Degree Name

Master of Science (MS)

Semester of Degree Completion


Thesis Director

U. Douglas Zimmerman


Gross primary productivity, heterotrophic bacterial numbers, and net phytoplankton densities of seven ponds in Coles County, Illinois, were studied in relation to physical, chemical, and biological habitat variables (light intensity and duration, turbidity, water temperature, pH, dissolved oxygen, sulfur, nitrogen, phosphorus, production, bacteria, and phytoplankton). Ten observations were made for each pond (except where otherwise noted) from 17 June through 25 August 1974. Stepwise multiple linear regression analyses of the data were used in order to determine those environmental factors which were important in predicting (i.e., significantly correlated with) bacterial and phytoplankton densities, and production. A multiple linear regression equation for each biological parameter analyzed as the dependent variable (Y) was then constructed from the intercept and the regression coefficients for the significant independent variables (X's).

The regression analyses utilized in this study suggested that there is usually no one habitat variable responsible for (i.e., correlated with) productivity and bacterial and phytoplankton densities. When considering the controlling element(s) of an ecosystem at any one time, it appears that the interactions between the various habitat variables are probably more important in determining (or predicting) environmental conditions than any one habitat variable alone.

Each of the seven ponds sampled appeared to be unique with respect to the habitat variables correlated with those biological parameters analyzed as the dependent variables. However, a program for the analysis of variance for factorial design indicated that the dependent variables themselves (and some of the independent variables) were not entirely unique to the ponds sampled. In order to gain some insight into the nature of these differences, a program for the Duncan's new multiple range test was utilized.

This study has indicated that the use of stepwise multiple linear regression analysis may be a valuable tool in aquatic research for evaluating the relative improtance of habitat variables in predicting variation in biological parameters. However, in order to develop more reliable predictive measures, additional studies should be made concerning biological rate data and nonlinear expressions of variable interactions based on the significantly correlated habitat variables suggested in the regression analyses. Systems analysis procedures could then be used in building effective simulation models.