Near-infrared reflectance (NIR) spectroscopy as a high-throughput screening tool for pest and disease resistance in a sugarcane breeding programme

摘要

Pests and diseases cause major production and economic losses in sugarcane cropping systems. The most effective form of long-term protection is through the use of resistant varieties. However, phenotyping sugarcane genotypes for pest and disease resistance is difficult and costly. Near-infrared reflectance (NIR) spectroscopy was investigated for its potential to predict the constitutive components of resistance to pests and diseases in germplasm in the South African sugarcane breeding programme. Two hundred and twenty-two genotypes were scanned over the 1100-2300 nm wavelength range using a fiber-optic probe. Partial least square (PLS) regressions were applied to bud, internode and leaf spectra that were pretreated (second derivative) and scatter-corrected (SNV and de-trending). Calibration models resulting from the correlations between NIR measurements and existing ratings, gave coefficients of determination for calibration (R~2c, the closer to one the better) and standard errors of prediction by leverage correction (SEP, the lower the better) of 0.72 (SEP 1.19) for resistance to the African stalk borer (Eldana saccharina), 0.62 (SEP 1.50) for smut (Sporisorium scitamineum), 0.62 (SEP 1.07) for sugarcane thrips (Fulmekiola serrata) and 0.67 (SEP 1.02) for brown rust (Puccinia melanocephala) ratings, respectively. Performance of the calibration models in prediction are encouraging and demonstrate the potential of NIR spectroscopy as a high-throughput screening method to evaluate sugarcane genotypes for resistance to pests and diseases. We believe that NIR spectroscopy can be used as an additional screening method, in the early selection stages of the breeding programme, which should increase the proportion of resistant genotypes carried forward to later selection stages.