72 



Fishery Bulletin 97(1), 1999 



In our analyses offish community dynamics in Lake 

 Ontario and Oneida Lake, New York, we used bioener- 

 getics to evaluate trophic transfer and tissue growth 

 (Rand et al., 1995; Lantry, 1997). To model trophic en- 

 ergy flux accurately, we needed to estimate fish energy 

 content throughout the years being simulated. Because 

 of the difficulty encountered in drying large numbers 

 of fish and in drying large individuals, we sought an 

 alternative method — measurement of whole-body wa- 

 ter content. Our objective here was to evaluate the use 

 of TOBEC as an alternative to drying, for estimating 

 the whole-body water content of yellow perch (Perca 

 flavescens), an important prey and sport fish compo- 

 nent of both lakes, and alewife (Alosa pseudoharengus ), 

 the dominant planktivore and prey fish in Lake Ontario. 



Materials and methods 



We used the "EM-SCAN Inc. SA2 Small Research Ani- 

 mal Body Composition Analyzer" to obtain TOBEC 

 values for yellow perch and alewife. The measure- 

 ment principle of the EM-SCAN has been published 

 elsewhere (Fiorotto et al., 1987; Walsberg, 1988; 

 Brown et al., 1993). The scanner uses a radiating 

 coil to set up a low-frequency electromagnetic field 

 to measure the electrical conductivity of an animal. 

 Because electricity is conducted by the ions dissolved 

 in body water, the most direct relationship that can 

 be drawn fi-om the TOBEC values is the amount of 

 water contained within an animal. By initially mea- 

 suring the wet weight of a fish and then measuring the 

 whole-body water content with the scanner, we could 

 obtain the dry weight of the fish by difference, and cal- 

 culate the percentage dry weight, which is the key pa- 

 rameter we needed to estimate energy density. 



Sample collection and processing 



We collected 43 yellow perch in bottom trawls from 

 Oneida Lake, NY, over four dates in 1992: 23 April, n = 

 10; 9 June, n=5; 30 September, /; = 10; and 23 Novem- 

 ber, « = 18. We collected 47 alewife from the New York 

 waters of Lake Ontario with a 3-m bag seine on 2 May 

 (Ai = 13) and 4-5 July (?!=34) 1993. Yellow perch were 

 kept alive in lake water during sampling, and alewife 

 were placed on ice immediately after capture. All fish 

 were fi"ozen in water upon arrival at the laboratory. 



During processing, fish were thawed under warm 

 running water, blotted dry, weighed to the nearest 

 0.1 g and measured for total length (mm). We ob- 

 tained TOBEC values according to the procedures 

 outlined in the scanner manual (EM-SCAN Inc., 

 1991). Because the positioning of the animal is im- 

 portant in obtaining consistent conductivity readings, 



fish were placed on their right sides headfirst on the 

 animal carrier trays with the portion of their bodies 

 anterior to the distal end of the shortest ray of the 

 pelvic fin lined up in front of the scribed mark on the 

 tray. Each fish was scanned five times in the 

 scanner's fixed mode and the readings were aver- 

 aged to produce a mean TOBEC value. After scan- 

 ning, all fish were dried to a constant weight at 65°C. 



Statistical analysis 



We constructed simple and multiple linear regres- 

 sion models to predict whole-body water content (WC, 

 g). Following earlier fish studies (Brown et al., 1993; 

 Bai et al., 1994; Jaramillo et al., 1994) we included 

 wet weight (WWT, g), total length (TL, mm), and 

 TOBEC as independent variables in regressions. We 

 graphed these variables against fish WC (determined 

 from drying individuals) to evaluate the shape of the 

 relationships. We transformed independent variables 

 when relationships with WC departed from linear 

 trends. We used SYSTAT 5.03 for Windows ( 1993) to 

 perform stepwise multiple linear regressions with 

 backwards elimination on all variables (inclusion 

 probability: P>0.15) to determine which independent 

 variables accounted for most of the variation in pre- 

 dictive equations (Zar, 1984; Neter et al., 1985). The 

 resulting regression models were then used to pre- 

 dict WC and calculate percentage error (PE): 



PE = {(|actual - predicted] / actual) x lOO}. 



Finally, we calculated percentage dry weight from 

 both actual (determined from drying) and predicted 

 (from regression equations) water content and com- 

 pared the variation between these values to the range 

 of percentage dry weights commonly observed for 

 yellow perch and alewife (Tables 1 and 2). 



Results 



Significant (P<0.05 ) positive relationships were found 

 in simple linear regressions between the dependent 

 variable WC and the independent variables TOBEC, 

 WWT, and TL for both yellow perch and alewife 

 (r-^=0.66 to 0.99). Simple linear regressions oi TOBEC 

 on WC produced r'^ values of 0.933 and 0.667 for yel- 

 low perch and alewife, respectively. Multiple linear 

 regressions with two independent variables (WWT 

 and TOBEC) gave excellent fits to the data for both 

 species. The equation for yellow perch was 



WC = 3.46239 + 0.69844 x WWT + 0.00559 x TOBEC 



(r-=0.998). 



