Application of near Infrared Transmittance Spectroscopy to the Estimation of Protein and Lipid Contents in Single Seeds of Soybean Recombinant Inbred Lines for Quantitative Trait Loci Analysis

The evaluation of genetic and environmental components related to protein and lipid contents requires a large number of seeds to be analysed. In this study, the use of non-destructive analysis using near infrared (NIR) transmittance spectroscopy was investigated for the determination of protein and lipid contents in intact single seeds of soybean [Glycine max (L.) Merrill] recombinant inbred lines (RIL). Calibration equations were developed from spectral information gathered between 700 and 1100 nm. Prior to the calculation, the original spectra were transformed to the second derivative spectra. This work demonstrated that separating seed into two groups (≥ or <6 mm), followed by second derivative transformation and selection of the firstwavelength couldreduce the sample size effect remarkably. Furthermore, calibration and validation equations were developed from the combination of F₈ and F₉ seeds in order to correct the bias which occurred when F₉ seeds were analysed using the F₈ only calibration. All the equations were obtained with four optical terms. The protein content equation in the ≥6 mm group resulted in correlation coefficient (R) of 0.88 and standarderror of prediction (SEP) of 1.32% protein, while inthe <6 mm group theresults were0.87for R and 1.57% protein for SEP. For the lipid constituent, R and SEP were 0.80 and −0.14% in the ≥6 mm group and 0.72 and −0.09% in the <6 mm group, respectively. These equations were then applied to estimate the protein and lipid contents of soybean RIL F₉ generation and the implications of these results are discussed.