Lantry et al.: Electrical conductivity to estimate water content of Perca flavescens and Alosa pseudoharengus 



75 



25 30 



Actual water content (g) 



Figure 1 



Comparison of the predicted and measured values of water content (g) of 

 yellow perch and alewife. Predicted values are from multiple linear regres- 

 sions with wet weight and total body electrical conductivity (TOBEC) as 

 mdependent variables and from simple linear regressions with wet weight as 

 the only independent variable. 



The equation for alewife was 



WC = -0.17206 + 0.8471 x WWT - 0.0439 x TOBEC 



(r2=0.990). 



The slope coefficient for wet weight and the inter- 

 cept in the yellow perch regression model were both 

 significant (P<0.01), whereas the TOBEC coefficient 

 was not (P=0.529). The slope coefficient for wet 

 weight in the alewife regression was significant 

 (P<0.0001), whereas both the TOBEC coefficient and 

 the intercept were not (P=0.295 and 0.679, respec- 

 tively). Both the yellow perch and alewife predictive 

 equations provided good fit to the data for water con- 

 tent with mean PEs fi-om all values in each data set of 



3.33% and 1.67%, respectively. Wet weight, however, 

 accounted for most of the variance within each rela- 

 tionship. When wet weight was used as the only inde- 

 pendent variable in each regression model, r^ values 

 were both about 0.99, and were greater than the r^ 

 values from regressions that used only TOBEC as the 

 independent variable. Predictions fi"om these simple 

 linear regressions ( WWT as the only independent vari- 

 able) provided nearly as good a fit to the data as the 

 multiple linear regressions (r^ values decreased only 

 by about 0.00002 to 0.0005, Fig. 1). 



Total length and TOBEC were not linearly related 

 to water content for yellow perch. Natural log trans- 

 formations for total length ihnTL) and square root 

 transformations of TOBEC values (sqTOBEC) re- 



