Cox et al.: Measurements of resistance and reactance in fish with the use of bioelectrical impedance analysis 
35 
of a substance is proportional to the voltage of an ap- 
plied current as it passes through a substance, or R = 
V/C, where V is applied voltage (volts), and C is current 
(amps). When the current is low enough, the current 
does not pass through the cell membrane (owing to the 
nonconductive lipid bilayer sandwiched between two 
conductive protein layers). The low current allows R 
to be reflective of everything extracellular. Reactance 
is the opposition to alternating current by a capacitor 
(cell membranes), and can be mathematically expressed 
by the following equation: X c =l/(2nfC), where f is fre- 
quency in Hertz, and C is capacitance in Farads (Keller 
et ah, 1993). Higher current frequencies will cause cell 
membranes to become capacitive so that X c becomes 
reflective of the total amount of cell membrane material 
within the current. Both values are thus related to the 
cross sectional area of the entire fish, conductor length 
of the organism, and the signal frequency of the current 
(Lukaski, 1987). The phase angle is the ratio of R to 
X c of tissue and has been found to be sensitive to the 
health and condition of fish (Cox and Heintz, 2009). By 
using R and X c , one can estimate the composition and 
condition of fish. 
In order for BIA to be accepted in fisheries science, it 
is necessary first to identify sources of error, and then to 
use that knowledge to minimize errors. With the use of 
BIA in studies of human nutrition and body composition, 
the identification of error sources was used to establish 
protocols that minimized errors (Rallison et al., 1993). 
More specifically, predictions of proximate composition 
parameters with BIA were found to be accurate with es- 
tablished procedures, but without them, these estimates 
became inaccurate (Ursula et al., 2004). In previous fish 
research, BIA protocols were established to minimize 
any unforeseen sources of error (Cox and Hartman, 
2005). Although protocols were established and error 
may have been minimized, the actual sources of error 
were not identified. More recently, studies with BIA 
methods have shown inconsonant results. For example, 
in a study of cobia ( Rathycentron canadum ), there was a 
high correlation between BIA and most cobia proximate 
composition values (Duncan et ah, 2007); whereas in 
another study involving yellow perch (Perea flavescens), 
walleye (Sander vitreus ), and lake whitefish (Coregonus 
clupeaformis), it was concluded that considerable work 
needs to be completed before BIA can provide reliable 
predictions of whole body energy and percent lipid con- 
tent (Pothoven et al., 2008). Furthermore, it was indi- 
cated that there needs to be an understanding of how 
temperature, locations where the electrode needle is 
placed (possible sources of error), and lipid distribution 
within a fish affect BIA measures. 
The objective of our study was to identify sources of 
error in measurements of fish with BIA and errors of R 
and X c . The cumulative effects of both significant and 
nonsignificant errors were examined through sensitiv- 
ity analysis modeling. We conclude by identifying a pro- 
tocol that will minimize the sources of error and maxi- 
mize the potential of BIA in providing measures of body 
composition and condition in the field and laboratory. 
Methods 
We conducted laboratory experiments to identify sources 
of errors within BIA measurements of R and X c . Spe- 
cifically, we considered how electrode needle location, 
procedure deviation, user training, time after death, 
temperature of the fish, and stomach fullness affected 
measurements of R and X c . For a comparison, we used 
a reference (control) that followed the protocol outlined 
by Cox and Hartman (2005). For all experiments, a 
handheld Quantum X impedance analyzer (RJL Sys- 
tems, Point Heron, MI) was used, except for temperature 
measurements, for which a desktop Quantum II analyzer 
was used. In either case, a fixed current at 800 pA, AC, 
and 50 kHz was used. Electrode needles were either 
“standard” 12 mmx28 gauge subdermal stainless steel 
disposable low-profile EEG needle electrodes (Grass 
Technologies, West Warwick, RI) as used in Cox and 
Hartman (2005), or “nonstandard” 38 mmxl4 gauge 
standard hypodermic needles with a polypropylene hub. 
Brook trout (Salvelinus fontinalis) used in this study 
were obtained from the Bowden West Virginia State Fish 
Hatchery, Bowden, WV, and Chinook ( Oncorhynchus 
tshawytscha), pink (O. gorbuscha), and coho (O. kisutch ) 
salmon were obtained from the Sheldon Jackson College 
(SJC) salmon hatchery, Sheldon Jackson College, Sitka, 
AK. Approximately 100 brook trout were maintained 
in a living stream tank at West Virginia University at 
15°C and fed standard hatchery pellets at a rate of 3% 
body weight per day until used in experiments. Juvenile 
Chinook, pink, and coho salmon used in this study were 
taken from the SJC hatchery round pens and adult 
salmon used in this study were selected from returning 
adults. Treatment methods are those described below for 
each particular experiment. Fish that were sacrificed 
were killed by a blow to the head. In all experiments, 
sample size was determined by iterative power analysis 
with a significance of 0.05 and a power of 0.96 (Zar, 
1996). In cases where variances of sample sets were not 
available from previous data, sample data were collected 
for power analysis before testing. 
Linear mixed-effects (LME) models were used to test 
for the effects of electrode needle location, procedure de- 
viation, user experience, and time on R and X c measures 
(Pinheiro and Bates, 2000). The effects of temperature 
and gut fullness were tested with regression analysis 
to test for differences in slopes. In each experiment, 
measured R and X c values were compared between 
treatment and controls. Statistical tests on response 
measures were performed by using program R, vers, 
2.4.1 (R Development Core Team, Vienna, Austria). 
Significance (a) was set at 0.05. 
Location of the electrode needle 
To determine if different electrode needle locations 
influence impedance, comparisions of R and X c measure- 
ments were made between different electrode locations 
within an individual fish (Fig. 1A). A location refers to 
the simultaneous location of all four (tetrapolar) elec- 
