Prevalence, Incidence, and Spatial Dependence of Soybean mosaic virus in Iowa

Abstract

The prevalence of soybean fields with plants infected with Soybean mosaic virus (SMV) in Iowa is assumed to be random, because the primary source of the virus is SMV-infected seed. Data collected from 2,500 soybean fields sampled over a 3-year period as part of the Iowa Soybean Disease Survey (2005 to 2007) were used to evaluate this assumption. SMV was first detected in early June of each year but counties in which it was first detected varied among years. Prevalence at the county scale at end of season was 32.3, 27.3, and 89.9% in 2005, 2006, and 2007, respectively. End-of-season incidence of SMV within SMV-positive counties was 1.5 to 25.0, 1.7 to 24, and 1.8 to 58% in 2005, 2006, and 2007, respectively. The number of fields in which plants infected with SMV were detected increased at the linear rate of approximately one new field every 2 days in 2007, compared with one new field every 22 days (2005) and 21 days (2006), with coefficients of determination (R2) of 93.2 to 96.8% using the linear model. Weak spatial dependence for end-of-season SMV incidence was detected using Moran's Index, indicating that the risk for SMV incidence at the county scale within Iowa at the end of the growing season is not random.

Detection and discovery of plant viruses in soybean by metagenomic sequencing

“…Because a systematic sampling strategy was not used, these data should not be interpreted to indicate virus prevalence. However, it is interesting that SMV, which is generally regarded as one of the most common viruses of soybean [ 69 ], was not found in any sample. Surprisingly, we discovered another potyvirus, ClYVV, in commercial soybean in two different years, which is significant, because cultivated soybean is not generally considered to be a host.…”

Landscape-Scale Disease Risk Quantification and Prediction

“…Most GIS software allows the creation of data surfaces through interpolation of point data, using statistical methods such as variograms and kriging (7,48,50,68). Spatially linked data surfaces then can be analyzed using statistical approaches, such as multiple regression, that account for spatial dependency and autocorrelation (31,48,68). Parametric methods, particularly logistic regression (9,42,69), have been extensively used because they are well known for balancing model parsimony and prediction robustness while providing interpretation of straightforward ecological relationships.…”