Comparison of Multispectral imaging and Near infrared Reflectance Spectroscopy methods for phenotyping resistance of cereals to Fusarium Head Blight (FHB) and Deoxynivalenol (DON)
Résumé
To encourage development of cereal varieties resistant to Fusarium Head Blight and mycotoxins for the National List, two new methods of phenotyping were developed by GEVES on winterwheat. They are more reliable and faster than in-field visual assessment and cheaper than LCMS-MS for measuring DON content. Multispectral imaging is being developed to predict /quantify damage from diseases on several species. Near-Infrared Spectroscopy is an efficient and non-destructive method for seed quality analysis. The aim was to assess the potential of these technologies for quantifying Fusarium graminearum damage, compared to visual assessment and DON content in kernels. i) With Multispectral imaging (VideometerR), the algorithm “Fusa-spectrale wheat”, was developed to evaluate the percentage of Fusarium Damaged Kernels (FDK) at maturity for varietal classification using Canonical Discriminant Analysis between kernels infected/not infected, on three years at several sites. FDK is strongly correlated with percentage of scabbed spikelets assessed by visual scorings (R>0.87) and with DON content (R>0.88). Correlation between FDK and Fusarium graminearum biomass assessed by real-time PCR was stronger in trials with only F. graminearum (R=0.97) than trials with a complex Fusarium-Microdochium. For durum wheat and triticale, studies are ongoing. ii) NIR calibration was developed from a sample set of varieties produced in two locations over two crop years. Calibration models (visual scoring, DON content) were developed using Partial Least Square Regression (PLS). Each model was checked by internal cross validation. Their performance was assessed by the coefficient of determination in calibration (R2cal) and standard error in cross validation. The R2cal observed was 0.90 for “DON content” model and 0.79 for “visual scoring” model. These two technologies show potential to evaluate FHB and DON resistance on cereals. The NIRS models require additional data to include geographical, genotyping and years effects. With multipectral and hyperspectral imaging, research is ongoing to quantify Microdochium spp.
| Origine | Fichiers éditeurs autorisés sur une archive ouverte |
|---|---|
| Licence |