46 

 treatments was linear with respect to GA 3 and both linear and quadratic 

 in response to 2,4-D. There also existed a significant undescribed 

 relationship above the quadratic function for the effects of 2,4-D. 

 There was no significant interaction between GA 3 and 2,4-D (a=0.05). 

 Thus, 2,4-D treatment levels had a much greater influence over the 

 response observed for waterhyacinths in the barrel studies than did com- 

 binations of GA 3 and 2,4-D. 



Tables 1-3 and 1-4 present summaries of regression analyses and 

 regression coefficients for the treatment levels of GA 3 and 2,4-D, 

 excluding the 2.24 kg/ha 2,4-D treatment, on the percent change biomass 

 and number of plants, respectively. These analyses demonstrated that 

 (1) the response to GA 3 was linear, (2) the negative response to 2,4-D 

 was both linear and quadratic, and (3) the interaction coefficient for 

 GA 3 and 2,4-D was insignificant, suggesting that at best the effect of 

 GA 3 was only additive. The resulting regression models for the response 

 of waterhyacinths to GA 3 and 2,4-D accounted for 82.9 percent of the 

 variability of mean change in biomass (Table 1-3) and 85.1 percent of 

 mean change in number of plants (Table 1-4). In order to account for an 

 additional portion of the variability of the response to treatments, 

 pretreatment weights of the waterhyacinths was included in the model but 

 found to be insignificant. 



Tables 1-5 and 1-6 provide another representation of the results in 

 terms of minimum and maximum values observed, treatment means, standard 

 error of the mean, and Waller-Duncan groupings for percent change in 

 biomass and number of plants. Figures 1-1 and 1-2 provide a graphical 

 representation of the percent change in biomass and number of plants, 

 respectively. The Waller-Duncan comparisons detected significant 



