Gilbert Sigua1, J. N. Rutger1, R.J. Bryant2, Harold Bockelman3, R.G. Fjellstrom4, M.H. Chen4, Thomas Tai5, and Anna Mcclung1. (1) USDA-ARS, USDA-ARS-STARS, 22271 Chinsegut Hill Rd., Brooksville, AR 34601, (2) USDA-ARS, DB NRRC, PO Box 1090, Stuttgart, AR 72160, (3) USDA-ARS, NSGC, 1691 S 2700 W, Aberdeen, ID 83210, (4) USDA-ARS, Rice Research Unit, 1509 Aggie Drive, Beaumont, TX 77713, (5) "USDA-ARS CPGRU, U. of CA Davis", "1308 PES, 1 Shields Avenue", Davis, CA 95616, United States of America
The USDA rice germplasm collection is managed by GRIN at
www.ars-grin.gov. A core subset including about 10% of total accessions was assembled in 2002. Frequency distributions for 14 descriptors: days to flower, plant height, awn type, panicle type, plant type, kernel length, width, length/width, and weight, hull color, hull cover, bran color, amylose content and alkali spreading value (ASV) were used to analyze the association of the core to the whole collection. Eight descriptors or 57% had correlation coefficients greater than 0.9, or coefficient of determination greater than 0.8. Descriptor ASV, amylose, kernel length and ratio had correlations less than 0.8, relatively lower than others. This is probably due to the fact that the data available in the whole collection span a narrower period of time when accessions were introduced and a narrower geographic range (fewer countries) of origin than the core collection. The 14 descriptors proved a high correlation of r=0.94 (P<0.0001), or determination of r
2=0.88 between the whole and core collections for frequency distributions. The high correlation demonstrated that information drawn from the core subset containing 10% of accessions in the whole collection could be effectively used to assess the whole collection with 88% certainty.