ZOOPLANKTON COHORT ANALYSIS USING SYSTEMS IDENTIFICATION TECHNIQUES

The linear-transfer and lag-Manly models of zooplankton cohort develop ment were examined using data generated from a third more realistic mo del. The more realistic multi-transfer model included variance in deve lopment rate among individuals. The linear-transfer model produced hig hly biased estimates of development rate under conditions of rapidly c hanging recruitment. Although its performance was improved by increasi ng the number of modeled stages and thus decreasing the rate of change in recruitment compared to stage duration, a positive bias remained. The lag-Manly model also produced positively biased estimates of stage duration given non zero variance in development rates. A comparison o f the models' performances under different simulated sampling regimes recommended the multi-transfer model. Use of the multi-transfer model was illustrated by determining the development and mortality rates of the brine shrimp, Artemia monica reared under three different conditio ns of food and temperature corresponding to natural regimes in Mono La ke, California. The experimental conditions and sampling regime result ed in high relative standard errors (mean, 33%) in stage abundance est imates not atypical of zooplankton sampling regimes in lakes. A Monte Carlo analysis was used to determine the uncertainty in estimated para meters and determine the level of stage aggregation which maximized th e amount of information derived from the experiments.