ECOSYSTEM RESPONSES TO STRESS 117 



certain parameters. It is perhaps worth noting that the total relative 

 stability of all three metabolic parameters was less for the second 

 stress than for the first. 



For each of the nine treatments, median coefficients of variation 

 among replicate microcosms were also compared before and after 

 perturbations. The binomial test again indicated significant 

 (P< 0.100) trends. The thermal stress increased replicate variation 

 for net day production but reduced variation for the chlorophyll a, 

 phosphorus, and magnesium parameters. Although no general trends 

 are obvious from these studies, we must certainly consider the 

 influence of prior events (i.e., history) when assessing an ecosystem's 

 stress response. 



SUMMARY 



Most of the current theory that seeks to describe and predict the 

 response of ecosystems to stress conditions has been developed by 

 mathematical or intuitive reasoning. I attempted to clarify the stress 

 concept and advocated an empirical approach complementing these 

 methodologies. This approach requires that strict operational defini- 

 tions be developed to clarify theoretical concepts. Relations between 

 concepts must be expressed as falsifiable hypotheses. Five measures 

 of relative stability can be used to quantify a system's response to 

 stress. Microcosms are proposed as living models of large natural 

 ecosystems. Since in any type of model (mathematical, verbal, or 

 physical) some behavior is specific to the model, results must be 

 extrapolated cautiously. Microcosms are potentially valuable tools, 

 however, both conceptually and pragmatically, for developing and 

 testing theories of ecosystem dynamics. 



The two microcosm experiments briefly described to illustrate 

 how such models can be used failed to support several hypotheses 

 that were derived by mathematical or intuitive approaches. The 

 relative stability of these experimental systems was not related 

 simply to their species diversity, nutrient— energy subsidies, or mass. 

 Constancy stability, however, was directly related to nutrient 

 availability. No relationship could be demonstrated between resis- 

 tance and resilience of stressed microcosms or between their 

 steady-state variability and any measure of their perturbed behavior. 

 These empirically derived results suggest that some hypotheses that 

 seek to describe an ecosystem's stress response are not correct as 

 generalities and require further clarification. 



Several factors influence an ecosystem's response to stress. 

 Among these are the type of perturbation, the parameters used to 



