Evaluation of ecological network analysis for ecosystem-based management

A critical assessment of ecological network analysis (ENA), a modeling technique increasingly being used to address resource management issues, was conducted. The major objectives were to evaluate the effectiveness of ENA in detecting differences in food web properties, and to validate ENA models using independent methods.; Quantitative trophic networks (n=12) representing four high marsh ponds during three times (corresponding to low stress, high stress, and post-disturbance) were constructed from an extensive field sampling program augmented by literature values. A null hypothesis was tested to determine how values of twelve indices from ENA output differed among the three stress/disturbance conditions (H o: Low Stress = High Stress = Post-Disturbance). Statistical differences were determined using repeated measures ANOVA (with contrasts) and Friedman's Tests (with multiple comparisons). Covariance of each pair of indices was evaluated with Spearman's Rank Correlation Tests.; Hypothesis testing suggested ENA was effective in detecting differences in the food web properties examined. ANOVA results indicated mean values of 10 of 12 ENA indices were significantly different among the three stress/disturbance conditions, and results from the Friedman's Test were generally in agreement (mean rankings in 11 of 12 indices showed significant differences). Confidence in these results was given by a relatively low amount of covariance among the indices (7 of 66 were significant).; Four separate aspects of selected models were then validated (respiration, aggregation of taxa, trophic levels, extended diet) by comparison to results derived from field measurements and stable isotope data (delta13 C, delta15N, delta34S). Validation was based on paired t-tests and graphical comparisons, and explanations were made for instances where there was not agreement. These involved assumptions associated with the models and inherent differences in the methods used for validation relative to ENA.; Implications of the validation were then examined by modifying selected models to be in agreement with the validation methods, and comparing their output to the unmodified version. Findings from this analysis support the hypothesis testing, and suggest ENA output was sensitive to changes in the amount of material available for energy flow as well as structural aspects of that flow.