95 



In the present study, subfossil assemblages integrated over 

 several recent years showed a clear positive correlation between 

 periphyton production and trophic state for a limnologically diverse 

 set of Florida lakes. Log-transformed values of planktonic and 

 periphytic diatom concentrations were positively correlated (r = 

 0.643, p < 0.001, n = 28), as were log -transformed planktonic and 

 periphytic diatom accumulation rates (r = 0.717, p < 0.001, n = 26). A 

 positive response of both periphyton and plankton communities to 

 water-column nutrient concentrations was observed in another study 

 in Florida that examined sedimentary diatom concentrations and 

 accumulation rates along a trophic gradient (Whitmore in press). 

 Although phytoplankton and periphyton may be negatively 

 correlated within a lake (Hodgson et al. 1986), this interpretation is 

 also influenced by the scale of analysis: they are both positively 

 affected by water-column nutrient concentrations when considered 

 over a broad scale of trophic state. 



Recommended Predictive Models for Macrophyte Variables 

 During construction of the predictive models in this study, 

 particular attention was paid to the potential covariant effects non- 

 target environmental variables might have on the predictive models. 

 Equation 3.1, for example, which predicted floating-leaved biomass, 

 was shown to be confounded by pH. Equation 3.3 was a multivariate 

 regression equation predicting percent-area coverage that was 

 confounded by TSI(AVG). In general, I do not recommend historical 

 application of diatom predictive models that might relate to more 



