Co x et al.: Measurements of resistance and reactance in fish with the use of bioelectricai impedance analysis 
43 
Figure 7 
Comparisons of estimates of water mass parameters 
with resistance errors ranging from -58% to 10% (A) 
and with resistance errors ranging from -3% to 3% (B). 
The data were plotted in three-dimensional matrix plots 
with the x and y axes describing the range of values 
for either the length between detectors, R , X c , or R in 
parallel, and with the z axis depicting the difference 
(as a percentage) between predicted estimates, with 
and without errors. 
Figure 8 
Comparisons of estimates of dry mass parameters with 
reactance in parallel [■X c + (7? 2 /A f )] errors ranging from 
-58% to 173% (A) and -11% to 4% (B), and length errors 
in both ranging from 0% to 5%. The data were plotted 
in three-dimensional matrix plots with the x and y axes 
describing the range of values for reactance in parallel 
and the length between detectors, and 2 axis depicts the 
difference (as a percentage) between predicted estimates 
of dry mass, with and without errors. 
0% to 21% (Fig. 8B). If length errors were not included, 
parameter estimations of DW were inaccurate by -3% 
to 11%. The addition of length errors compounded the 
parameter estimation errors. 
Individual errors of R and X c errors affected phase 
angle measures, but combined errors tended to offset 
one another (Fig. 9, A and B). The introduction of sig- 
nificant errors of R and X c (-58% to 10%, and -37% to 
45%, respectively) caused phase angle measurements to 
vary from -60% to 129%. When inserted errors were in 
the same direction (i.e., both errors are either negative 
or non-negative numbers), they offset one another and 
resulting phase angle errors were closer to 0% (Fig. 9A). 
When inserted errors were opposite of one another (i.e., 
when one error was a positive number and the other 
negative), phase angle errors increased and decreased 
to their maximum values and did not offset one another, 
but rather increased errors. Identical errors for both R 
and X c showed symmetry in that phase angle errors 
equaled 0% when inserted R and X c errors were the 
same (Fig. 9B). Reactance errors by themselves were 
inversely correlated with phase angle errors and R er- 
rors were positively correlated with phase angle values. 
Discussion 
The ability to accurately estimate physiological param- 
eters including proximate composition, condition, and 
energy content with BIA will permit increased preci- 
