Regression Analyses for Evaluating the Influence of Bacillus cereus on Alfalfa Yield Under Variable Disease Intensity

Abstract

We developed and tested regression methods to exploit the variability in disease inherent in field experiments, and applied the methods to evaluate strains of Bacillus cereus for biocontrol efficacy. Four B. cereus strains were tested for their effect on alfalfa (Medicago sativa) performance in 16 field trials planted during 1993 to 1996 at multiple sites in Wisconsin. To evaluate performance of the strains, we used the ratio of (metalaxyl response)/(untreated control response) as a measure of disease intensity within the experiments. The ratio of (Bacillus response)/(untreated control response) was then regressed as a function of disease intensity. The slope of the resulting line provides a statistical test to compare performance of the Bacillus strain with that of the untreated seed (Ho: slope = 0) and metalaxyl controls (Ho: slope = 1). Under conditions in which disease occurred, forage yield of plots planted with seed treated with B. cereus strain AS4–12 exceeded yield from the untreated control plots (P = 0.002) and was similar to yield of plots planted with metalaxyl-treated seed (P = 0.14). Yield gain associated with AS4–12 and metalaxyl seed treatment averaged 6.1 ± 2.8% (±standard error) and 3.0 ± 2.8%, respectively. In contrast to the regression approach, means analysis by analysis of variance did not detect differences among treatments. Three other B. cereus strains either did not increase alfalfa yield or increased yield less than did AS4–12. Metalaxyl and three of the Bacillus strains increased seedling emergence, but the improved stands were not predictive of increased forage yield. In six additional studies conducted for one season in 1997, AS4–12 enhanced yield of two cultivars at diverse locations in Wisconsin, but there was an apparent cultivar-location interaction. A strong correlation between response to AS4–12 and metalaxyl treatment suggests that these treatments controlled similar pathogens, most likely the oomycete pathogens Phytophthora medicaginis and Pythium spp.