Sudekum et al.: Life history and ecology of Caranx ignobilis and Caranx melampygus 



509 



Measured ration of fish feeding ad libitum while 

 swimming actively in a relatively undisturbed captive 

 environment was used as an estimate of the natural, 

 long-term, average ration offish in the wild. If feeding 

 were not reduced by trauma or behavioral effects of 

 captivity, such feeding should provide an estimate of 

 maximum natural ration, given superabundant, easily 

 accessible food. Since the measured growth rate in cap- 

 tivity closely approximated the growth rate estimated 

 for wild fish by the von Bertalanffy model, it may be 

 that the effects of captivity depressed the ad libitum 

 feeding activity from the maximum rate, to approach 

 the natural, long-term, average rate. In any case, there 

 is no clear direction of bias and no obvious way to im- 

 prove the estimate of natural ration. 



For the parameters k and y, there seem to be no data 

 specifically for these jacks or for closely related species. 

 Published values for a considerable number of rather 

 large, active, perciform fishes are similar to those 

 chosen in this study (about 0.8 for both parameters) 

 (e.g., Winberg 1956). The value of k chosen affects the 

 entire calculation of captive a and subsequently of C 

 for all weight-classes. The value of y affects the entire 

 calculation of captive a, but affects only one term (the 

 largest) in the subsequent calculation of C. For each 

 of these parameters, because it occurs in the calcula- 

 tion of captive a and subsequently of C, the effects of 

 any inaccuracy in the choice of the value tend to cancel. 



Use of a single value of a for all sizes of fish intro- 

 duces another assumption. In previous work, this coef- 

 ficient has often been treated as constant over a con- 

 siderable range of sizes (Winberg 1956). However, for 

 most species, a has not been determined for large 

 sample sizes or large size ranges. A rather extreme 

 range of sizes is included in the present calculations. 

 No adequate measurements were available to produce 

 a direct estimate of a for C. ignobilis. The use in its 

 place of the a calculated for C. melampygus represents 

 an approximation, and thus the estimated components 

 of the energy budget for C. ignobilis may be less 

 reliable. However, the similarity of a values presented 

 by Winberg (1956) for a wide variety of less closely 

 related species suggests that the approximation used 

 is acceptable. Temperatures in the seawater feeding 

 tanks were close to normal sea temperatures through- 

 out the year, and therefore no correction of a for 

 temperature seemed necessary. (The full annual range 

 of tank and sea temperatures was small.) 



The quality of the estimate of growth rate depends 

 upon the quality of the fitted von Bertalanffy model 

 as a descriptor of actual growth. Both regressions were 

 good fits, and t and L^ values were realistic. The use 

 of dW/dt, evaluated for each chosen weight-class, is 

 probably the best estimate of growth rate available 

 from the von Bertalanffy model. 



The estimate of reproductive output is crude, but 

 little else is available. Since whole body weights were 

 used in the von Bertalanffy model regressions, at least 

 part of the reproductive energy may be viewed as in- 

 cluded in the growth rate term, G (on some sort of 

 average basis across all specimens and seasons). How- 

 ever, since the development of reproductive products 

 is seasonal, it seemed reasonable to add a term repre- 

 senting the weight of a fully developed gonad to repre- 

 sent additional annual reproductive energy demand. 

 The effects from both these terms may somewhat 

 overestimate reproductive demand. However, multiple 

 clutches may be developed over the course of a year, 

 so the gonad weight alone may underestimate the 

 actual reproductive energy. 



As in all such energy budgets for active animals 

 beyond the early juvenile stages, the respiratory 

 metabolism term, Q, was dominant (see Table 5). The 

 growth term, G, was never more than 10% of Q for 

 the size-classes calculated here, and the reproductive 

 term, S, was always substantially less than G. There- 

 fore, factors affecting the accuracy of the components 

 of Q (e.g., assumptions about the parameters a and y) 

 produce the greatest effects on the total estimate of 

 the ration, C. 



Components of the basic energy budget derived in 

 this study, e.g., growth parameters (L^, W M , K, and 

 t ), maximum GSI, and a, should be similar in other 

 tropical and subtropical localities where these species 

 occur. The value of a (and corresponding metabolism 

 and ration) can be readily adjusted for other temper- 

 atures by using Krogh's (1959) "normal curve" or cor- 

 responding mathematical functions (e.g., Winberg 

 1956:32, Ursin 1967:2396). 



The estimate of population consumption in the spe- 

 cific case of FFS was of particular interest because of 

 the extensive community trophic work done there. A 

 mathematical trophic model of the shallow marine 

 ecosystem there has been produced (Polovina 1984), 

 with the biota lumped into a dozen large trophic com- 

 partments. Jacks (especially these two Caranx species) 

 are major members of one of the top trophic-level 

 compartments. Comparison of our numbers for con- 

 sumption with those from the model suggests that 

 these jacks provide an even more important trophic 

 path than the large-scale model predicts. 



The census methodology used to produce the popula- 

 tion estimates in Table 6 was not designed to adequate- 

 ly census such wide-ranging, highly mobile species, and 

 these estimates must be considered rough. They prob- 

 ably represent maximum estimates. No method was 

 found to check them directly. Extrapolation of the 

 catch data for the year 1900 from the main Hawaiian 

 islands produced an estimated "catch" for FFS of 

 about 44 C. melampygusfkm 2 and 26 C. ignobilislkm 2 . 



