Link to home

Explaining Loss Caused by Tomato spotted wilt virus on Tobacco with Boreal Winter Weather: A Bayesian Approach

April 2011 , Volume 101 , Number  4
Pages  462 - 469

A. L. Mila

Department of Plant Pathology, North Carolina State University, Campus Box 7405, Raleigh 27695.

Go to article:
Accepted for publication 11 November 2010.

In North Carolina, Tomato spotted wilt virus (TSWV) has regularly been reported since 1997, with incidence being the highest in 2002. At the end of each season, a questionnaire is sent to the county agents to report disease losses. TSWV reported losses in 1993 to 2007 from 58 counties were available. A county–year combination was considered a case and, in total, 494 cases were analyzed. The winter months' temperature and precipitation significantly explained the reported TSWV loss (R2 = 0.82). Specifically, the monthly average air temperature for December to February had a positive association with TSWV loss (P < 0.0001) whereas the total precipitation for the same months had a negative effect (P < 0.0001). Bayesian hierarchical models were implemented to include spatial and nonspatial random effects to investigate if there were significant spatial correlations or unexplained variability, respectively, and, thus, other significant variables that were ignored in the model development. The spatial random effects were not significant but the nonspatial random effects were significant in 36 cases. The importance of spring weather to dispersal of thrips and TSWV has been previously identified. Winter weather also may be a good indicator of potential available TSWV inoculum for the upcoming season.

Additional keywords: posterior distribution, random error.

© 2011 The American Phytopathological Society