SYSTEMATIC SOURCES OF BIAS IN A BIOENERGETICS MODEL - EXAMPLES FOR AGE-0 STRIPED BASS

Laboratory experiments were conducted to measure prey consumption and metabolism of age-0 striped bass Morone saxatilis for a bioenergetics model for this species. In many bioenergetics models, resting metaboli sm is multiplied by a factor of one to two to account for energy expen ded in active metabolism. We estimated this activity multiplier (ACT) indirectly from laboratory experiments and found it to average 2.47. W e validated the bioenergetics model in the laboratory and found estima tes of growth and consumption to be within 6.1% of measured values at 20-degrees-C, but growth was underestimated and consumption overestima ted by as much as 33% at 6.9-degrees-C. The bioenergetics model was us ed to evaluate the influence of parameter uncertainty upon model resul ts. In general, parameters with the greatest uncertainty (largest 95% confidence interval relative to the mean) tended to be least sensitive to errors, whereas those measured with greater precision were more se nsitive. Changing the value of ACT from the mean to the 95% confidence limits had little effect on consumption estimates, but uncertainty in ACT could mean a 1.3-degrees-C range in the uncertainty of the upper temperature that supports positive growth. If prey type and prey densi ty influence parameter estimates, confidence limits of model parameter s may be larger than the levels of variation normally used in model er ror analyses. Estimates of maximum consumption (C(max)) for striped ba ss fed live age-0 spot Leiostomus xanthurus, live polychaetes Nereis s pp., or thawed bay anchovies Anchoa mitchilli were not significantly d ifferent; however, when we used mean values to estimate the intercepts for C(max) functions derived from the different prey types, the estim ates did differ. Predator density had a positive effect on C(max) of a ge-0 striped bass. Fish at densities greater than four per tank (0.85 m3 volume) had significantly higher maximum consumption rates than one or two fish per tank. We conclude that bioenergetics modelers must pa y particular attention to the methods used to generate data for model development and attempt to match methods to the particular application of the model.