Bean pod mottle virus: Spatial and temporal dynamics at different spatial scales and the impact of time of infection on soybean yield and quality
A comprehensive state-wide soybean disease survey was carried out during the 2005- 2007 growing seasons to determine seasonal and geospatial patterns of soybean diseases in Iowa. Bean pod mottle virus (BPMV) was one of the viruses assessed in the survey. To quantify BPMV prevalence and incidence at the field and county scales, 30 soybean plants per field were sampled from 8-20 soybean fields/county using a systematic sampling design. The GPS coordinates for each soybean field sampled were recorded. Soybean plants were tested for the presence of BPMV using enzyme linked immunosorbent assay (ELISA). Field- and county-scale BPMV prevalence and incidence data were mapped using ArcGIS (ESRI, Redlands, CA). County-scale prevalence of BPMV was 40.6% in 2005, 90.1% in 2006, and 74.7% in 2007. On the field basis, the prevalence of BPMV in soybean fields was 9.5%, 40.4% and 27.4% for 2005, 2006, and 2007, respectively. BPMV incidence within soybean fields from 2005-2007 was 4.4%, 24.8%, 9.8%, respectively. Moran's I analysis revealed significant spatial dependence for BPMV incidence among counties, indicating a nonrandom distribution of BPMV among Iowa counties. Based on GIS point maps for the three growing seasons, kriged maps revealed an increased risk for BPMV from the southern border of the state to the northern border of the state. Regression analysis for relationship between BPMV incidence and county latitude indicated a strong linear relationship with BPMV incidence increasing from 0.24 to 2% for every 10 km increase in latitude during the 3-year study. To quantify the temporal and spatial spread of BPMV within soybean fields, 30-cm-long quadrats were established within field plots of the soybean cultivar NE3001 (150 quadrats per plot) in 2006 and 2007. Quadrats were sampled every 8-11 days throughout each growing season, beginning 25 days after planting. Sap from the youngest, fully-expanded leaflet from each of the four plants per quadrat was group tested for the presence of BPMV using ELISA. To determine the relationship between day of BPMV detection and soybean yield, 35 soybean quadrats representing each BPMV detection time were randomly selected to obtained yield, yield components, and grain quality data. For sampling dates with < 35 quadrats that were BPMV positive, all quadrats were harvested to obtain yield data. Yield, yield components, and grain quality data from quadrats were then plotted with respect to day of year that BPMV was first detected in a quadrat, and linear regression was used to quantify the relationships between time of BPMV detection with yield, yield components (number of pods per plant, seeds per pod, and 100-seed weight), and grain quality (percent mottled seed, protein content and oil content) (y). BPMV was first detected in quadrats as early as the first sampling date (30 May 2006 and 12 June in 2007). The onset of BPMV epidemic (5% BPMV incidence) occurred on 29 June 2006 and on 18 July in 2007. Time to 50% BPMV incidence ranged from day of year 207 (26 July) to 215 (3 August) in 2006. In 2007 time to 50% occurred well past senescence. Final BPMV incidence was between 85.8 and 94.4% in 2006 and between 10.0 and 37.1% in 2007. In 2006, temporal rates of BPMV infection ranged from 0.09 to 0.12 logits/day with R2 values ranging from 97.8% to 98.3%, indicating that BPMV incidence within soybean plots doubled every 5.3 to 6.9 days. Rates of BPMV temporal spread in 2007 were significantly slower (0.05 to 0.07 logits/day, R2 values ranged from 91.6% to 99.7%), with doubling times ranging from 13.8 to 17.3 days. Plots with the earliest onset of BPMV had the highest BPMV incidence at the end of the growing season and the largest areas under the BPMV progress curves in both 2006 and 2007 (the slope for both years was 1.1 and R2 values of 65.4% and 76.9%, respectively). Spatial analyzes using ordinary runs revealed that BPMV-infected quadrats had spatial patterns that were highly aggregated throughout both growing seasons. Time of BPMV detection explained 89.7% and 57.9% of the variation in soybean yield in 2006 and 2007, respectively. The yield damage function (slope) was -0.23 bushels/acre/day in 2006 and -0.12 bushels/acre/day, indicating that for every 4.3 and 8.3 days BPMV detection was delayed, in 2006 and 2007, respectively, soybean yield would increase by one bushel. The linear relationship between number of pods per plant and time of BPMV detection was significant in 2006 (slope = 0.15, R2 = 72.8%), but not in 2007. The number of seeds per pod was not impacted by time of BPMV detection in either year, and 100-seed weight was impacted only in 2006 (slope = 0.013, R2 = 78.5%). There was a significant linear relationship between time of BPMV detection and the percentage of mottled seeds in the two years (slope = 0.34, 0.15; R2 = 82.8, 48.3%); earlier BPMV detection was associated with a higher percentage of mottled seeds. Time of BPMV detection in a quadrat did not influence protein and oil content in either year. This research was the first to document the nonrandom distribution of BPMV prevalence and incidence in Iowa, and the first to show that BPMV spread within soybean fields is highly aggregated over time, which has important implications for yield losses and BPMV sampling designs. This project was also the first to quantify the relationship between BPMV date of detection and reduction in yield and impact on yield components and soybean quality.