FISHERY BULLETIN: VOL. 84, NO. 4 



(/> 

 </l 

 O 



e 



g 



"E 



k. 



CD 

 Q. 



C 



o 



V- 

 Q. 



£ 



Z3 



in 



c 

 o 

 o 



o 

 Q 



Wa 



Body weight (g) 



Figure 4.— Size-specific estimates of food consumption per unit biomass in Epinephelus guttatus, 

 as obtained by integrating Equation (22) over narrow ranges of weight, then plotting the resulting 

 Q/B estimates against the midranges of the weights. Note definition of maintenance ration as 

 "Q/B at W m ". 



tenance ration of 0.5% body weight per day is ob- 

 tained (Fig. 4), while the corresponding value for 

 H. bermudensis is 1.73%. 



Using the computed output of Equation (22) one 

 can also obtain an estimate of population trophic ef- 

 ficiency (E T ) from 



E T = Z  (B/Q) 



(25) 



which expresses production per unit food consumed, 

 production being expressed here as total mortality 

 (i.e., production/biomass ratio) times biomass (Allen 

 1971). 



For E. guttatus, the estimated value of trophic 

 efficiency is E T = 0.23, i.e., slightly less than one 

 quarter of the fish food eaten by a population of 

 E. guttatus is turned into production. The cor- 

 responding value for H. bermudensis is E T = 

 0.08, which is low, as should be expected in an 

 herbivore. 



ACCOUNTING FOR 

 MULTIFACTOR EFFECTS ON K x 



Experimental data allowing for the estimation of 

 values of W m and p corresponding exactly to those 

 to be expected in nature cannot be obtained, since 

 no experimental design can account for all the en- 

 vironmental factors likely to affect the food conver- 



sion of fishes in nature. Among the factors which 

 can be experimentally accounted for are 



1) ration size (Paloheimo and Dickie 1966; but see 

 Condrey 1982), 



2) type of food (see below), 



3) temperature (Menzel 1958, Taylor 1958, Kinne 

 1960, and see below), 



4) salinity (Kinne 1960). 



Also, "internal states" affecting food conversion 

 efficiency, such as the sex of the fish, previous ther- 

 mal history, and stress undergone during an experi- 

 ment, can be accounted for given a suitable 

 experimental design. 



One method of incorporating some of these fac- 

 tors into a linear form of the basic model (Equation 

 (5)) is to extend the model into a multiple regres- 

 sion of the form 



C = a- H log 10 W + b,V, + b 2 V 2 . . . b n V n (26) 



in which V lt V 2 , and V n are factors which affect C 

 (= -log 10 (1 - K{)) after the effect of weight on C 

 has been accounted for. 

 For example, 



C = 0.363 - 0.0419W - 0.0116T 



+ 0.0156S + 0.0488M 



(27) 



834 



