112 LEFFLER 



cient. The probability of a significant correlation (P< 0.100) was 

 also calculated. The binomial test was used to detect trends across all 

 parameters. The only significant correlation was a positive relation- 

 ship between particulate matter and constancy stability during the 

 first perturbation (P = 0.020). I must conclude that a generaliza- 

 tion that attempts to explain an ecosystem's stress response as a 

 function of its mass is not justified by these results. Neither 

 resistance nor resilience showed any tendencies toward such a 

 correlation. This contradicts the hypotheses of GoUey (1974) and 

 Webster, Waide, and Patten (1975). Harwell, Cropper, and Ragsdale 

 (1977) reported similar conclusions based on mathematical argu- 

 ments. 



Relationships Among Types of Stability 



The relationships among the five different measures of stability 

 were also investigated. Kendall's rank correlations were calculated 

 for each measured parameter and each pairing of stability results 

 based on the stability ordering of treatments for both the diversity 

 and the nutrient— energy studies. Few significant correlations 

 (P < 0.100) were discovered. Again, binomial tests across all parame- 

 ters from both studies were used to detect significant (P< 0.100) 

 trends. No such relationships were found. The most highly correlated 

 (P = 0.132) pair of stabilities was resistance and total relative 

 stability. This was not unexpected considering their definitions. No 

 significant relationship between resistance and resilience was discern- 

 ible based on all 22 parameters. This appai'ently contradicts the 

 generality of the inverse relationship postulated by Webster, Waide, 

 and Patten. 



It has been suggested that an ecosystem's ability to resist and to 

 recover from a stress may be a function of habitat predictability 

 (O'Neill, 1976). Such ecosystems as estuaries, which are subjected to 

 highly variable environments, typically exhibit low constancy stabil- 

 ity but high levels of resistance and resilience (Boesch, 1974; 

 Copeland, 1970; Larsen, 1974). It may be worthwhile to consider 

 whether a system's response to stress can be predicted by monitoring 

 its variability under unstressed steady-state conditions. An inverse 

 relationship between constancy stability and the perturbation 

 response stabilities might be postulated. Correlation results, however, 

 indicate that such a generalization is not justified, at least in terms of 

 the two microcosm studies considered. Constancy stability fails to 

 correlate with any of the measures that characterize an ecosystem's 

 stress response. 



