106 LEFFLER 



(1975) hypothesized that resistance would increase and resihence 

 would decrease. They believed that resistance is a function of the 

 mass of the system and resilience is related to functional dynamics, 

 especially turnover rates. Thus resistance and resilience might be 

 considered inverse concepts. Other investigators have discussed the 

 effect of environmental constancy on ecosystem stability. Systems 

 adapted to less variable environments often exhibit high levels of 

 constancy stability but very poor resistance and resilience in the 

 event of a major perturbation (Copeland, 1970; Jernelov and 

 Rosenberg, 1976; Larsen, 1974). Estuaries, on the other hand, 

 typically exhibit low constancy stability because of their highly 

 variable environment but high levels of resistance and resilience 

 (Boesch, 1974). 



THE MICROCOSM APPROACH TO ECOSYSTEM THEORY 



Ecologists have had difficulty in testing the validity of hypothe- 

 ses that describe an ecosystem's response to stress. Most of these 

 propositions were derived either by an intuitive synthesis of field 

 observations or by mathematical simulation. Rigorous experimental 

 evaluation is rare because of problems typical of all ecosystem-level 

 studies, e.g., long time scales, lack of replication, difficulty of 

 ecosystem parameter measurement, lack of control over extraneous 

 environmental conditions and system histories, and economic feasi- 

 bility. One means of testing the generality of a hypothesis is to use 

 microcosms as living analogs of ecosystems. I have used several types 

 of aquatic microecosystems to evaluate hypotheses regarding ecosys- 

 tem stress responses. Microcosms can be designed to meet all 

 requirements of standard ecosystem definitions (Odum, 1971) and to 

 overcome the examples cited of experimental difficulties in field 

 studies. By designing microcosm systems in which extraneous 

 variables can be controlled, we can focus on the specific concept to 

 be evciluated. No effort is made to duplicate any specific natural 

 ecosystem; the microcosms are merely models of general ecosystem 

 properties, such as energy flow through a trophic structure, nutrient 

 cycling, and species diversity. In addition, the hypothetical factors 

 influencing stability can be modeled by subjecting microcosms to 

 different treatment levels of a factor, e.g., nutrient availabihty. 

 Ecosystems developed under each treatment level can be subjected to 

 identical stresses, and their respective responses can be compared. 



The goal of such experiments is not to extrapolate from 

 microcosm results to all ecosystems but to determine whether a 

 hypothesis derived by nonempirical methods is empirically falsifi- 



