Estimating sorghum leaf dhurrin content using a handheld near infrared instrument

Dhurrin is a toxic cyanogenic glycoside present in sorghum as a secondary metabolite. As such, dhurrin content of plants is important when sorghum is used as a forage crop. Moreover, leaf dhurrin content may indicate pre- and post-flowering drought resistance of sorghum germplasm. Standard method of analysis by HPLC is expensive and time consuming. Therefore, a feasibility study was conducted to measure dhurrin content in sorghum leaves by NIR spectroscopy using a handheld NIR spectrometer.

Partial least squares (PLS) regression, back propagation neural network (BPNN) and deep learning artificial neural network (ANNDL) models were developed to estimate dhurrin content in fresh and dry sorghum leaves. However, it was impossible to develop useful models with fresh leaf scans. NIR spectroscopy models could estimate the dhurrin content of dry sorghum leaves with R² = 0.711-0.719, RMSEP = 5.89 - 5.96 µg/mg and bias of -0.26 - 0.11 µg/mg in a population of leaves with dhurrin content ranging from 0.65- 46.52 µg/mg.

Therefore, NIR spectroscopy may be used as a rapid and cost-effective technique for pre-screening sorghum germplasm for leaf dhurrin content in plant breeding programs.