ONTARIO GRAIN FARMER 18 AGRONOMY Ralph Pearce Dual-edged technology helping wheat growers A smartphone app can monitor for FHB and estimate head counts The WheatScanR app is a new technological tool to help researchers and wheat growers monitor Fusarium head blight (FHB) and estimate wheat head counts for yield predictions. In 1996, Fusarium head blight (FHB) decimated more than 90 per cent of the Ontario wheat crop, at a loss of more than $120 million. With improved genetics, timely fungicide applications, and retuning harvest methods, the effects of FHB have been reduced, yet it remains a considerable threat. Now, 30 years later, plant breeders and growers stand to benefit from a new app conceived and built by a University of Guelph PhD candidate. WheatScanR is available on iOS and Android platforms, free of charge. Developed primarily to help researchers screen varieties and assess the extent of FHB in wheat cultivars, it can also help growers monitor their fields for the disease and perform head counts for yield estimates. Riley McConachie is the developer of the WheatScanR and began work on the platform while immersed in his master’s degree. He started in the Undergraduate Student Experiential Learning (USEL) program and has continued with his postgraduate studies. The app was born out of painstaking labour from many test plots, inoculating thousands of wheat heads, then estimating levels of infection by physically examining the heads. THE ACTUAL PROCESS With the WheatScanR, McConachie used an open-source software to segment images of the heads. Wheat-head detection and segmentation combine shapes and colours, providing an accurate image of what’s present. A spectral index is used to determine the extent of bleaching from FHB. Green colouring indicates a diseasefree head. The app removes much of the variability from human estimates of FHB, which were previously assumed to be 100 per cent accurate, though even experienced researchers find distinguishing 60 to 70 per cent infection a challenge. The app is capable of capturing millions of pixels in an image, which eliminates variability, adding to its accuracy in predicting FHB in a field. “Some of our testing has shown that it’s within a five per cent severity estimation; if you say the plot is 10 per cent infected, it’s plus or minus five per cent from that number,” says McConachie. “That might seem like a little bit, but studies have shown that with person-to-person (counts), there’s about a 10 per cent sway. The app’s accuracy is a bit more consistent.” That’s a huge advantage for researchers and plant breeders, particularly in assessing Fusarium resistance in breeding lines. In a breeding program, the focus is on germplasm improvement by introducing new traits and selecting for them amongst thousands of breeding lines.
RkJQdWJsaXNoZXIy MTQzODE4