246 J. AMER. SOC. HORT. SCI. 128(2):246–252. 2003. J. AMER. SOC. HORT. SCI. 128(2):246–252. 2003. Comparison of AFLPs, RAPD Markers, and Isozymes for Diversity Assessment of Garlic and Detection of Putative Duplicates in Germplasm Collections Meryem Ipek and Ahmet Ipek 1 Department of Horticulture, University of Wisconsin, 1575 Linden Drive, Madison, WI 53706 Philipp W. Simon 2 U.S. Department of Agriculture–Agricultural Research Service, Vegetable Crops Research Unit, 1575 Linden Drive, Madison, WI 53706 ADDITIONAL INDEX WORDS. Allium sativum, amplified fragment length polymorphism, genetic diversity, molecular marker comparisons, duplicate accessions ABSTRACT. Garlic (Allium sativum L.) is an asexually propagated crop that displays much morphological diversity. Studies which have assessed garlic diversity with isozymes and randomly amplified polymorphic DNA (RAPD) markers generally agreed with the morphological observations but sometimes failed to discriminate clones. To discriminate among closely related garlic clones in more detail, we introduced amplified fragment-length polymorphism (AFLPs) to evaluate the genetic diversity and phenetic relatedness of 45 garlic clones and three A. longicuspis clones and we compared AFLP results with RAPD markers and isozymes. Three AFLP primer combinations generated a total of 183 polymorphic fragments. Although similarities between the clusters were low (≥0.30), some clones within the clusters were very similar (>0.95) with AFLP analysis. Sixteen clones represented only six different banding patterns, within which they shared 100% polymorphic AFLPs and RAPD markers, and likely are duplicates. In agreement with the results of other investigators, A. longicuspis and A. sativum clones were clustered together with no clear separation, suggesting these species are not genetically or specifically distinct. The topology of AFLP, RAPD, and isozyme dendrograms were similar, but RAPD and isozyme dendrograms reflected less and much less polymorphism, respectively. Comparison of unweighted pair group method with arithmetic averaging (UPGMA) dendrograms of AFLP, RAPD, and isozyme cluster analyses using the Mantel test indicated a correlation of 0.96, 0.55, and 0.57 between AFLP and RAPD, AFLP and isozyme, and RAPD and isozyme, respectively. Polymorphic AFLPs are abundant in garlic and demonstrated genetic diversity among closely related clones which could not be differentiated with RAPD markers and isozymes. Therefore, AFLP is an additional tool for fingerprinting and detailed assessment of genetic relationships in garlic. longicuspis clones were discriminated morphologically from A. sativum clones by generally having exerted and purple anthers, higher flowering rate and seed production, and smaller bulbils in inflorescence (Etoh and Simon, 2002). The species have the same karyotype and A. sativum and A. longicuspis clones have been crossed successfully (Etoh, 1984; Pooler and Simon, 1994). Garlic has been propagated by vegetative means for centuries and the presence of many closely related or duplicated garlic clones is likely in germplasm collections. Large numbers of molecular markers are needed to discriminate closely related clones or to identify duplicates. One particularly useful tool to achieve this goal is amplified fragment-length polymorphism (AFLP), which could produce high multiplex informative mark- ers in a single reaction and, therefore, could generate sufficient markers to assess genetic diversity among garlic clones. AFLP is a PCR-based DNA fingerprinting technique (Vos et al., 1995) and like randomly amplified polymorphic DNA (RAPD), it generates primarily dominant markers. Neither AFLP nor RAPD require any probe or sequence information (Barker et al., 1999) and they can be applied to any species after some minor modifications. Isozymes, RAPD, and AFLP have been used extensively for determining genetic diversity and relatedness within and between plant species (Knerr et al., 1989; Sharma et al., 1996; Spooner et al., 1996; Tohme et al., 1996; Zhu et al., 1998). Garcia-Mas et al. (2000) compared AFLP, RAPD, and RFLP for measuring genetic diversity in melon and found that AFLP, RAPD, and restricted fragment length polymorphism (RFLP) markers were equally informative, although AFLP de- tected polymorphism with highest efficiency. Recieved for publication 11 Apr. 2002. Accepted for publication 3 Dec. 2002. We thank J. Bradeen and D. Spooner for their valuable comments on the manuscript, and D. Senalik for technical assistance. 1 Current address: Dept. of Horticulture, Uludag Univ., Bursa, Turkey. 2 Corresponding author; e-mail psimon@facstaff.wisc.edu. The genus Allium consists of at least 500 species (Vvedensky, 1944) and garlic (Allium sativum L.) is the species in this genus under cultivation for the longest time. Garlic has been cultivated for at least 5000 years (Hahn, 1996) presumably having origi- nated in central Asia (Hong and Etoh, 1996; Vavilov, 1951). Garlic is an asexually propagated crop, and it displays great morphological diversity in bulb and leaf size, color and shape, scape presence and height, and flower color, fertility, and bulbil (topset) development in inflorescence (Pooler and Simon, 1993). Until recently, characterization of garlic germplasm has been based primarily on morphological data (Jones and Mann, 1963). However, morphological characters may differ under varying environmental conditions. This can complicate characterization of garlic clones (Al-Zahim et al., 1997; Bradley et al., 1996). To eliminate some of these complications, isozyme and RAPD markers have been used to reveal genetic diversity in garlic germplasm (Al-Zahim et al., 1997; Bradley et al., 1996; Lallemand et al., 1997; Maass and Klaas, 1995; Pooler and Simon, 1993). Flower-related morphological traits were correlated with isozyme data, while bulb-associated traits and geographic origin were not (Pooler and Simon, 1993). Furthermore, isozyme and RAPD studies indicated that A. longicuspis Regl., the proposed ancestor of garlic, is not a separate species from A. sativum (Al-Zahim et al., 1997; Maass and Klaas, 1995; Pooler and Simon, 1993). A.