Prediction of grain protein content in winter wheat (Triticum aestivum L.) using plant pigment ratio (PPR)

The applicability of the hyperspectral data from the canopy to the prediction of wheat grain quality was assessed for winter wheat. A training experiment and a validation experiment with contrasting nitrogen (N) levels and different cultivars were conducted, respectively, at different locations in Beijing, China. The wheat canopy spectral reflectance over 350–2500 nm, leaf N concentration and chlorophyll (Chl) concentration were measured at different growth stages, and the grain protein content was also determined after harvest. Eight vegetation indices (VIs) were compared relating to leaf N concentration, and the result indicated that the plant pigment ratio (PPR, <F>(R550-R450)/(R550+R450)</F>), a Chl-based index, was most applicable to predict wheat grain protein due to its significant correlation with leaf N concentration at the post-anthesis stage. Based on the relationships among PPR, leaf Chl concentration, leaf N concentration, and grain protein content, the statistical prediction models of grain protein content for Zhongyou9507 (a hard winter wheat) and Jingdong8 (a semi-hard winter wheat) were developed. The root mean square error (RMSE) of the 18 DAA (days after anthesis) model of Zhongyou9507 was 0.175; those of the anthesis model and the 11 DAA model of Jingdong8 were 0.238 and 0.982, respectively. Taking both the precision and accuracy into account, the 18 DAA model of Zhongyou9507 and the anthesis model of Jingdong8 were recommended to predict grain protein content for each cultivar. The result demonstrated that PPR could be used to assess grain quality of winter wheat. [Copyright &y& Elsevier]