42 
Fishery Bulletin 109(1 ) 
500 
4 6 8 10 
Temperature (°C) 
Figure 5 
Influence of temperature on repeated measure- 
ments of resistance (A) and reactance (B) on dead 
Chinook ( Oncorhynchus tshawytsclia ) (n = 2) and 
coho salmon (O. kisutch) (n = 3). All fish except 
the juvenile fish were measured from warm to 
cold temperatures. Solid circle=juvenile fish 
that was measured from cold to warm, and open 
circle = adult salmon that were measured from 
warm to cold. 
0% to 9%. Compounding both maximum R (-3%) and 
length (5%) errors resulted in an overestimation >12%. 
Both significant and nonsignificant errors inserted 
into the derived electrical volume equation X c in parallel 
caused inaccuracies in subsequent parameter estima- 
tions of dry mass (Fig. 8, A and B). The insertion of 
significant errors impacted parameter estimations more 
substantially than nonsignificant ones. The addition of 
significant R errors (-58% to 10%) and X c errors (-35% 
to 47%) into X c in parallel (nonvolumetric) resulted in 
errors ranging from -58% to 173%. The subsequent 
addition of this range of errors into predictive models 
of DW caused estimations to be inaccurate by -45% to 
349%, and length errors compounded the error (Fig. 
800 - 
700 - 
600 - 
500 - 
400 - 
A 
▲ 
Half 
A 
Whole 
A A 
1 1 1 1 — 
~ Full Empty Full Empty 
O 
Figure 6 
Boxplots describing resistance (A) and reactance 
(B) measurements in four groups of live brook 
trout (Saluelinus fontinalis ) for different electrode 
locations and stomach fullness (n = 5 per group), 
where full=full stomach, empty= empty stomach. 
“Whole” refers to a whole-body length (the entire 
length of the fish as described in Cox and Hart- 
man [2005]). “Half “refers to a half-body length 
(one set of electrodes was placed towards the 
head region and the second being placed around 
mid-point [under the dorsal fin] ). Open circles 
(O) represent outliers determined by a Grubbs 
test. Closed circles (•) represent mean values. 
Other symbols indicate differences as determined 
by the applied statistical tests. 
8A). The addition of nonsignificant R errors (-3% to 
3%) and X c errors (0% to 9%) into X c in parallel (non- 
volumetric) resulted in X c in parallel errors ranging 
from -11% to 4%. The subsequent addition of these 
errors plus the length errors (0% to 5%) into the volu- 
metric equation to predict dry weight (DW) resulted in 
parameter estimations of DW that were inaccurate by 
