Abstract

The evaluation of seed vigor is significant in seed quality evaluation for cultivation and plant breeding. However, seed testing is a time-consuming and laborious process that requires specialized expertise. Consequently, it is necessary to design and fabricate methods that reduce labor, increase precision in seed vigor testing, and reduce labor costs. The objective of this study was to classify seed vigor using the radicle emergence test and the K-means clustering algorithm. Radicle emergence of maize seeds, 120 lots, was tested using between paper technique at 25 °C in the dark with 4 replications of 50 seeds. The radicle emergence was scored at 2 mm in length daily for seven days from the start of seed testing. GERMINATOR® then used to calculate the radicle emergence indices such as maximum radicle emergence (MaxRE), radicle emergence speed (t₅₀RE), area under the curve of the radicle emergence fitted curve (AUC), mean radicle emergence time (MRET) and uniformity of radicle emergence (U₇₅₂₅). Principle component analysis (PCA) was analyzed with K-means by Matlab®, which calculated the optimal number of clusters using the Elbow and Silhouette methods. The results showed that using MaxRE, t₅₀RE, AUC, MRET and U₇₅₂₅ by normalizing the data with the STANDARDIZE function, based on the calculations of Matlab®, OptimalK is equal to 2, which can be grouped into 2 clusters which is appropriate for practical use. Therefore, the evaluation of seed vigor using the radicle emergence test and the K-means clustering technique using MaxRE, t₅₀RE, AUC, MRET and U₇₅₂₅ can appropriately classify maize seed vigor.