Low Genetic Diversity Indicates the Need to Broaden the Genetic Base of Cultivated Watermelon

Genetic diversity and relatedness were assessed among 46 American cultivars of watermelon (Citrullus lanatus var. lanatus), and 12 U.S. Plant Introduction accessions (PIs) of Citrullus sp. using 25 randomly amplified polymorphic DNA (RAPD) primers. These primers produced 288 distinct reproducible bands that could be scored with high confidence among cultivars and PIs. Based on the RAPD data, genetic similarity coefficients were calculated and a dendrogram was constructed using the unweighted pairgroup method with arithmetic average (UPGMA). The cultivars and C. lanatus var. lanatus PIs differentiated at the level of 92% to 99.6% and 88% to 95% genetic similarity, respectively. In contrast, the C. lanatus var. citroides, and C. colocynthis PIs were more divergent and differentiated at the level of 65% to 82.5% and 70.5% genetic similarity, respectively. The low genetic diversity among watermelon cultivars in this study emphasizes the need to expand the genetic base of cultivated watermelon. diversity and relatedness were examined among U.S. Plant Introduction accessions (PIs) of C. lanatus var. lanatus, C. lanatus var. citroides, and C. colocynthis using RAPD analysis. RAPD markers were also used in the construction of an initial genetic linkage map for watermelon (Hashizume et al., 1996), and to determine genetic relatedness among Asian watermelon cultivars and breeding lines (Lee et al., 1996). Three hundred and fourteen American cultivars are stored at the USDA, ARS, National Seed Storage Laboratory (NSSL, Fort Collins, Colo.), and are considered an essential germplasm resource for watermelon breeding programs. Among them are ‘Allsweet’, ‘Au-Producer’, ‘Charleston Gray’, ‘Crimson Sweet’, ‘Jubilee’, and ‘Peacock’, that are grown throughout the world and are widely used as parents for many hybrids. Currently, there are no molecular data assessing the genetic relatedness and diversity among these cultivars. The objectives of this study were to: 1) estimate genetic relatedness and diversity among American cultivars; and 2) compare their genetic diversity with that in PI accessions of Citrullus. Materials and Methods Plant material. Forty-six watermelon cultivars (Table 1) and twelve PIs representing the three major Citrullus groups (C. lanatus var. lanatus, C. lanatus var. citroides, and C. colocynthis) (Table 2) were evaluated. Seeds of cultivars were obtained from commercial seeds companies, and the USDA NSSL in Fort Collins, Colo. (Table 1). Seeds of all PIs were obtained from the USDA Plant Genetic Resources Conservation Unit in Griffin, Ga. All plants were grown in the greenhouse. Young leaves were collected from four to five, 3week-old-plants of each watermelon accession and stored at –80 °C. Marker data collection. DNA was isolated from young watermelon leaves as described by Levi and Thomas (1999). One hundred and thirty eight primers (60% to 80% GC content) were screened for polymorphism using the cultivars Blackstone and Stone Mountain #5. These two cultivars were chosen for the initial screening of primers because of their difference in parentage (Table 1). RAPD-PCR reactions were performed as described by Levi et al. (1993). Amplification products were separated by electrophoresis in 1.4% agarose gels in × 0.5 Tris borate buffer (Sambrook et al., 1989). The gels were stained with 0.5 mg·mL ethidium bromide solution for 30 min and destained for 15 min in distilled water. DNA fragments were visualized under UV light and photographed using a still video system (Gel Doc 2000; Bio-Rad, Hercules, Calif.). The molecular weights of amplification products were calculated using the “1-Kb Plus DNA Ladder” standards (Gibco BRL, Gaithersburg, Md.). Data analysis. A pairwise similarity matrix was generated using the Nei-Li similarity index (Nei and Li, 1979) according to the equation: Similarity = 2 Nab / (Na +Nb), where Received for publication 27 Nov. 2000. Accepted for publication 5 May 2001. We thank Susan Fox and Laura Pence for their technical assistance. We also thank G. Elmstrom, R. Fery, J. Thies, and J. Staub for reviewing this manuscript. The use of trade names in this publication does not imply endorsement by the USDA or the NCARS of the products named, or criticism of similar ones not mentioned. The cost of publishing this paper was defrayed in part by the payment of page charges. Under postal regulations, this paper therefore must be hereby marked advertisement solely to indicate this fact. To whom requests for reprints should be addressed. E-mail address: [email protected] The xerophytic genus Citrullus Schrad. ex Eckl. & Zeyh. belongs to the Cucurbitaceae family. It comprises four known diploid (n = 11) species found in the temperate regions of Africa, Central Asia, and the Mediterranean (Jeffrey, 1975; Whitaker and Davis, 1962). Among these species is C. lanatus (Thunb.) Matsum & Nakai, from which the cultivated watermelon (C. lanatus var. lanatus) originated (Whitaker and Bemis, 1976; Whitaker and Davis, 1962). Watermelon has been cultivated in Central Africa for at least 5000 years, and in Egypt and in the Middle East for over 4000 years. By the 10 century it was introduced to China, which is the world’s greatest producer and consumer of watermelon. By the 13 century, watermelon was grown in Europe, and the crop was introduced into North America during the 17 century (Jeffrey, 1975; Whitaker and Davis, 1962). Major U.S. production areas are in Florida, California, Texas, Georgia, and Arizona. U.S. watermelon production has increased from 1.2 million tons in 1980 to 3.9 million tons in 1999 with a farm value of $270 million [U.S. Dept. of Agriculture (USDA), Agricultural Statistics, 2000]. Over 500 diploid cultivars were developed in the United States during the last two centuries, but there is an ongoing need to improve watermelon, particularly with respect to development of disease and pest resistant cultivars. There is little information regarding the ancestries of many American watermelon cultivars developed during the 19 and early 20 centuries (G.W. Elmstrom, personal communication). Identification of watermelon cultivars and determination of their genetic purity and relatedness relies mainly on fruit characteristics. Molecular markers can be an effective means to determine genetic relatedness among cultivars and among selections used in watermelon breeding programs. In previous studies designed to examine genetic diversity and phylogenetic relationships among watermelon cultivars using isozymes, most isozymes tested produced monomorphic patterns (Biles et al., 1989; Zamir et al., 1984). In contrast, RAPD procedure provided a sufficient number of informative markers that could distinguish among watermelon cultivars (Hashizume et al., 1993; Zhang et al., 1994). In a recent study (Levi et al., 2000), genetic Nab is the number of RAPD fragments shared by two genotypes (a and b) and Na and Nb are the total number of RAPD fragments analyzed in each genotype. A dendrogram was constructed based on the similarity matrix data by applying unweighted pair-group method with arithmetic average (UPGMA) cluster analysis using the Numerical Taxonomic and Multi-Variant Analysis System for PC (NTSYS-PC version 2) (Rohlf, 1993). Results and Discussion Of the 138 primers that were initially screened, only 35 produced polymorphic RAPD patterns. Of these, 25 primers that produced distinct polymorphic bands were used for further analysis with all cultivars and PIs. The 25 primers produced 288 reproducible RAPD bands that ranged in molecular size from 100 to 3000 base pairs (bp) (Fig. 1). Of these bands, 26 were monomorphic for all cultivars and PIs, 9 were polymorphic among cultivars, but were monomorphic for all PIs. One hundred and sixty eight bands were polymorphic among PIs, but were monomorphic for all cultivars, while 85 bands were polymorphic among all cultivars and PIs (Table 3). Table 1. Watermelon cultivars evaluated in this study, including source of seeds, parental background, year of introduction, and fruit characteristics. Source of Breeding Year Fruit Fruit Flesh Flesh Rind Rind Cultivar seeds parentage introduced shape wt color firmness color firmness Maturity Allsweet Sunseeds (Miles x Peacock) x 1972 Long 25 Pink Firm Green w/deep Firm Late Charleston Gray green wide stripes Astrakanski Seed Savers ------------------AU-Golden Producer Hollar Selection from AU-Producer 1993 Globe 20 Light red Firm Light green w/green Firm Mid-early wide stripes AU-Jubilant Hollar Jubilee x PI 271778 1985 Long 25 Light red Firm Light green w/green Firm Midseason narrow stripes AU-Producer Hollar Crimson Sweet x PI 189225 1985 Globe 20 Light red Firm Light green w/green Firm Mid-early wide stripes Black Diamond Sunseeds ----Globe 15 Light red Soft Dark green Medium Early Blackstone Hollar Florida Giant, Fairfax --Globe 15 Light red Soft Dark green Medium Mid-early Calhoun Gray Sunseeds ----Long 20 Light red Firm Light green/gray Firm Late CalSweet Sunseeds ----Oblong 20 Deep red Firm Green w/deep Firm Late green wide stripes Charleston Gray NSL-5267 Africa 8, Iowa Belle,Garrison, 1954 Long 20 Light red Firm Light green/gray Firm Late and NKL&G Hawkesbury, Leesburg Coles Early NSL-5270 --1892 --------------Congo Syngenta (African x Iow Belle) x Garrison 1949 Long 25 Light red Firm Deep green w/dark Firm Late green narrow stripes Crimson Sweet Hollar (Miles x Peacock) x 1963 Globe 20 Light red Firm Light green w/green Firm Mid-early Charleston Gray wide stripes Dixielee Hollar Texas W5, Wilt resistant Peacock, 1979 Globe 20 Deep red Firm Light green w/green Firm Late Fairfax, Summit narrow stripes Dixie Queen Sunseeds --1890 Globe 20 Deep red Firm Light green w/green Firm Late narrow stripes Dunbarton NSL-6637 (African x Iowa Belle) x Garrison 1953 --------------Fairfax Sunseeds Garrison, African, Iowa Belle, 1952 Long 20 Light red Firm Light green w/green Firm Late Leesburg, Hawkesbury narrow stripes Family Fun Syngenta --1973 --------------Garrison NSL-2053 ----Long 20 Light red Firm Light green w/green Firm Late narrow stripes Garrisonian Willhite Africa 8, Iowa Belle, Garrison, 1957 Long 20 Light red Firm Light green w/green Firm Late Hawkesbury, Leesburg narrow stripes Georgia Rattlesnake Seed Savers ------------------Golden Honey Hollar --1954 Globe 12 --Soft --Explosive Early Golden Midget NSL-5288 Mixed strain from Japan Seed Co. 1959 Globe 12 Yellow Soft --Tender Early Hawkesbury Syngenta New Hampshire Midget x 1936 --------------Pumpkin Rind Iopride Syngenta ----Globe 20 Light red Firm Deep green w/dark Firm Midseason green narrow stripes The RAPD marker data were used to construct a genetic similarity matrix among cultivars and PIs (data not shown) based on the Nei-Li estimate of similarity (Nei and Li, 1979). The similarity matrix was used in a UPGMA cluster analysis to produce a genetic similarity dendrogram (Fig. 2). High genetic similarity values (92% to 99.6%) were detected among watermelon cultivars and among PIs of C. lanatus var. lanatus (88% to 95%) which is considered the progenitor of the cultivated watermelon. Lower similarity values were found among the C. lanatus var. citroides and among the wild species C. colocynthis PIs (65% to 82.5%, and 70.5%, respectively) (Fig. 2). In an additional study designed to elucidate genetic diversity in Citrullus sp. (Levi et al., 2000), genetic diversity and relatedness were estimated among 17 C. lanatus var. lanatus PIs, 12 C. lanatus var. citroides PIs, 13 C. colocynthis PIs, and five cultivars (‘Allsweet’, ‘Charleston Gray’, ‘Ironsides’, ‘Mickylee’, and ‘New Hampshire Midget’) using RAPD analysis. In that study, the five watermelon cultivars grouped in a distinct cluster, indicating that they were derived from common parents. However, the low genetic diversity among C. lanatus var. lanatus PIs in the previous study (Levi et al., 2000), and in the present study (Fig. 2) indicates that the lack of genetic diversity among cultivars is due to a narrow genetic base in C. lanatus var. lanatus. Navot and Zamir (1987) found little isozyme variation among watermelon accessions, and suggest that this is due to the domestication of watermelon outside of its center of origin. They base this assumption on the probability that only a small fraction of plants of the progenitor species (a few plants that have desired qualities) were selected and used at the early stages of domestication (Ladizinsky, 1985). Katzir et al. (1996) could not detect any polymorphism among the watermelon cultivars Sugar Baby and Malali (diploids), and Tri-X-313 (triploid) using seven simple sequence repeat (SSR) markers. However, Jarret et al. (1997) could detect genetic diversity among Citrullus PIs using the same SSR markers. In that study, PIs of C. lanatus var. citroides were slightly more divergent than PIs of C. lanatus var. lanatus. Navot and Zamir (1987) considered C. lanatus var. citroides as the wild progenitor of C. lanatus var. lanatus. In contrast with the low DNA polymorphism, extensive variation in morphological characteristics existed among watermelon culTable 1 continued on next page Ironsides NSL-7369 (Leesburg x Hawkesbury) x 1950 --------------Garrison Jubilee Hollar Africa 8, Iowa Belle, Garrison, 1963 Long 25 Light red Firm Light green w/green Firm Mid-season Hawkesbury, Leesburg narrow stripes King and Queen Hollar ----Globe 12 Light red Soft Light green Medium Early Kleckely’s Sweet Seed Savers ----Globe 12 Light red Soft Deep green Medium Early Klondike Syngenta --1959 Long 20 Light red Firm Dark green Firm Late Klondike Striped Sunseeds ----Long 20 Light red Firm Light green w/green Firm Late RS 57 narrow stripes Klondike StripedHollar --1939 Long 20 Light red Firm Light green w/green Firm Late Blue Ribbon narrow stripes Leesburg NSL-7368 Selection from Kleckley Sweet 1936 --------------Melitoplisky Seed Savers ------------------Mickylee Hollar Texas W5, Fairfax, Summit, 1986 Globe 15 Deep red Firm Light green/gray Firm Midseason Graybelle Miles NSL-6688 Dixie Queen x Klondike R-7 1948 --------------Minilee Hollar Texas W5, Fairfax, Summit, 1986 Globe 8 Deep red Firm Light green/gray Firm Early Graybelle New Hampshire Syngenta Favorite Honey x Dakota Sweet 1951 Globe 6 Red Soft Light green/gray Explosive Early Midget Northern Sweet Syngenta --1932 Globe 12 Red Medium Green w/lines Medium Early Parker Willhite F1 ----------------Peacock Hollar --1939 Oblong 20 Red Firm Dark green Firm Late Prince Charles Syngenta F1 1978 Long 20 Red Firm Light green/gray Firm Midseason Sangria Syngenta F1 1985 Long 25 Deep red Firm Green w/deep Firm Late green wide stripes Sugar Baby Sunseeds Selection from Tough Sweets 1955 Globe 15 Orange red Soft Dark green w/dark Medium Early stripes Stone Mountain Hollar --1924 Oval 20 Light red Firm Light green w/green Firm Midseason narrow stripes Stone Mountain #5 Syngenta Stone Mountain x Iowa Belle 1936 --------------Sunseeds Co. (Acampo, Calif.). Seed Savers Exchange (Decorah, Iowa). Information unavailable. Hollar Seeds (Rocky Ford, Colo.). Accession number provided for each of the heirloom cultivars kept at the USDA National Seed Storage Laboratory (Fort Collins, Colo.). NK Lawn & Garden (Chattanooga,Tenn.). Syngenta Seeds (Naples, Fla.). Willhite Seed (Poolville, Texas). Table 1. Continued. Source of Breeding Year Fruit Fruit Flesh Flesh Rind Rind Cultivar seeds parentage introduced shape wt color firmness color firmness Maturity Table 2. Twelve U.S. plant introduction accessions (PIs) examined in the present study, the Citrullus group to which they belong, the country from which they were collected, and their fruit characteristics as described by the Germplasm Resources Information Network (www.ars-grin.gov). Citrullus Country of Rind color Rind color Flesh Fruit Fruit Frit PI # type origin background pattern color diameter shape maturity 162667 lanatus Argentina Medium green Solid Red 15/30 Oblong Midseason 165451 lanatus Mexico Medium green Solid Pink 18/25 Oblong Midseason 169289 lanatus Turkey Dark green Solid Red 36/36 Round Early 185636 lanatus Ghana Medium green Solid White 10/10 Round Late 189316 lanatus Nigeria Dark green Striped White 12/12 Round Late 203551 lanatus U.S.A. Light green Striped White 12/12 Round Midseason 271778 lanatus S. Africa Medium green Striped Yellow 22/27 Oblong Midseason 270564 citroides S. Africa Light green Solid Yellow 20/25 Oblong Early 482251 citroides Zimbabwe Medium green Striped Yellow 12/12 Round Midseason 271779 citroides S. Africa yellow Striped Yellow 30/45 Oblong Midseason 386014 colocynthis Iran Light green Striped White 9/9 Round Early 388770 colocynthis Morocco Light green Striped White 12/12 Round Early tivars used in this study (Table 1). These characteristics included rind color and thickness, fruit shape and size, flesh texture and color, sugar content, seed shape and color, days to fruit maturity, and disease resistance. Most of these characteristics are qualitative traits affected by a single or a few gene mutations (Rhodes and Dane, 1999; Rhodes and Zhang, 1995) that could not be readily detected by the RAPD markers in this study. However, some of these mutations may be detected through a bulked segregant analysis procedure (Michelmore et al., 1991) using a large number of RAPD primers. The watermelon cultivars in the present study are open-pollinated or F1 hybrid diploid types (n = 11). The parentage records for most of these cultivars are incomplete (Table 1). However, records available for some of the cultivars are consistent with the RAPD-based results (Fig. 2). For instance, the gray-green rind and round fruit type cultivars Mickylee and Minilee were developed from sister plants that had the cultivars Fairfax, Summit, and Texas-W5 in their genetic background (Table 1) (Crall, 1986). ‘Mickylee’ and ‘Minilee’ appeared closely related in the present analysis (Fig. 2). The cultivar Garrison contributed to the genetic background of the oblong fruited cultivars Congo, Charleston Gray, and Garrisonian (Table 1). In the present analysis, ‘Congo’, ‘Garrison’, and ‘Garrisonian’ are in the same group, while ‘Charleston Gray’ is in a closely related group (Fig. 2). The cultivar Allsweet has a green-striped rind and oblong fruit (Table 1), and according to the records it Table 3. The nucleotide sequences of RAPD primers used in the present study, and the number of polymorphic (PM), and monomorphic (MM) band markers produced by each primer. Primer names are according to manufacturer’s identification system (Operon Technologies; OP, and University of British Columbia; UBC). Nucleotide Primer Sequence (5'—>3') PM MM OPB-05 TGCGCCCTTC 11 2 OPB-11 GTAGACCCGT 24 OPB-14 TCCGCTCTGG 14 1 OPC-05 GATGACCGCC 18 2 OPC-20 ACTTCGCCAC 15 1 OPD-02 GGACCCAACC 16 OPD-07 TTGGCACGGG 17 OPD-20 ACCCGGTCAC 25 OPE-04 GTGACATGCC 19 OPJ-06 TCGTTCCGCA 26 OPJ-13 CCACACTACC 19 OPK-14 CCCGCTACAC 21 OPK-20 GTGTCGCGAG 17 OPT-01 GGGCCACTCA 26 OPT-05 GGGTTTGGCA 29 UBC106 CGTCTGCCCG 32 3 UBC115 TTCCGCGGGC 25 1 UBC137 GGTCTCTCCC 21 UBC147 GTGCGTCCTC 8 UBC149 AGCAGCGTGG 25 UBC152 CGCACCGCAC 22 4 UBC155 CTGGCGGCTG 16 1 UBC157 CGTGGGCAGG 24 1 UBC159 GAGCCCGTAG 10 UBC186 GTGCGTCGCT 28 2 UBC199 GCTCCCCCAC 27 4 UBC212 GCTGCGTGAC 21 1 UBC218 CTCAGCCCAG 20 2 UBC222 AAGCCTCCCC 29 1 UBC228 GCTGGGCCGA 31 Fig. 1. PCR-RAPD patterns (on 1.4% agarose-gel) of watermelon cultivars produced by primer UBC-222. Lanes are: 0) 1-Kb Plus DNA ladder markers (Gibco BRL, Rockville, Md.), 1) ‘Allsweet’, 2) ‘Astrakanski’, 3) ‘AU-Golden Producer’, 4) ‘AU-Jubilant’, 5) ‘AU-Producer’, 6) ‘Blackstone’, 8) ‘Calhoun Gray’, 9) ‘Calsweet’, 10) ‘Charleston Gray’ (NSL-5267), 11) ‘Charleston Gray’ (NK Lawn & Garden Co), 12) ‘Coles Early’, 13) ‘Congo’, 14) ‘Crimson Sweet’, 15) ‘Dixielee’, 16) ‘Dixie Queen’, 17) ‘Dunbarton’, 18) ‘Fairfax’, 19) ‘Family Fun’, 20) ‘Garrison’, 21) ‘Garrisonian’, 22) ‘Georgia Rattlesnake’, 23) ‘Golden Honey’, 24) ‘Golden Midget’. is comprised of ‘Charleston Gray’ (50%), ‘Miles’ (25%), and ‘Peacock’ (25%) (Table 1). The present analysis shows that ‘Allsweet’ is closer to ‘Charleston Gray’ than to the two other cultivars (Fig. 2). The cultivar AUGolden Producer (globular fruit with green, wide-striped rind) is reported to be an orange fleshed mutation of ‘AU-Producer’ (Table 1). Predictably, these two cultivars are closely related (Fig. 2). ‘AU-Producer’ is derived from a cross of ‘Crimson Sweet’ (globular fruit with green-striped rind) with PI 189225 that is reported to have resistance to gummy stem blight (Sowell and Pointer, 1962). In the present study, ‘Crimson Sweet’ is indeed closely related to ‘AU-Producer’. Like ‘Allsweet’, ‘Crimson Sweet’ also is comprised of ‘Charleston-Gray’ (50%), ‘Miles’ (25%), and ‘Peacock’ (25%) (Table 1). ‘Allsweet’ is indeed in the same group with ‘Crimson-Sweet’ and its descendents ‘AU-Producer’ and ‘AU-Golden Producer’ (Fig. 2). The cultivar Blackstone (globular fruit with dark green rind) contains ‘Black-Diamond’ (also known as FloridaGiant) and ‘Fairfax’ in its genome (Table 1). Indeed, ‘Black-Diamond’ and ‘Blackstone’ have similar characteristics (Table 1) and are in the same group (Fig. 2). ‘Dixilee’ is in the same group with ‘Charleston-Gray’ (Fig. 2). A detailed review (data not shown) reveals that these two cultivars are derived from common parents (‘Peacock’, ‘Garrison’, ‘Leesburg’, and ‘Hawkesbury’, Table 1). Not all the results are fully consistent with parental records or with morphological characteristics. For instance, the cultivar Leesburg is a selection from ‘Kleckely-Sweet’ (Table 1). The present analysis shows that these two cultivars are in closely related groups, but are not clustered together (Fig. 2). Also, ‘StoneMountain # 5’ is relatively distant from ‘StoneMountain’ (Fig. 2), which is considered one of its original parents, the other being ‘Iowa Bell’ (Table 1). The cultivar Jubilee is distant from ‘Charleston Gray’ and ‘Garrisonian’ (Fig. 2), which supposedly have similar parental backgrounds (Table 1). ‘Jubilee’ is also distant from its descendent ‘Au-Jubilant’ (Fig. 2). Using a contingency test, we could not detect any significant association between RAPD markers and fruit shape. Therefore, it is likely that the RAPD markers here do not provide the resolution required to elucidate gene loci that control fruit characteristics, but elucidate random parts of the watermelon genome. Another indication for this assumption is that cultivars with different morphological characteristics appear closely related. For instance, the cultivar Calhoun-Gray, which has an elongated fruit with light green-gray rind, is in the same group with ‘Blackstone’ and ‘Black Diamond’ (Fig. 2), which have globular fruit and dark rind (Table 1). Most cultivars in the present study have not been examined in previous studies of genetic relatedness using molecular markers. However, three cultivars (‘Family Fun’,’Sugar Baby’, and ‘New Hampshire Midget’) were examined together with watermelon cultivars and breeding lines developed in Asia (Lee et al., 1996). The genetic distance between ‘Sugar Baby’ and ‘New Hampshire Midget’ in the present study is comparable with that in the previous study. However, ‘Family-Fun’ appeared to be distant from most watermelon cultivars (Fig. 2), and not as closely related to ‘Sugar Baby’ as reported by Lee et al. (1996). This difference may be due to the use of different RAPD primers and RAPD procedures, or it may be due to use of seeds of cultivars that were produced by different sources. Due to the open-pollinated nature of watermelon, different genotypes may occur in a cultivar during seed increase, resulting in different RAPD patterns. To test this possibility, we examined ‘Charleston Gray’ plants from the original seed stock (NSSL) vs. ‘Charleston Gray’ plants from seeds provided by NK Lawn & Garden, Co. (Chattanooga, Tenn.). The RAPD patterns (as shown in Fig. 1) confirmed that these two types are alike, but not identical (Fig. 2). Two distinct RAPD markers (470 and 790 bp) produced by primer OPD-07 and UBC-155, respectively, were present in the original Charleston Gray cultivar (NSSL), but not in the type obtained commercially. The 790 bp marker was unique to the ‘Charleston Gray’ provided by NSSL and was not found in any other cultivar or PI in this study. These differences between two sources of the same cultivar indicate that changes in cultivar genotype may occur during seed increase. Thus, the source of the seeds for each cultivar tested should be taken into account in genetic studies and in DNA fingerprinting of watermelon cultivars. The high genetic similarities among C. lanatus var. lanatus PIs, and among watermelon cultivars indicate that many cultivars developed in the United States over the last two centuries have a narrow genetic background. Therefore, it is essential to broaden the genetic base of the cultivated watermelon to reduce its vulnerability to diseases and insect pests. The recent study of Levi et al. (2000) indicated that the wild species C. colocynthis, which has the widest geographic Fig. 2. Dendogram of watermelon cultivars and U.S. Plant Introduction accessions (PIs) produced by UPGMA cluster analysis of similarity matrix. ‘AU-Golden Producer’ = AU-G. Producer, ‘Charleston Gray’ (NSL-5267) = Charleston Gray, ‘Charleston Gray’ (NK Lawn & Garden Co.) = Chrles.-Gray-N, ‘Georgia Rattlesnake’= Georgia-RS, ‘Klondike Striped Blue Ribbon’ = Klondike-SBR, ‘New Hampshire Midget’ = New-Hampshire-M., ‘Klondike Striped RS 57’ = RS57. distribution, also has the highest genetic diversity among Citrullus species. That study also indicated higher genetic diversity within the wild subspecies C. lanatus var. citroides, than in C. lanatus var. lanatus. Accessions of C. lanatus var. lanatus are preferred in watermelon breeding programs because of their horticultural qualities and their close genetic proximity to the cultivated watermelon. Previous studies indicate that resistance to anthracnose (Boyhan et al., 1994; Sowell et al., 1980), or watermelon mosaic virus (Gillaspie and Wright, 1993) exists among accessions of C. lanatus var. lantus. Although the two former Citrullus types are wild and do not have desirable fruit qualities, they might be an essential source of genes that would confer resistance to major diseases and pests. Resistance to gummy stem blight and Fusarium wilt, which are major diseases of watermelon, may exist among accessions of C. lanatus var. citroides (Dane et al., 1998; Martyn, and Netzer, 1991; Sowell, and Pointer, 1962). In a recent study (Simmons and Levi, 2000), C. colocynthis PIs had high resistance, while all C. lanatus PIs were highly susceptible to whitefly (Bemisia tabaci). The Citrullus germplasm collection at the USDA, Agricultural Research Service, Plant Genetic Resources Conservation Unit (Griffin, Ga.), contains 1480 C. lanatus var. lanatus PIs, but only 102 C. lanatus var. citroides PIs, and 21 C. colocynthis PIs (Germplasm Resources Information Network, www.ars-grin.gov). Thus, in order to broaden the genetic base of watermelon cultivars, further efforts are required to expand the collection of C. lanatus var. citroides and C. colocynthis PIs, and to evaluate additional accessions for resistance to important diseases and pests of watermelon.