Field Crops Research 119 (2010) 277–289 Contents lists available at ScienceDirect Field Crops Research journal homepage: www.elsevier.com/locate/fcr Establishing a soybean germplasm core collection Marcelo F. Oliveira a,∗ , Randall L. Nelson b , Isaias O. Geraldi c , Cosme D. Cruz d , José Francisco F. de Toledo a a Embrapa Soybean, P.O. Box 231, Londrina, PR 86001-970, Brazil b USDA-Agricultural Research Service, Soybean/Maize Germplasm, Pathology, and Genetics Research Unit, Dept. of Crop Sciences, 1101 W. Peabody Dr., Univ. of Illinois, Urbana, IL 61801, United States c University of São Paulo-Faculdade de Agronomia-ESALQ, Dept. of Genetics, P.O. Box 83, Piracicaba, SP 13400-970, Brazil d Federal University of Vic ¸ osa, Faculdade de Agronomia, Avenida Peter Henry Rolfs, s/n Campus Universitário, Vic ¸ osa, MG 36570-000, Brazil article info Article history: Received 9 March 2010 Received in revised form 23 July 2010 Accepted 25 July 2010 Keywords: Glycine max Genetic diversity Germplasm bank Sampling strategies abstract Core collections are of strategic importance as they allow the use of a small part of a germplasm collection that is representative of the total collection. The objective of this study was to develop a soybean core collection of the USDA Soybean Germplasm Collection by comparing the results of random, proportional, logarithmic, multivariate proportional and multivariate logarithmic sampling strategies. All but the ran- dom sampling strategy used stratification of the entire collection based on passport data and maturity group classification. The multivariate proportional and multivariate logarithmic strategies made further use of qualitative and quantitative trait data to select diverse accessions within each stratum. The 18 quantitative trait data distribution parameters were calculated for each core and for the entire collection for pairwise comparison to validate the sampling strategies. All strategies were adequate for assembling a core collection. The random core collection best represented the entire collection in statistical terms. Proportional and logarithmic strategies did not maximize statistical representation but were better in selecting maximum variability. Multivariate proportional and multivariate logarithmic strategies pro- duced the best core collections as measured by maximum variability conservation. The soybean core collection was established using the multivariate proportional selection strategy. © 2010 Elsevier B.V. All rights reserved. 1. Introduction Access to genetic variability is critical for plant breeding. Germplasm conservation centers were created to preserve the available genetic variability before it is lost due to the widespread use of modern, improved cultivars (Brown, 1989b). In the 1980s, the International Board for Plant Genetic Resources (IBPGR) pro- vided substantial financial support for germplasm preservation that resulted in an increase in the number of collections. The emphasis placed on conservation led to the establishment and preservation of large collections, which were not totally or even partially evaluated or characterized. Although large germplasm collections are desirable from the perspective of genetic variabil- ity preservation (Frankel and Bennett, 1970), their usefulness and accessibility can be inversely related to their size (Frankel and Soulé, 1981). The increase in the number of accessions that are not adequately evaluated can diminish the effectiveness of collections (Holden, 1984; Marshall, 1989). ∗ Corresponding author. Tel.: +55 43 3371 6263; fax: +55 43 33716001. E-mail address: embrapamfo@gmail.com (M.F. Oliveira). China has the largest Glycine collection, with approximately 26,000 accessions of Glycine max and 6200 accessions of Glycine soja, located in the Institute of Crop Germplasm Resources of the Chinese Academy of Agricultural Science in Beijing (Chang et al., 1999; Carter et al., 2004) and has developed a core collection of this accessions (Zhang et al., 2003). A core collection of the perennial Glycine species has also been established (Brown et al., 1987). The soybean germplasm collection of the United States Depart- ment of Agriculture (USDA) is the second largest in the world with 16,999 accessions of introduced G. max, 1116 accessions of G. soja and 919 accessions of perennial Glycine species. Detailed origin data are available for most entries in the USDA Soybean Germplasm Collection as are data for many descriptive, agronomic, and seed composition traits. For most germplasm collections, there is a gap between the germplasm availability and its use (Peeters and Galwey, 1988) due to the collection size and financial limi- tations. In general, germplasm conservation programs have been more successful in ensuring long-term preservation then in facili- tating germplasm use. The establishment of core collections, as proposed by Frankel and Brown (1984), is an effective strategy to optimize human, mate- rial, and financial resources by providing greater efficiency in the use of germplasm collections (Spagnoletti-Zeuli and Qualset, 1993; 0378-4290/$ – see front matter © 2010 Elsevier B.V. All rights reserved. doi:10.1016/j.fcr.2010.07.021