawed for 1 and 18 h
respectively at room temperature prior to elution or pumping.
BLOOD SAMPLING AND PROCESSING
Blood samples were collected in heparinized syringes by caudal venipuncture from fish
given a lethal dose of the anaesthetic benzocaine (012 g l1) to prevent a cortisol stress
response to handling (Olsen et al., 1995). The fish were completely anaesthetized in
<1 min, and all blood samples were taken within 5 min of removal from the tank.
Blood samples were centrifuged (10 400g for 3 min) to separate the blood cells, and the
plasma decanted and stored frozen. Cortisol was extracted by adding 1 ml ethyl acetate to
100 ml plasma. The sample and solvent were vortex mixed, shaken for 5 min, and then
centrifuged (10 400g for 2 min). The ethyl acetate layer was decanted and evaporated at
45 C under nitrogen, and the residue re-dissolved in 1 ml RIA buffer and stored frozen.
The efficiency of extraction (per cent recovery) from plasma was assessed by adding
radioinert cortisol to charcoal-stripped plasma to give predicted concentrations of 0, 1,
10, 50 and 100 ng ml1, and then extracting 100 ml aliquots (n ¼ 6) for RIA.
F I S H E X P ER I M E N T S
Rainbow trout were purchased as eggs from the Gwen Wyllin Trout Hatchery (Isle of
Man) (all female) and ongrown at the CEFAS Weymouth laboratory. For experiments,
the fish were held at 12 C in aerated flow-through circular tanks supplied with dechlorinated mains water (pH 79). Water flow rates were monitored with in-line flowmeters
(KDG 2000, KDG Mobrey, Crawley, U.K.) that were calibrated gravimetrically. The
fish were fed commercial pellets at a rate of 18% body mass per day. Fish were, however,
not fed on the day of stressor application. To produce a standardized response, an acute
handling stress (90 s aerial exposure) was applied. A crowding screen was used to
concentrate all the fish within a tank into a net. This net was removed from the tank,
and held in the air for 90 s before returning the fish to the tank. Similar netting and aerial
exposure procedures have been widely used to ensure a reproducible stress response
applied to all fish in a tank (Kebus et al., 1992).
A S S ES S M E N T O F B A S A L A N D P O S T - S T R E S S W A T E R
CORTICOSTEROID CONCENTRATION
The aim of the experiment was to establish whether water concentrations of cortisol
and cortisone changed in response to stress. Twelve tanks (mean S.E. volume
1460 24 l and mean S.E. flow rate 19 00 l min1) were each stocked with 25 rainbow trout. After a 2 month acclimation period, the mean S.E. individual mass, stocking
density and biomass loading were 176 3 g, 30 1 kg m3 and 23 00 kg l1 min1. The
experiment involved four replicates of three treatments: an unstressed control, a single
acute stress at 0 h and an acute stress repeated at 0, 1 and 2 h. Such repeated application
of an acute stressor has previously been shown to result in a stepwise increase in plasma
#
2004 Crown copyright, Journal of Fish Biology 2004, 65, 1233–1252
A NON-INVASIVE STRESS ASSAY FOR FISHES
1237
cortisol concentration in salmonids (Barton et al., 1986). Water samples were taken at 0 h
(immediately prior to application of stress), 05 h, and at hourly intervals from 1 h until
8 h, and processed using the original (methanol/ether) elution procedure.
The amount of hormone released (Ht in ng) by the fish within a given time interval (t)
was calculated by adapting an equation derived for the analogous problem of estimating
food consumption rate from stomach content data and gastric evacuation rate (Adams &
Breck, 1990), Ht ¼ Vkt(Ct C0ekt)(1 ekt)1, where V is the water volume (i.e. tank
volume minus fish biomass), C0 and Ct are the hormone concentrations at the beginning
and end of the sampling period (over a time interval t) and k is the instantaneous rate of
decrease due to dilution from the inflow water. Values for k were derived as R V1, where
R is the water inflow rate. Landau (1992) provides a dilution equation predicting the
fraction of water replaced (F) in a well-mixed tank over
a time interval (t):
1
R ¼ ln(1 F)(V t1). This can be rearranged as: 1 F ¼ et(R V ). This equation demonstrates that the term R V1 represents the instantaneous rate of decrease due to dilution
from the inflow water, i.e. k. The hormone release rate (ng g1 h1) was then calculated
from Ht, the fish biomass and time interval.
CORRELATION OF PLASMA CORTISOL CONCENTRATION
A N D R E L E A S E R A TE
This experiment sought to confirm that water cortisol concentration and calculated
release rates were related to plasma cortisol concentration. Rainbow trout (67 per tank)
were acclimated to 274 4 l tanks (28 002 l min1 inflow) for 1 month, after which
time the mean individual mass, stocking density and biomass loading were 164 3 g,
419 09 kg m3 and 396 005 kg l1 min1. The experiment involved three replicates
of three treatments: an unstressed control, a single handling stress at 0 h and a repeated
handling stress at 0, 075 and 15 h. Water samples were taken at 0, 1 and 2 h and
processed using the revised (ethyl acetate) elution procedure. At 2 h, 12 individuals per
tank were sampled for blood. Cortisol release rate was estimated as described above for
the hour preceding blood sampling, i.e. from the water cortisol concentrations at 1 and
2 h.
V A L I D A T I O N O F T H E IN ST A N T A N E O U S R A T E O F
D E C R E A SE O F C O R T I SO L C O N C E N T R A T I O N DU E T O
DILUTION
The aim of the experiment was to confirm whether the Landau (1992) equation reliably
predicted the rate of decrease in cortisol concentrations in flowing tank water. After
removal of the 12 fish for blood sampling from the nine tanks in the above experiment,
all the remaining fish were removed and the inflow rates were changed to either 0, 25 or
5 l min1, one flow treatment at each stress level (control, single, repeated). The static
treatment was used to assess the possible contribution of biodegradation and adsorption
to the disappearance of cortisol from the tanks (Jürgens et al., 2002). Water samples were
taken for measurement of cortisol concentration at intervals based upon the time for 50%
loss as predicted by the dilution equation. Water samples were processed using the
revised (ethyl acetate) elution procedure. The data were analysed by regressing ln water
cortisol concentration against time. The observed gradients were compared to predicted
values derived from the tank volume and inflow rate (i.e. R V1) for each tank using a
Wilcoxon matched pairs test.
ST A TI S T ICA L A NA LY S E S
Values of corticosteroid concentrations in water and plasma are expressed as
mean S.E. in the text and figures. All values of corticosteroid concentration in water
and plasma, and estimates of release rate were ln transformed for analyses. Treatment
effects were analysed by appropriate ANOVA models.
#
2004 Crown copyright, Journal of Fish Biology 2004, 65, 1233–1252
1238
T. ELLIS ET AL.
RESULTS
RADIOIMMUNOASSAYS
The TLC separations showed that >97% of the cross-reaction of the cortisol
antiserum was in the elution position of cortisol, but only 78% of the cross-reaction
for the cortisone antiserum was in the elution position of cortisone (Fig. 1). The
interassay CV for cortisol was 11% (n ¼ 24) for both line standards (means 12 and
118 pg). The intra-assay CV for cortisol was 6, 7 and 10% for means of 14, 49 and
184 pg, (n ¼ 15, 24 and 18). The slopes for per cent binding of the serially diluted
samples were indistinguishable from the standard curve (Fig. 2).
(a)
Cortisol (ng)
2
1
cortisol
0
1
3
5
7
9
3
5
7
9
11
cortisone
13
15
17
19
21
23
19
21
23
(b)
Cortisone (ng)
0·2
0·1
cortisol
0
1
11
cortisone
13
15
17
Distance from origin (cm × 0·4)
FIG. 1. Thin-layer chromatographic separation of material in water extracts from rainbow trout tanks
that cross-reacted with either (a) the cortisol antiserum or (b) the cortisone antiserum. The elution
positions of cortisol and cortisone standards are indicated.
#
2004 Crown copyright, Journal of Fish Biology 2004, 65, 1233–1252
A NON-INVASIVE STRESS ASSAY FOR FISHES
1239
(a)
40
30
20
Per cent binding
10
0
0
2
4
6
8
2
4
6
8
(b)
40
30
20
10
0
0
Number of 1 : 2 dilutions
FIG. 2. Parallelism of the cortisol standard curve to serial (doubling) dilutions of four water samples in two
separate assays (a) (&, standard 1; , sample 1; ~, sample 2) and (b) (&, standard 2; , sample 3; ~,
sample 4). Values for the rainbow trout water samples are means of six replicates.
W A T E R S A M P L I N G A N D P R O C E S SI N G
There was no effect of sample volume (one-way ANOVA, F2,27, P > 05) on
cortisol concentrations measured in stock tank water samples. The mean water
cortisol concentrations were 200 012, 210 009 and 216 010 ng l1 for
the 250, 500 and 1000 ml volumes.
The cortisol concentration in unspiked stock tank water samples was estimated at
18 02 ng l1. The cortisol concentration in samples that had been spiked with
5 ng l1 was estimated at 62 02 ng l1. The combined recovery of the methodology
(extraction, methanol/ether elution and RIA) was therefore estimated at 87%.
There was no significant effect of freeze storage on cortisol concentration
(two-way ANOVA, F2,30, P > 02), or interaction between storage and elution
#
2004 Crown copyright, Journal of Fish Biology 2004, 65, 1233–1252
1240
T. ELLIS ET AL.
method (F2,30, P ¼ 05), indicating that freeze storage of the water sample or the
extraction cartridge did not cause a significant loss of cortisol. There was,
however, a slight but significant effect of elution methodology on the measured
cortisol concentration (Fig. 3; F1,30, P < 0001), with the revised (ethyl acetate)
method giving higher values (35 v. 31 ng l1). The revised (ethyl acetate)
method also gave less variable results (CV 6 v. 12%).
PLASMA SAMPLING AND PROCESSING
The amount of cortisol measured in the unspiked stripped plasma was below
the limit of detection of the assay. The mean recovery efficiency of the extraction (and RIA) of the spiked plasma was 82%, with no difference between the
spiking concentrations (one-way ANOVA, F3,19, P > 02).
A S S ES S M E N T O F B A S A L A N D P O S T - S T R E S S W A T E R
CORTICOSTEROID CONCENTRATION
Elution and sample storage method
Before application of stress, both cortisol and cortisone were measured in
water samples at concentrations of 11 01 and 07 00 ng l1 respectively
(Fig. 4). There was no difference between treatments in water concentrations
of either corticosteroid prior to application of the stress (one-way ANOVA,
F2,9, P > 05). After 30 min, concentrations of cortisol and cortisone were greatly
elevated in the stressed tanks, over the unstressed controls (one-way ANOVA,
F2,9, P < 0001; Fig. 4). Concentrations of corticosteroids in the single and repeat
stress tanks had diverged by 2 h (Fig. 4). The water cortisol concentration in the single
stress treatment peaked at 252 34 ng l1 at 2 h, whereas concentrations in the
Original elution,
no storage
Original elution,
water frozen
a
Original elution,
cartridge frozen
Revised elution,
no storage
Revised elution,
water frozen
b
Revised elution,
cartridge frozen
0
1
2
Water cortisol (ng
3
4
l–1)
FIG. 3. Cortisol concentrations measured in rainbow trout tank water samples using different sample
storage and cartridge elution methodologies (original methanol/ether and revised ethyl acetate).
Values are means of six replicates S.E. and different lower case letters (a and b) indicate significant
effect of elution methodology.
#
2004 Crown copyright, Journal of Fish Biology 2004, 65, 1233–1252
1241
A NON-INVASIVE STRESS ASSAY FOR FISHES
140
(a)
Water cortisol (ng l–1)
120
100
80
60
40
20
0
0
Water cortisone (ng l–1)
30
2
4
6
8
(b)
20
10
0
0
2
4
6
8
Time (h)
FIG. 4. Mean þ S.E. (a) cortisol and (b) cortisone concentrations in water samples from rainbow trout
tanks (n ¼ 4 per treatment) after exposure to a repeated handling stress (at 0, 1 and 2 h, -&-), a single
handling stress (at 0 h, --) or no stress (control, -~-).
repeat stress treatment peaked between 3 and 5 h, reaching 1072 159 ng l1.
Differences between all three treatments in the water concentrations of both steroids
remained for the duration of the sampling (Fig. 4). Concentrations of cortisol in the
water were higher than cortisone at all sampling times (Fig. 4).
The estimated cortisol and cortisone release rates for the unstressed control
fish were relatively stable over the 8 h period of the experiment (Fig. 5). Release
rates of both cortisol and cortisone for the stress treatments were greater than
for the control over the first period (0 to 05 h) after application of the stress at
0 h (Fig. 5: one-way ANOVA, F2,9, P < 0001). The two stress treatments
diverged during the third period (1 to 2 h), with higher release rates in the fish
exposed to the repeat stress. In the single stress treatment, the release rate of
#
2004 Crown copyright, Journal of Fish Biology 2004, 65, 1233–1252
1242
T. ELLIS ET AL.
(a)
Cortisol release rate (ng g–1 h–1)
8
6
4
2
0
0
4
6
8
(b)
1.0
Cortisone release rate (ng g–1 h–1)
2
0·8
0·6
0·4
0·2
0
0
2
4
6
8
Time (h)
FIG. 5. Mean þ S.E. release rates calculated for the release of (a) cortisol and (b) cortisone into the water in
rainbow trout tanks (n ¼ 4 per treatment) after exposure to a repeated handling stress (at 0, 1 and 2 h, -&-),
a single handling stress (at 0 h, --) or no stress (control, -~-). Values are plotted at the midpoint of the time
interval that release rates were measured over. Values for 025 and 075 h represent estimates over 30 min
periods, i.e. 0 to 05 h and 05 to 1 h respectively. All other values represent estimates over 1 h periods.
both corticosteroids peaked within the first hour, and were not significantly
different from the control treatment after 3 h (Fig. 5). The release rates in the
repeat stress treatment peaked between 2 and 5 h and remained higher for the
duration (8 h) of the experiment (Fig. 5). Estimated release rates were consistently higher for cortisol than cortisone (Fig. 5).
C O R R E L A T I O N O F P L A SM A C O R T I SO L C O N C E N T R A TI O N
AND RELEASE RATE
Prior to the application of the handling stress, the mean water cortisol concentration was 27 03 ng l1 and there was no difference between the different
#
2004 Crown copyright, Journal of Fish Biology 2004, 65, 1233–1252
1243
A NON-INVASIVE STRESS ASSAY FOR FISHES
treatments (one-way ANOVA, F2,6, P > 05). At 2 h, there was a significant
treatment effect [one-way ANOVA, F2,6, P < 0001, Fig. 6(a)], with water cortisol concentrations ranging from 2 to 93 ng l1. There was a significant effect of
Water cortisol (ng l–1)
(a)
a
b
b
Control
Single stress
Repeat stress
80
40
0
Cortisol release rate (ng g–1 h–1)
(b)
3
a
b
c
Control
Single stress
Repeat stress
2
1
0
Plasma cortisol (ng ml–1)
(c)
a
b
c
Control
Single stress
Repeat stress
120
80
40
0
FIG. 6. (a) Water cortisol concentrations at 2 h, (b) estimated cortisol release rate into the water for the
time period 1 to 2 h and (c) plasma cortisol concentrations at 2 h in tanks of rainbow trout exposed
to either no stress (control, n ¼ 3), a single handling stress at 0 h (n ¼ 3) or a repeated handling stress
at 0, 075 and 15 h (n ¼ 3). Values in (a) and (b) represent single observations for each tank, and
values in (c) represent the mean S.E. of 12 fish for each tank. Different lower case letters (a, b and
c) indicate significant differences between treatments.
#
2004 Crown copyright, Journal of Fish Biology 2004, 65, 1233–1252
1244
T. ELLIS ET AL.
treatment on estimated cortisol release rate [one-way ANOVA, F2,6, P < 0001,
Fig. 6(b)], which ranged from 003 to 250 ng g1 h1.
The effects of stress treatment, sampling order, fish mass and tank (stress
treatment) on plasma cortisol concentration was assessed using a four-way
generalized linear model ANOVA. Plasma cortisol was unaffected by sampling
order (F1,97, P > 05) or fish mass (F1,97, P > 02). There was however a highly
significant effect of stress treatment (F2,97, P < 0001). Mean plasma cortisol
concentrations ranged from 2 ng ml1 in unstressed tanks to 127 ng ml1 in
repeat stress tanks [Fig. 6(c)]. There were highly significant positive correlations
between mean plasma cortisol concentration at 2 h and both water cortisol
concentration at 2 h (rs ¼ 085, n ¼ 9, P ¼ 0004) and cortisol release rates
between 1 and 2 h (rs ¼ 093, n ¼ 9, P ¼ 0001).
V A L I D A T I O N O F T H E IN S T A N T A N E O U S R A T E O F
DECREASE OF CORTISOL CONCENTRATION DUE TO
DILUTION
Plots of ln water cortisol concentration against time conformed to straight
line relationships for all three flow regimes, apart from when the concentrations
approached the limit of sensitivity of 03 ng l1 [Fig. 7(a)–(c)]. These values were
excluded from the analysis. All regression equations represented good fits to the
data (r2 93%) and the gradients were significant in all cases (P < 0001),
showing significant decreases in the cortisol concentration over time in all
flow regimes, including the static tanks. The gradients of the three flow treatments differed (one-way ANOVA, F2,6, P > 0001), and the gradients of the 5
and 25 l min1 flow regimes were not significantly different from predicted
values [(R V1) (Wilcoxon matched pairs test, P > 01; Fig. 7(d)].
DISCUSSION
WATER CORTISOL MEASUREMENT AS A NON-INVASIVE
ASSAY
This study confirmed the release of free corticosteroids, both cortisol and
cortisone, into the water by rainbow trout. Cortisone in fish plasma is thought
to originate mainly from oxidation of cortisol. Plasma cortisone concentrations
have been shown to be similar to, and sometimes exceed, those of cortisol
(Weisbart & McGowan, 1984; Patiño et al., 1987; Pottinger & Moran, 1993).
Although cortisone is an important component of the stress response, and may
have a physiologically active role (Patiño et al., 1987), there appears little to be
gained from measuring both steroids on a routine basis. Water cortisol concentrations appear to be consistently higher than those of cortisone both under
basal and stressed conditions (Fig. 4). Cortisol is also the principal, and most
commonly measured, corticosteroid in other fish studies (Mommsen et al.,
1999). Also, antisera for cortisol RIA are more readily available than those
for cortisone and appear to be more specific.
In the present study, the reliability of every stage of the procedure for
measuring cortisol in water from sample collection to RIA was established.
#
2004 Crown copyright, Journal of Fish Biology 2004, 65, 1233–1252
A NON-INVASIVE STRESS ASSAY FOR FISHES
1245
(a)
100
10
1
0·1
Water cortisol (ng l–1)
0
100
1000
2000
1000
2000
(b)
10
1
0·1
0
(c)
100
10
1
0·1
0
1000
2000
Instantaneous rate of decrease
in cortisol concentration (min–1)
Time (min)
0·02
(d)
NS
NS
5
2·5
0·01
0
Static
Inflow rate (l min–1)
FIG. 7. (a)–(c) Decrease in water cortisol concentrations in individual tanks after stressed or unstressed
fish had been removed from the tanks, and the flow rates had then been adjusted to either (a) 5, (b)
25 or (c) 0 (static) l min1. Values (*), close to limit of detection, were not used in calculations of
the rate of disappearance of cortisol. (d) Observed (&, gradients) v. predicted (&, R V1) rates of
decrease of mean S.E. cortisol concentration for each flow regime.
#
2004 Crown copyright, Journal of Fish Biology 2004, 65, 1233–1252
1246
T. ELLIS ET AL.
Buchanan & Goldsmith (2004) have recently pointed out the importance of
carrying out thorough validation during the development of non-invasive
procedures. Despite evidence for some loss of cortisol in static water, cortisol
was flushed out at a predictable rate in running water, thus enabling estimation of cortisol release rates. Water cortisol concentrations and release rates
have been shown to be related to the degree of stress that the fishes are
subjected to. The observed time course of water cortisol concentrations and
release rates, with a rapid rise within 1 to 3 h of a handling stress and a
subsequent decay (Figs 4 and 5), resemble previous studies of plasma cortisol
response (Pickering et al., 1982; Patiño et al., 1987; Barton, 2000). Water
cortisol concentrations and release rates were found to be correlated with
plasma cortisol concentrations (Fig. 6). The procedure therefore provides a
good basis for a non-invasive stress assay for fishes. Water cortisol concentrations (ng l1) can be used directly as a relative measure of stress status in
experimental tank systems with matching fish biomass and flow regimes.
Alternatively, estimated cortisol release rates (ng g1 h1) can be used as an
absolute measure. Release rates will better reflect the plasma status after an
acute stress, as there may be a slight delay in the cortisol concentration peak,
due to the need for accumulation (Figs 4 and 5).
Following their demonstration of accumulation of endogenous cortisol metabolites in the bile of rainbow trout, Pottinger et al. (1992) suggested that bile
samples may provide a means to identify chronically stressed fishes when
sampling stress is unavoidable. The time delay required for accumulation in
the bile would circumvent the problem of the rapid rise in plasma cortisol
concentrations associated with sampling. The disadvantages of measuring cortisol in the bile, however, are that the fishes have to be killed, the time delay is
of little use for studies of short-term stress and the production and storage of
bile is directly affected by feeding (Pottinger et al., 1992).
The logical extension to measurement of cortisol metabolites in bile is the
measurement of cortisol metabolites in faeces. As mentioned in the Introduction, two such studies have already been carried out on fishes by Oliveira et al.
(1999) and Turner et al. (2003). Although this method looks to have much
promise for field studies, it does have several drawbacks. The major one is that
faeces need to be collected very soon after defecation to prevent their disintegration and to minimize the leaching of cortisol metabolites into the water.
Another drawback is that the rate of defecation (and hence of cortisol concentrations) is affected by food quantity and feeding rate. Yet another drawback is
that, without relatively sophisticated technology, it is difficult to collect faeces
without at the same time disturbing the fishes.
Although the amounts of free cortisol in the water that are released via the
gills appear to be a smaller proportion of plasma-derived cortisol than the
metabolites in the bile, the advantages of measuring free cortisol in the water
are numerous. As cortisol in the water comes straight from the gills (unlike
metabolites in the faeces that could have been stored in the gall bladder for
several hours to days), water cortisol concentrations are likely to give a better
picture of what is currently happening to the fishes. There will inevitably be a
time lag between plasma and water cortisol concentrations. The present experiments, however, have demonstrated a seven-fold increase in water cortisol
#
2004 Crown copyright, Journal of Fish Biology 2004, 65, 1233–1252
A NON-INVASIVE STRESS ASSAY FOR FISHES
1247
concentration within 30 min of stressor application, indicating that this time lag
is a matter of only minutes.
A major advantage of the procedure outlined in the present paper is that, as
water sampling does not disturb the fish, it will allow replacement of the stressful
and harmful procedure of blood sampling. This has the additional benefit that
repeat samples can be taken from the same population in time-course experiments, thereby reducing the numbers of tanks and animals that might otherwise
be required. Another potential advantage is that stress experiments can be carried
out on fishes that are too small to be bled. The stress response of small fishes is
currently assessed by making whole body extracts (Pottinger et al., 2002).
Following the publication of the concept for a non-invasive assay for stress in
rainbow trout (Scott et al., 2001), Ruane & Komen (2003) applied the original
extraction procedure of Scott & Sorensen (1994), but without the water/diethyl
ether phase separation, to water from tanks containing carp Cyprinus carpio L.
at different densities. They found a reasonable agreement between cortisol
concentration and fish density. They did not find, however, such a good agreement between plasma and water cortisol levels as in the present study. This
could be due to the failure to separate the free cortisol from the far more
abundant conjugated steroids (many of them metabolites of cortisol) that may
have entered the water via the bile and urine. It is possible that some of these
conjugates could cross-react to a certain degree with the cortisol antiserum and
increase apparent concentrations in the RIA.
Measurement of cortisol in water samples assumes a homogeneous concentration of cortisol within the tank. The validity of this assumption in the present
tank systems is demonstrated by the low variation in cortisol concentrations in
replicate water samples (CV ¼ 6% for revised elution method) due to adequate
mixing from aeration and fish movement. Heterogeneity, however, would need
to be considered in less dynamic or linear flow systems.
A potential drawback of the non-invasive methodology is that it effectively
integrates the cortisol release of all members of a population. Large interindividual differences in plasma cortisol concentration are well documented in
rainbow trout (Laidley & Leatherland, 1988), and individuals with a high
plasma cortisol concentration may make a greater contribution to water cortisol
concentration than predicted from their biomass. It is possible, however, to
apply the methodology to assess the stress level of individuals (unpubl. data).
Another potential drawback to the methodology is that the rate of cortisol
release into the water is likely to depend not only upon the plasma cortisol
concentration, but also on the rate of passage of cortisol through the gills.
Passage through the gills will presumably be influenced by a variety of factors
including gill surface area, permeability, branchial blood flow and ventilatory
water flow. The ratio of gill surface area to body mass decreases with increasing
size (Pauly, 1994), so smaller fishes would potentially release more cortisol for a
similar plasma cortisol concentration. Since temperature and oxygen affect
ventilation frequency, they would also probably affect the cortisol release rate.
The epidermal structure of the gill can also be affected by water quality that
may in turn affect permeability (Anderson & Mitchum, 1974). Stress is also
known to directly increase gill blood flow, ventilation rate and the permeability
of gill epithelia (Mazeaud & Mazeaud, 1981; Bonga, 1997). The release of
#
2004 Crown copyright, Journal of Fish Biology 2004, 65, 1233–1252
1248
T. ELLIS ET AL.
cortisol via the gills is therefore complex and may not respond in direct relation
to changes in plasma cortisol concentration. Although these factors affecting
release may potentially reduce the applicability of this method as an absolute
measure of stress status, they may make it more sensitive as a relative measure.
At present, in order to work out cortisol release rates, accurate information is
required on biomass, water flow rate and tank volume. If any of these factors is
unknown (or not identical for all treatments), then the measurement of cortisol in
water is of limited value. Mammalian researchers have had the same problem
when measuring cortisol in urine as the flow rate of urine is highly unpredictable.
They have resolved this problem by measuring the urinary concentration of
creatinine (Wasser et al., 2000), a compound that is formed continuously in the
body at a rate that is not affected by stress. By relating the concentration of
cortisol to that of creatinine, it becomes unnecessary to know the flow rate of the
urine. The next step in the development of the non-invasive procedure is to find a
similar compound that is produced by fishes, is easy to measure and is released
into the water in the same way as cortisol (i.e. via the gills rather than the urine).
If such a compound can be identified, it will hopefully become unnecessary to
know details of fish biomass, water flow rates, gill permeability, gill surface area,
oxygenation or temperature in order to interpret cortisol concentrations in water.
RELEASE OF FREE CORTICOSTEROIDS INTO THE WATER
The release of cortisol into the water represents a loss of the steroid from the
plasma. There are three main routes for the clearance of free and conjugated
steroids, i.e. via the bile, the urine and the gills (Pottinger et al., 1992; Vermeirssen
& Scott, 1996; Sorensen et al., 2000). The study by Vermeirssen & Scott (1996)
showed that, when rainbow trout were injected with either a free, sulphated or
glucuronidated steroid, the free steroid appeared in the water via the gills, the
sulphated steroid appeared rapidly in the urine and the glucuronidated steroid
accumulated in the bile. By ‘bisecting’ goldfish in specially constructed tanks,
Sorensen et al. (2000) were also able to show that free steroids were released
only from the anterior (gill) region of the fish and sulphated and glucuronidated
steroids only from the posterior region. The current findings indicate that the
release route of cortisol and cortisone in rainbow trout is also the gills. This
conclusion is supported by the fact that cortisol concentrations in the water
increase very soon after the application of stress and are unlikely to be of biliary
origin as bile corticosteroid concentrations do not rise until after 1 h of stress
application (Pottinger et al., 1992). Although they could be of urinary origin,
a preliminary experiment was conducted following the same procedure as
Vermeirssen & Scott (1996), in which three rainbow trout were fitted with a
catheter that prevented urine from entering the water and were then injected
with radioactive cortisol. Free cortisol (some of it converted to cortisone)
appeared in the water (but not the urine) and HPLC analysis of the water
showed little or no evidence for the presence of corticosteroid conjugates
(unpubl. data). Only a very small amount of radioactivity (all of it conjugated)
appeared in the urine, as suggested by Idler & Truscott (1972).
Despite the evidence for the release of free cortisol and cortisone via the gills,
the hepato-biliary route is still considered the major route for cortisol clearance
#
2004 Crown copyright, Journal of Fish Biology 2004, 65, 1233–1252
A NON-INVASIVE STRESS ASSAY FOR FISHES
1249
in fishes (Mommsen et al., 1999). The present results are unfortunately insufficient to estimate just how important are the gills as a route of excretion for
cortisol. To do this, information would be required on cortisol secretion and
metabolic clearance rates of individuals. Nevertheless, Donaldson & Fagerlund
(1968) have estimated the production rate of cortisol in sockeye salmon
Oncorhynchus nerka (Walbaum) to be between 07 and 102 ng g1 h1, which is
an order of magnitude higher than the present estimates of branchial release of
cortisol in rainbow trout [003–67 ng g1 h1; Figs 5(a) and 6(b)]. Previous
estimates of metabolic clearance rates of cortisol by rainbow trout range from
008 to 026 ml g1 h1 (Mommsen et al., 1999) which are again at least an order
of magnitude greater than the present estimates of branchial clearance rate (001
to 007 ml g1 h1), derived from the combination of branchial release rate
(ng g1 h1) and plasma cortisol concentration (ng ml1) (Fig. 6). With more
directed experiments, the validated methodology could contribute to the assessment of the relative importance of the gills for excretion of cortisol.
One question still to be answered is whether the clearance of cortisol via the
gills represents passive ‘leakage’ or active ‘excretion’. Patiño et al. (1985) suggested that the gills may play an active role in cortisol excretion because plasma
clearance rate of cortisol metabolites in coho salmon Oncorhynchus kisutch
(Walbaum) was correlated to gill NaþKþATPase activity, implying an energy
consuming process. Nevertheless, passive leakage is more likely because, for
lipophilic compounds such as free steroids, the gills behave like a permeable
membrane with two solutions (water and plasma) on each side; with the concentration gradient almost always favouring the passage of cortisol from the
plasma to water. Acting as a brake on the rate of diffusion is the ability of most
steroids to bind to plasma proteins. In fact, Vermeirssen & Scott (1996) have
speculated that the differential release of various free sex steroids into the water
by fishes may depend upon their affinity for, and the concentrations of, sexsteroid binding proteins in the plasma. Fishes do not appear to possess specific
corticosteroid-binding proteins in the plasma (Mommsen et al., 1999). If they
did so, cortisol and cortisone would probably be released in much smaller
amounts. Cortisol release may, however, vary according to the amounts of
non-specific steroid binding proteins in the plasma.
Vermeirssen & Scott (1996) showed that fishes were able not only to release
free steroids via the gills but also to take them up. In flow-through systems as
used in this study, this is likely to be of little significance due to the 1000-fold
difference in water and plasma cortisol concentrations. It is possible that
re-entry into the fish may become significant when cortisol accumulates in the
water in static or low-flow systems.
This work was funded by DEFRA, U.K. We gratefully acknowledge the assistance of
E. Vermeirssen and L. Greenwood during the early stages of this study.
References
Adams, S. M. & Breck, J. E. (1990). Bioenergetics. In Methods for Fish Biology
(Schreck, C. B. & Moyle, P. B., eds), pp. 389–415. Bethesda, MD: American
Fisheries Society.
#
2004 Crown copyright, Journal of Fish Biology 2004, 65, 1233–1252
1250
T. ELLIS ET AL.
Anderson, B. G. & Mitchum, D. L. (1974). Atlas of Trout Histology. Cheyenne, WY:
Wyoming Game & Fish Department.
Barton, B. A. (2000). Salmonid fish differ in their cortisol and glucose response to
handling and transport stress. North American Journal of Aquaculture 62, 12–18.
Barton, B. A., Schreck, C. B. & Sigismondi, L. A. (1986). Multiple acute disturbances
evoke cumulative physiological stress responses in juvenile chinook salmon.
Transactions of the American Fisheries Society 115, 245–251.
Bonga, S. E. W. (1997). The stress response in fish. Physiological Reviews 77, 591–625.
Buchanan, K. L. & Goldsmith, A. R. (2004). Non-invasive endocrine data for
behavioural studies: the importance of validation. Animal Behaviour 67, 183–185.
Donaldson, E. M. & Fagerlund, U. H. M. (1968). Changes in cortisol dynamics of
sockeye salmon (Oncorhynchus nerka) resulting from sexual maturation. General
and Comparative Endocrinology 11, 552–561.
Ellis, T., North, B., Scott, A. P., Bromage, N. R., Porter, M. & Gadd, D. (2002). The
relationships between stocking density and the welfare of farmed rainbow trout.
Journal of Fish Biology 61, 493–531. doi: 10.1006/jfbi.2002.2057
Gerwick, L., Demers, N. E. & Bayne, C. J. (1999). Modulation of stress hormones in
rainbow trout by means of anaesthesia, sensory deprivation and receptor blockade.
Comparative Biochemistry and Physiology 124A, 329–334.
Greenwood, L. N., Scott, A. P., Vermeirssen, E. L. M., Mylonas, C. C. & Pavlidis, M.
(2001). Plasma steroids in mature common dentex (Dentex dentex) stimulated with
a gonadotropin-releasing hormone agonist. General and Comparative Endocrinology
123, 1–12.
Idler, D. R. & Truscott, B. (1972). Corticosteroids in fish. In Steroids in Non-mammalian
Vertebrates (Idler, D. R., ed.), pp. 126–252. London: Academic Press.
Jeney, Z., Jeney, G. & Maule, A. G. (1992). Cortisol measurements in fish. In Techniques
in Fish Immunology (Stolen, J. S., Fletcher, T. C., Anderson, D. P., Kaattari, S. L.
& Rowley, A. F., eds), pp. 157–166. New Jersey: SOS Publications.
Jürgens, M. D., Holthaus, K. I. E., Johnson, A. C., Smith, J. J. L., Hetheridge, M. &
Williams, R. J. (2002). The potential for estradiol and ethinylestradiol degradation
in English rivers. Environmental Toxicology and Chemistry 21, 480–488.
Kebus, M. J., Collins, M. T., Brownfield, M. S., Amundson, C. H., Kayes, T. B. &
Malison, J. A. (1992). Effects of rearing density on the stress response and growth
of rainbow trout. Journal of Aquatic Animal Health 4, 1–6.
Koren, L., Mokady, O., Karaskov, T., Klein, J., Koren, G. & Geffen, E. (2002). A novel
method using hair for determining hormonal levels in wildlife. Animal Behaviour
63, 403–406.
van der Kraak, G., Sorensen, P. W., Stacey, N. E. & Dulka, J. G. (1989). Periovulatory
female goldfish release three potential pheromones: 17a,20b-dihydroxyprogesterone,
17a,20b-dihydroxyprogesterone glucuronide and 17a-hydroxyprogesterone. General
and Comparative Endocrinology 73, 452–457.
Laidley, C. W. & Leatherland, J. F. (1988). Cohort sampling, anaesthesia and stocking
density effects on plasma cortisol, thyroid hormone, metabolite and ion levels in
rainbow trout, Salmo gairdneri Richardson. Journal of Fish Biology 33, 73–88.
Landau, M. (1992). Introduction to Aquaculture. New York: John Wiley & Sons.
Mazeaud, M. M. & Mazeaud, F. (1981). Adrenergic response to stress in fish. In Stress
and Fish (Pickering, A. D., ed.), pp. 49–75. London: Academic Press.
Mommsen, T. P., Vijayan, M. M. & Monn, T. W. (1999). Cortisol in teleosts: dynamics,
mechanisms of action, and metabolic regulation. Reviews in Fish Biology and
Fisheries 9, 211–268.
Oliveira, R. F., Canario, A. V. M. & Bshary, R. (1999). Hormones, behaviour and
conservation of littoral fishes: current status and prospects for future research. In
Behaviour and Conservation of Littoral Fishes (Almada, V. C., Oliveira, R. F. &
Gonçalves, E. J., eds), pp. 149–178. Lisbon: Instituto Superior de Psicologia Aplicada.
Olsen, Y. A., Einarsdottir, I. E. & Nilssen, K. J. (1995). Metomidate anaesthesia in
Atlantic salmon, Salmo salar, prevents plasma cortisol increase during stress.
Aquaculture 134, 155–168.
#
2004 Crown copyright, Journal of Fish Biology 2004, 65, 1233–1252
A NON-INVASIVE STRESS ASSAY FOR FISHES
1251
Parrot, R. F. & Misson, B. H. (1989). Changes in pig salivary cortisol in response to
transport stimulation, food and water deprivation, and mixing. British Veterinary
Journal 145, 501–505.
Patiño, R., Schreck, C. B. & Redding, J. M. (1985). Clearance of corticosteroids during
smoltification of coho salmon, Oncorhynchus kisutch. Comparative Biochemistry
and Physiology 82A, 531–535.
Patiño, R., Redding, J. M. & Schreck, C. B. (1987). Interrenal secretion of corticosteroids
and plasma cortisol and cortisone concentrations after acute stress and during
seawater acclimation in juvenile coho salmon (Oncorhynchus kisutch). General and
Comparative Endocrinology 68, 431–439.
Pauly, D. (1994). A framework for latitudinal comparisons of flatfish recruitment.
Netherlands Journal of Sea Research 32, 107–118.
Pickering, A. D. (1992). Rainbow trout husbandry: management of the stress response.
Aquaculture 100, 125–139.
Pickering, A. D., Pottinger, T. G. & Christie, P. (1982). Recovery of the brown trout,
Salmo trutta L., from acute handling stress: a time course study. Journal of Fish
Biology 20, 229–244.
Pottinger, T. G. & Moran, T. A. (1993). Differences in plasma cortisol and cortisone
dynamics during stress in two strains of rainbow trout (Oncorhynchus mykiss).
Journal of Fish Biology 43, 121–130.
Pottinger, T. G., Moran, T. A. & Cranwell, P. A. (1992). The biliary accumulation of
corticosteroids in rainbow trout, Oncorhynchus mykiss, during acute and chronic
stress. Fish Physiology & Biochemistry 10, 55–66.
Pottinger, T. G., Carrick, T. R. & Yeomans, W. E. (2002). The three-spined stickleback as
an environmental sentinel: effects of stressors on whole-body physiological indices.
Journal of Fish Biology 61, 207–229. doi: 10.1006/jfbi.2002.2034
Ruane, N. M. & Komen, H. (2003). Measuring cortisol in the water as an indicator of
stress caused by increased loading density in common carp (Cyprinus carpio).
Aquaculture 218, 685–693.
Scott, A. P. & Sorensen, P. W. (1994). Time course of release of pheromonally active
gonadal steroids and their conjugates by ovulatory goldfish. General and Comparative
Endocrinology 96, 309–323.
Scott, A. P., Sheldrick, E. L. & Flint, A. P. F. (1982). Measurement of 17a,20bdihydroxy-4-pregnen-3-one in plasma of trout (Salmo gairdneri Richardson):
seasonal changes and response to salmon pituitary extract. General and Comparative
Endocrinology 46, 444–451.
Scott, A. P., Pinillos, M. & Ellis, T. (2001). Why measure fish steroids in plasma when you
can measure them in water. In Perspectives in Comparative Endocrinology: Unity
and Diversity (Goos, H. J. T., Rastogi, R. K., Vaudry, H. & Pierantoni, R., eds),
pp. 1291–1295. Bologna: Monduzzi Editore.
Sorensen, P. W. & Scott, A. P. (1994). The evolution of hormonal sex pheromones in
teleost fish: poor correlation between the pattern of steroid release by goldfish
and olfactory sensitivity suggests that these cues evolved as a result of chemical
spying rather than signal specialization. Acta Physiologica Scandinavica 152,
191–205.
Sorensen, P. W., Scott, A. P. & Kihslinger, R. L. (2000). How common hormonal
metabolites function as relatively specific pheromonal signals in the goldfish. In
Proceedings of the 6th International Symposium on Reproductive Physiology of Fish,
Bergen, Norway, July 4–9, 1999 (Norberg, B., Kjesbu, O. S., Taranger, G. L.,
Andersson, E. & Stefansson, S. O., eds), pp. 125–128. Bergen: University of Bergen.
Stacey, N. E., Sorensen, P. W., Van der Kraak, G. J. & Dulka, J. G. (1989). Direct
evidence that 17a,20b-dihydroxy-4-pregnen-3-one functions as a goldfish primer
pheromone: preovulatory release is closely associated with male endocrine
responses. General and Comparative Endocrinology 75, 62–70.
Truscott, B. (1981). An alternative method for the synthesis of 11-[3H]ketotestosterone
and 11b-[3H]hydroxytestosterone from [3H]cortisol. General and Comparative
Endocrinology 45, 409–411.
#
2004 Crown copyright, Journal of Fish Biology 2004, 65, 1233–1252
1252
T. ELLIS ET AL.
Turner, J. W., Nemeth, R. & Rogers, C. (2003). Measurement of fecal glucocorticoids in
parrotfishes to assess stress. General and Comparative Endocrinology 133, 341–352.
Vermeirssen, E. L. M. & Scott, A. P. (1996). Excretion of free and conjugated steroids in
rainbow trout (Oncorhynchus mykiss): evidence for branchial excretion of the
maturation-inducing steroid, 17,20b-dihydroxy-4-pregnen-3-one. General and
Comparative Endocrinology 101, 180–194.
Wasser, S. K., Hunt, K. E., Brown, J. L., Cooper, K., Crockett, C. M., Bechert, U.,
Millspaugh, J. J., Larson, S. & Monfort, S. L. (2000). A generalized fecal
glucocorticoid assay for use in a diverse array of nondomestic mammalian and
avian species. General and Comparative Endocrinology 120, 260–275.
Weisbart, M. & McGowan, L. K. (1984). Radioimmunoassay of cortisone in the adult
Atlantic salmon, Salmo salar L. General and Comparative Endocrinology 55,
429–436.
#
2004 Crown copyright, Journal of Fish Biology 2004, 65, 1233–1252