44 



year variation can not be handled in this way at all. Often long-term among-years variation strongly influences 

 marine coastal shelf communities, for example by infrequent events such as severe storms setting successional 

 processes back to an earlier stage (Glemarec 1978). 



What frequency of sampling is appropriate in a long-term monitoring study? If cost were no object, then 

 obviously the answer would be as frequently as possible. With cost factored in, the answer will be some 

 compromise between the frequency of sampling and the chance of detecting (or missing) a pollution impact 

 event if it occurred. This is the philosophy of a power analysis again: how does the probability of detecting a 

 pollution impact event of a given duration vary as a function of sampling frequency? I have developed a 

 computer program which will analyze time series monitoring data, or if such data are not available it will simulate 

 (and then analyze) a set of data having specified parameter values (i.e., mean frequency, temporal autocorrelation 

 which determines cyclic pattern, and length of the time series). I have repeatedly run this program to simulate 

 data covering a range of parameter values, and sampled each set of data using sampling frequencies covering a 

 range of frequencies. Thus, points on a response surface representing "power to detect a pollution impact event" 

 were estimated, as a function of sampling frequency, and of mean frequency and temporal autocorrelation of the 

 pollution impact events. A simple empirical response surface model was fit to the results. I intend to continue 

 with this approach, developing the model further, extending the parameter ranges, and re-fitting the response 

 surface model, using a variety of sets of real time-series pollution impact monitoring data to estimate power 

 curves, and adding to the program the capability of simulating background noise representing natural 

 environmental variation having its own mean frequency and temporal pattern. 



REFERENCES 



Bayne, B.L., K.R. Clarke, and J.S. Gray, eds. 1988. Biological effects of pollution: results of a practical 

 workshop. MEPS Special. Mar. Ecol. Progr. Ser. 46:278. 



Carney, R.S. 1987. A review of study designs for the detection of long-term environmental effects of offshore 

 petroleum activities, pp. 651-696. In D.F. Boesch and N.N. Rabalais, eds. Long-term environmental effects 

 of offshore oil and gas development. Elsevier, NY. 



Glemarec, M. 1978. Problemes d'ccologie dynamigue et de succession en baie de Concarneau. Vie Milieu 

 (Ser AB):l-20. 



Green, R.H. 1979. Sampling design and statistical methods for environmental biologists. John Wiley and Sons, 

 NY. 257 pp. 



Green, R.H. 1984. Statistical and nonstatistical considerations for environmental monitoring studies. Envir. 

 Monit. Assessm. 4:293-301. 



Green, R.H. 1987. Statistical and mathematical aspects: distinction between natural and induced variation, 

 pp. 335-354. In V.B. Voulk, G.C. Butler, A.C. Upton, D.V. Parke, and S.C. Asher, eds. Methods for 

 assessing effects of mixtures of chemical. SCOPE 33c. Wiley, Chichester, U.K. 



Green, R.H. 1989. Inference from observational data in environmental impact studies. Proc. Intern. Statist. 

 Instit., 47th Session, Paris, France. 



Green, R.H. In press. Power analysis and practical strategies for environmental monitoring. Envir. Research. 



Hurlbert, S.H. 1984. Pseudoreplication and the design of ecological field experiments. Ecol. Monogr. 54:187- 

 211. 



