|AL-AMERY, MAYTHEM - University Of Kentucky|
|GENEVE, ROBERT - University Of Kentucky|
|SANCHES, MAURICIO - Sao Paulo State University (UNESP)|
|LEE, CHAD - University Of Kentucky|
|VIEIRA, ROBERVAL - Sao Paulo State University (UNESP)|
|HILDEBRAND, DAVID - University Of Kentucky|
Submitted to: Seed Science Research
Publication Type: Peer Reviewed Journal
Publication Acceptance Date: 3/29/2018
Publication Date: 5/11/2018
Citation: Al-Amery, M., Geneve, R.L., Sanches, M., Armstrong, P.R., Maghirang, E.B., Lee, C., Vieira, R., Hildebrand, D. 2018. Near-infrared spectroscopy used to predict soybean seed germination and vigor. Seed Science Research. 28(3):245-252. https://doi.org/10.1017/S0960258518000119.
Interpretive Summary: Information on soybean germination and vigor of seeds is important for seed producers to know. This study looked into using near-infrared spectroscopy (NIR) as a method to measure germination, vigor and electrical conductivity of soybean seeds to determine seed health. Germination defines how many seeds actually sprout, vigor measures how quickly the seeds grow after sprouting and electrical conductivity is an indirect measurement of seed viability or long term survivability. Results showed that NIR spectroscopy can be used to discriminate between low and high levels of germination and vigor but not viability. As such NIR spectroscopy could be used to identify lots of poor seed quality from those of high quality which could make it a good screening tool as opposed to traditional sprouting tests which are laborious and time consuming.
Technical Abstract: The potential of using near-infrared (NIR) spectroscopy for differentiating levels in germination, vigor, and electrical conductivity of soybean seeds was investigated. For the 243 spectral data collected using the Perten DA7200, stratified sampling was used to obtain three calibration sets consisting of 183 spectra with the remaining 60 spectra used as independent validation sample sets. Results showed that NIR spectroscopy can be used for qualitative discrimination between low and high levels of germination and vigor. The percentage of correct classification of germination score was 84.1% for low and 87.2% for high level samples at a 1.7 cutoff. The vigor level was correctly classified at 86.7% for low and 75.6% for high levels at a 1.6 cutoff. Differentiation in viability was not possible using NIR spectroscopy with 100% correct classification for the low score and only 5.6% for the high score, indicating that most samples were classified as low viability. Validation of the NIR calibration models developed for quantitative measurement of germination, vigor and electrical conductivity had R2 of 0.600, 0.570, and 0.566, respectively, indicating that NIR spectroscopy can only explain approximately 60% of variation across seed quality parameters. Regression coefficients showed that constituents that were used for predictions included oil, protein, starch and water contents. This study showed the ability to differentiate low from high germination and/or vigor which will be useful for breeders as a tool for enabling segregation of desired seed quality that will assist in enhancing development of better quality seeds.