Sorghum grown as fodder, food and biofuel crop in about 100 countries is a dietary staple food for more than a million in subtropical and semi-arid regions of Asia and Africa. The reliable characterization of genotypes serves as a base for every crop improvement programme and a machine vision system is a prime option for efficient discrimination. The 39 sorghum landraces collected from various parts of Trichy district were measured for area, perimeter, width, length, aspect ratio, rectangularity, and circularity by Grain Scanner (Grain Scanner RSQI 10A). The maximum area, perimeter, length and width were recorded in Ammapatti local 2 (12.68 mm2), Keezhapuliyur local 1(13.27 mm), Muthiyampalayam local 3 (4.90 mm) and Ammapatti local 2 (3.82mm), respectively. Among the genotypes observed, Ammapatti local 2 and Keezhapuliyur local 1 showed significance for all the traits except aspect ratio. The cluster analysis grouped genotypes into four clusters and number of genotypes in each cluster was 15, 8, 3 and 13 in clusters I, II, III and IV, respectively. Based on aspect ratio, the members of cluster I and III were elliptical and II and IV were circular. The members of cluster II have high germination percentage, threshability and grain yield due to its boldness. Hence circular bold seeded high yielding sorghum varieties can be developed from members of cluster II. Relatively small-seeded genotypes come under cluster IV. The major contributors for the cluster formation were the length, width and area. The genotypes collected from different locations of the same geographical area were grouped into different clusters and this is due to minor variation that happens in same species during evolution. The minor variation which is not detected by general characterization can be easily scanned by Grain Scanner and so, it is an evident tool in characterizing genotypes based on grain morphology.
Read full abstract