ORDINATION OF MACROINVERTEBRATE COMMUNITIES 763 



station 4 and station 6 (Dayton sewage-treatment plant) (see Fig. 5). 

 In September, however, station 5 macroinvertebrates showed their 

 strongest affinities with the Dayton sewage-treatment-plant station 

 and the West Carrollton station (8). Both these stations are slightly 

 below the sewage-treatment plants. The probable interpretation of 

 these associations is that, with decreased river flow, a much higher 

 proportion of the water below the plant's discharge comes from the 

 plant discharge itself rather than from directly upstream. Therefore 

 plant effects, such as an increase in river temperature or chlorination, 

 are heightened at this time of year. 



Ordinations are usually two-dimensional representations of a 

 multidimensional configuration (Kaesler and Cairns, 1972). By the 

 use of a similarity coefficient, e.g., CC and ordination, a station is 

 reduced to a single multidimensional point in a space defined by axes 

 that reflect changes in species composition (Mcintosh, 1973). A 

 factor analysis (Nie et al., 1975) of variations in the macro- 

 invertebrate species and species abundances for the data collected in 

 this study showed 14 common factors to be significant in deter- 

 mining such variations (the factor analysis can estimate the number 

 of significant factors but unfortunately does not identify these 

 factors). It is evident, then, that an interpretation of causal factors 

 may be difficult. This difficulty, due to complexity of factors, can be 

 seen in the June ordination; it was not possible to envision an axis 

 corresponding to an ecocline, known community variables, gradients 

 of disturbance effects, or of community development in the manner 

 of Whittaker (1975). The August and September samples, however, 

 ordered quite well along an oblique axis corresponding to distur- 

 bance (shown as dashed lines in Fig. 5). Although other factors may 

 still be somewhat important in variation in species composition 

 during these stress periods, pollution effects become more robust in 

 the ordering of communities and overwhelm the effect of normal 

 variables on faunal composition. 



ACKNOWLEDGMENTS 



My thanks to Gary Hater, Brad Averill, Joe Redi, Tim Dineen, 

 Tom Federle, Pete Spatt, Dan Strome, and Doug Yeakel for their 

 assistance in the field and in the laboratory. The criticisms and advice 

 of Drs. Michael C. Miller and Thomas C. Kane were very helpful 

 throughout the study. 



This study was supported by a grant from Dayton Power and 

 Light Company, Dayton, Ohio. 



