104 LEFFLER 



case of an ecosystem in a nonoscillating steady state. This range can 

 be calculated as confidence bounds about the mean of all prepertur- 

 bation observations, assuming the system is in steady state. The 

 width of the region or the coefficient of variation of the observations 

 can be regarded as a measure of the relative constancy of the 

 ecosystem. 



Description of an ecosystem's normal operating range provides a 

 standard for judging whether a particular input should be considered 

 a stress. The criterion is whether the input causes the system's 

 trajectory to vary significantly from its normal operating range. Since 

 both the normal operating range and the system trajectory can be 

 described statistically, desired confidence levels can be placed on 

 such deviations. On the basis of this discussion and analysis, it is 

 obvious that the same quahtative input may or may not be a stress. 

 Temperature, for example, is an inevitable input to all ecosystems. A 

 normal operating range can be defined under conditions of tempera- 

 ture fluctuations that are typical for the system. Temperature can 

 become a stress, however, if the quantitative level of this input 

 assumes a value that causes system response to exceed the normal 

 operating bounds. 



The description of the normal operating range is also extremely 

 observer dependent. The width of such a region is a function of the 

 characteristics used to describe the ecosystem and of the ability of 

 the observer to discriminate differences in these parameters. For 

 example, an algal culture can be classified on the basis of "greenness" 

 or in terms of milligrams of chlorophyll per liter, depending on the 

 observer. Of perhaps greater importance is the choice of a time scale 

 for describing the normal operating range. Since the rate of change of 

 dissolved oxygen in aquatic systems typically displays diel variation, 

 an hourly description gives an oscillating normal operating range and 

 permits a dynamic analysis. In this case, an input can become a stress 

 on the system only at certain times in the daily cycle. Depending on 

 the observer's objective, the same system can be described on a daily 

 time scale. This gives a more static and much broader normal 

 operating range at all time points. Such a concept is flexible enough 

 to be applied to ecosystems undergoing oscillations, seasonal 

 changes, developmental sequences, or even cychc succession. Simi- 

 larly, the width of the normal operating range need not remain 

 constant at all times. It is plausible, for example, to believe that early 

 serai stages possess broader operating ranges than climax systems. 



It becomes evident that the concept of ecological stress cannot 

 be discussed in a general sense. Both the change of system inputs aiid 

 the properties of the system itself determine the ecosystem's stress 



