Utility of a multilevel modeling approach to investigate differences in isolation frequency of fusarium culmorum in agricultural soil across the inland pacific Northwest

Abstract

The plant pathogen Fusarium culmorum represents an inoculum source capable of contaminating grains with deoxynivalenol in the Inland Northwest region of the United States. A multilevel modeling approach utilizing varying intercepts for different sampling quadrats, fields, and iterations in the dataset was performed to characterize differences in isolation frequency of F. culmorum collected during a 2-year soil survey. Differences in the isolation frequency of F. culmorum varied the most by sampled field followed by quadrat and iteration, respectively. Higher relative elevation within the sampling region of a field limited the amount of F. culmorum recovered. The effect of annual climate variables was investigated using combinations of single-variable and multivariable model equations with linear and polynomial terms. The same data analysis approach was applied to an external dataset of F. culmorum isolation frequencies in grains from fields across eastern Australia. These results represent a case study for investigating variability within datasets containing overdispersed fungal counts and incorporating climate summaries as predictor variables.