Genetic diversity in watermelon (Citrullus lanatus) - Wiley Online Library

Hereditas 147: 142–153 (2010)

Genetic diversity in watermelon (Citrullus lanatus) landraces from Zimbabwe revealed by RAPD and SSR markers C. MUJAJU1, J. SEHIC1, G. WERLEMARK1, L. GARKAVA-GUSTAVSSON1, M. FATIH2 and H. NYBOM1 1

Department of Plant Breeding and Biotechnology, Swedish University of Agricultural Sciences, Balsgård, Kristianstad, Sweden 2 CBM The Swedish Biodiversity Centre, Alnarp, Sweden Mujaju, C., Sehic, J., Werlemark, G., Garkava-Gustavsson, L., Fatih, M. and Nybom, H. 2010. Genetic diversity in watermelon (Citrullus lanatus) landraces from Zimbabwe revealed by RAPD and SSR markers. – Hereditas 147: 142–153. Lund, Sweden. eISSN 1601-5223. Received December 1, 2009. Accepted May 19, 2010. Low polymorphism in cultivated watermelon has been reported in previous studies, based mainly on US Plant Introductions and watermelon cultivars, most of which were linked to breeding programmes associated with disease resistance. Since germplasm sampled in a putative centre of origin in southern Africa may harbour considerably higher variability, DNA marker-based diversity was estimated among 81 seedlings from eight accessions of watermelon collected in Zimbabwe; five accessions of cow-melons (Citrullus lanatus var. citroides) and three of sweet watermelons (C. lanatus var. lanatus). Two molecular marker methods were used, random amplified polymorphic DNA (RAPD) and simple sequence repeats (SSR) also known as microsatellite DNA. Ten RAPD primers produced 138 markers of which 122 were polymorphic. Nine SSR primer pairs detected a total of 43 alleles with an average of 4.8 alleles per locus. The polymorphic information content (PIC) ranged from 0.47 to 0.77 for the RAPD primers and from 0.39 to 0.97 for the SSR loci. Similarity matrices obtained with SSR and RAPD, respectively, were highly correlated but only RAPD was able to provide each sample with an individual-specific DNA profile. Dendrograms and multidimensional scaling (MDS) produced two major clusters; one with the five cow-melon accessions and the other with the three sweet watermelon accessions. One of the most variable cow-melon accessions took an intermediate position in the MDS analysis, indicating the occurrence of gene flow between the two subspecies. Analysis of molecular variation (AMOVA) attributed most of the variability to within-accessions, and contrary to previous reports, sweet watermelon accessions apparently contain diversity of the same magnitude as the cow-melons. Hilde Nybom, Balsgård, Department of Plant Breeding and Biotechnology, Swedish University of Agricultural Sciences, Fjälkestadsvägen 459, SE-291 94 Kristianstad, Sweden. E-mail: [email protected].

Citrullus lanatus, commonly known as watermelon and belonging to Cucurbitaceae, is an important food crop in many African countries. This annual diploid (2n 2x 22) (SHIMOTSUMA 1963) species grows as a vine with a climbing or sprawling growth habit, large green leaves with three to five deep lobes, medium-sized monoecious and often bee-pollinated flowers with short pedicels, medium to large fruit with smooth skin and flesh with a high water content, and oval to oblong seeds of a white, grey, red or brown colour. Watermelon has a centre of diversity in the southern part of the continent which could also be the area of origination (RUBATZKY 2001; DANE and LANG 2004). Two major forms of watermelon occur: C. lanatus var. lanatus, the sweet watermelons and C. lanatus var. citroides, the cow-melons (citron and tsamma types) which, although non-bitter, are not sweet. Citrullus lanatus var. citroides is often cultivated but a diversity of feral forms also exist. By contrast, C. lanatus var. lanatus is only known from cultivation and has emerged as a result of a domestication process involving selection for reddish colour and sweetness. Thus, less variation can probably © 2010 The Authors. This is an Open Access article.

be expected among sweet watermelons compared to cowmelons. In Africa, watermelon cultivation is prevalent in drought-prone, semi-arid areas with an annual rainfall below 650 mm. In these areas, watermelon is grown as a staple food (edible seeds), a dessert (edible flesh), and for animal feed. The fruit can be eaten fresh or cooked. The rind can be pickled or candied, while the seeds are baked or roasted for consumption. Cultivation is based on seedpropagated landraces and farmer varieties that have been integrated with the indigenous knowledge, agricultural practices, food habits and cultural dynamics of the rural communities. Traditionally grown sweet watermelons and cow-melons can be white-, yellow-, orange- or red-fleshed and have different fruit shapes and seed coat patterns including colour variation of both fruit rinds and seeds. Two main types of sweet watermelon are recognized: vulgaris and mucosospermus. The vulgaris types are the most widely cultivated forms and have red-fleshed and sweet fruits whereas the mucosospermus types belong to the egusi watermelon, grown in west Africa, where the DOI: 10.1111/j.1601-5223.2010.02165.x

Hereditas 147 (2010) soft seeds are used as a source of edible oil (JEFFREY 2001). In comparison to the sweet watermelons, cow-melons have a longer shelf life and can be stored for more than a year under shade. The cow-melons are consumed mainly after cooking to produce a meal called Nhopi in the Shona language, and are also used as livestock feed. In a recent study, DAVIS et al. (2007) screened worldwide watermelon germplasm for resistance to powdery mildew, and found that 36% and 15% of the 93 most resistant accessions originated from Zimbabwe and Zambia, respectively. This diversity is essential as it offers the opportunity for production diversification and the development of new farming systems and new quality products (BRUSH 2000). Although still very valuable in traditional agrosystems, C. lanatus is presently, however, regarded as a neglected and marginalized crop species in Africa, and therefore treated as a mandate species for conservation by the Southern African Development Community (SADC) Plant Genetic Resources Centre and the National Plant Genetic Resources Centre Regional Network. For effective conservation of watermelon, it is important to obtain information about genetic diversity within and between accessions. Few such studies have however been conducted in southern Africa, except for an investigation of morphological diversity in landraces of Citrullus in Namibia (MAGGS-KOLLING et al. 2000). Other diversity studies have been conducted on a global scale, based mainly on modern cultivars and United States plant introductions (PIs) of selected accessions from African countries. Thus, LEE et al. (1996) used RAPD markers to estimate genetic diversity among watermelon cultivars, and to construct an initial genetic linkage map for watermelon. JARRET et al. (1997) used SSR markers to determine genetic variation among PI accessions of C. lanatus var. lanatus, C. lanatus var. citroides and the wild species C. colocynthis, and delineated 4 groups at the 25% level of genetic similarity. The largest group contained C. lanatus var. lanatus accessions including the egusi watermelons from Nigeria, the second only wild and cultivated ‘citron’-type or C. lanatus var. citroides accessions, the third an accession tentatively identified as C. lanatus var. lanatus, presumably a hybrid between C. lanatus var. lanatus and C. lanatus var. citroides, and the fourth group consisted of a single accession identified as C. colocynthis. In a RAPD-based study, LEVI et al. (2001a) found low genetic diversity among 46 heirloom cultivars of watermelon and concluded that cultivated watermelon has a narrow genetic base. Furthermore, LEVI et al. (2001b) assessed RAPD diversity in PIs, and found three groups consisting of C. lanatus var. lanatus, C. lanatus var. citroides and C. colocynthis, respectively. Although to a considerable extent based on United States plant introductions originally stemming from southern Africa, the analysed material in the above-mentioned

Genetic diversity in watermelon

143

studies is unlikely to represent the actual variation presently nurtured on traditional African farms where landraces consist of variable but identifiable populations that lack ‘formal’ improvement. Therefore, objectives of the present investigation were to (1) to assess levels of intraand inter-accession diversity in some C. lanatus samples collected in Zimbabwe and to estimate relatedness among these accessions, and (2) to investigate the level of congruence between RAPD and SSR findings. MATERIAL AND METHODS Plant material and DNA extraction Seeds from 10 watermelon accessions were obtained from the National Plant Genetic Resources Center of Zimbabwe (Fig. 1). Each accession consists of a batch of seed obtained from a local farmer who allegedly has harvested this seed from a single plant grown on his farm. The accessions were collected in areas inhabited by two distinct groups of people, the Shona (provinces Mashonaland and Masvingo, cow-melons) and the Ndebele (provinces Matabeleland and Midlands, sweet watermelons). The seeds were germinated at 25°C in a greenhouse at Balsgård in Sweden but two of the sweet watermelon accessions never produced any seedlings. A total of 81 plants were chosen for this study (Table 1). DNA was extracted from young leaf tissue using the E.Z.N.A.TM SP Plant DNA Mini Kit (Omega Bio-Tek, Norcross, GA, USA). DNA concentration was estimated visually using DNA low mass ladder (InvitrogenTM Life Technologies, Carlsbad, CA, USA) and electrophoresis in a 2% agarose gel. RAPD analysis The PCR protocol for RAPD primers used a total volume of 25 μl, containing 0.2 μl of 5 U μl1 Taq DNA polymerase (Amersham Biosciences, Uppsala, Sweden), 3 μl of DNA template (10 ng μl1), 1.0 μl of each primer (5 μM) (Eurofins MWG Operon, Ebersberg, Germany), 16.2 μl dH2O, 0.5 μl of 10 mM dNTPs, 1.6 μl of 25 μM MgCl2 and 2.5 μl of reaction buffer (Thermo Fisher Scientific, Surrey, UK). PCR was performed with a P x 2 Thermocycler (Thermo Hybaid, Ulm, Germany) programmed for 45 cycles of 94°C for 15 s, 36°C for 45 s (with a ramp rate of 0.4°C s-1), and a final extension of 72°C for 1.5 min. The amplified products were separated by electrophoresis in a 1.8% agarose gel, stained with ethidium bromide and photographed under UV illumination. Only clearly visible DNA fragments with a length between 150 and 2200 bp were used as markers. Scoring for the presence or absence of DNA fragments was aided by the use of a 1 kb DNA ladder and a control sample, which was run in triplicate to check for reproducibility. Initially, a total of 27 RAPD

0.46 (0.07) 0.60 (0.06) 0.60 (0.10) 0.38 (0.06) 0.31 (0.07) – 0.60 (0.10) 0.48 (0.09) 0.76 (0.04) 0.63 (0.07) 0.46 (0.08) 0.53 (0.07) 0.56 (0.09) 0.36 (0.08) 0.39 (0.09) – 0.36 (0.10) 0.39 (0.11) 0.45 (0.03) 0.38 (0.07) NB. *Accession 2839 with a single plant was included in cluster analysis and ordination, ** NEI’s expected heterozygosity.

0.30 (0.05) 0.39 (0.04) 0.37 (0.06) 0.24 (0.04) 0.21 (0.05) – 0.37 (0.06) 0.30 (0.06) 0.41 (0.02) 0.37 (0.04) 66.67 88.89 77.78 66.67 55.56 – 77.78 77.78 88.89 100 84.71 79.66 78.66 90.26 94.99 – 73.50 81.62 71.60 71.51 0.25 (0.09) 0.33 (0.09) 0.22 (0.08) 0.20 (0.07) 0.21 (0.08) – 0.21 (0.10) 0.28 (0.09) 0.38 (0.03) 0.29 (0.07) 0.17 (0.06) 0.22 (0.06) 0.15 (0.06) 0.13 (0.05) 0.14 (0.05) – 0.14 (0.07) 0.19(0.06) 0.24 (0.02) 0.19 (0.05) 50.00 63.93 40.98 44.26 40.98 – 39.34 50.82 48.03 45.08 11 11 12 15 13 1 8 10 62 18 CM-2643 CM-2645 CM-2650 CM-2746 CM-2768 *SWM-2839 SWM-2854 SWM-2879 Cow-melon (CM) Group Sweet watermelon (SWM) Group

81.41 73.49 81.51 82.46 82.11 – 80.19 72.60 69.46 68.87

HO HE ** %JSC

%PL

SSR

I HE ** %PL %JSC

The PCR protocol for SSR primers used a total volume of 10 μl, containing 0.1 μl of 5 U μl-1 Taq DNA polymerase (Amersham Biosciences, Uppsala, Sweden), 0.1 μl of each primer (100 μM) (Eurofins MWG Operon, Ebersberg, Germany), 7.1 μl dH2O, 1 μl of DNA template (10 ng μl1), 0.2 μl of 10 mM dNTPs, 0.4 μl of 25 mM MgCl2 and 1 μl of reaction buffer (Thermo Fisher Scientific, Surrey, UK). PCR was performed with a VWR Unocycler (VWR, Stockholm, Sweden), programmed as: 94°C for 4 min, 34 cycles of 30 s at 94°C, 30 s at the appropriate annealing temperature (Table 2), 30 s at 72°C, and a final extension of 7 min at 72°C. 4.0 μl of the reaction volume from ten randomly selected samples was checked for successful amplification on 2% agarose gels with subsequent visualization of fragments using UV illumination. Nine SSR primer pairs originally published by JOOBEUR et al. (2006) were chosen based on recommendations by Brita Dahl Jensen of the Dept of Agricultural Sciences, Univ. of Copenhagen (Table 4). For proper separation of fragments and their size determination, primers were fluorescently labeled at the 5´-end with either FAM (MCP1-07, MCP1-13, MCP1-21, MCP1-32, MCP1-37) or HEX (MCP1-03, MCP1-12, MCP1-14, MCP1-28). The PCR products were separated and analysed with capillary electrophoresis on a 3730 DNA Analyser (Applied Biosystems, Carlsbad, CA, USA). The size of the amplified products was calculated based on an internal standard

NPL

SSR analysis

Accession no. (NPGRC)

primers were screened using four DNA samples. Ten primers were used on the entire material (Table 2).

RAPD

Fig. 1. A map of Zimbabwe showing collection sites. Provinces indicated are: MSV – Masvingo, MTS – Matabeleland South, MDL – Midlands, MSC – Mashonaland Central and MSW – Mashonaland West.

Hereditas 147 (2010)

I

C. Mujaju et al. Table 1. Within-accession genetic variation of watermelon (CM cow-melon, SWM sweet watermelon) collected in Zimbabwe, estimated as mean value for Jaccard’s similarity coefficient (%JSC), percentage polymorphic bands/alleles, expected heterozygosity (HE ), observed heterozygosity (HO) and Shannon’s index (I). Standard errors are indicated in parenthesis. NPL is number of plants sampled.

144



Table 2. Nucleotide sequences of RAPD primers used in the present study, number of polymorphic (PM) and monomorphic (MM) bands produced by each primer, PIC values and marker index values. Primer

Nucleotide sequence (5′—3′)

Table 3. Partitioning of genetic variation using GST and AMOVA on both RAPD and SSR data taking into account (a) no prior grouping of accessions, and (b, c) grouping into two major forms (cow-melons and sweet watermelons)

PM

MM

PIC

RMI*

Source of variation

16 6 16 14 15 11 10 12 11 11 122

0 2 1 3 0 6 0 1 2 1 16

0.47 0.53 0.65 0.65 0.68 0.70 0.71 0.72 0.73 0.77

7.52 3.18 10.40 9.10 10.20 7.70 7.10 8.64 8.03 8.47

(500 ROXTM Size Standard) with GeneMapper® Software ver. 3.0 (Applied Biosystems, Carlsbad, CA, USA). A manual binning step was included to assign all detected alleles to repeat unit equivalents.

(a) Partitioning all accessions GST ΦST (b) Partitioning with two major forms of cow-melons and sweet watermelons Between group diversity (AMOVA) Between accessions within groups (AMOVA) Within accession diversity (AMOVA) (c) Partitioning per each major form Cow-melons GST ΦST Sweet watermelons GST ΦST

Data analysis

*Significant at 1%, P 0.01

OPT-01 OPE-04 OPK-14 OPD-20 OPK-20 OPC-05 OPB-11 OPJ-13 OPT-05 OPJ-06 Total

GGGCCACTCA GTGACATGCC CCCGCTACAC ACCCGGTCAC GTGTCGCGAG GATGACCGCC GTAGACCCGT CCACACTACC GGGTTTGGCA TCGTTCCGCA

*RAPD marker index.

RAPD data Each RAPD band was considered as an independent locus, and polymorphic bands were scored as absent (0) or present (1) for all the 81 plants. Polymorphic bands were then used in the subsequent analyses. To evaluate the informativeness of each RAPD primer, a polymorphic index content (PIC) was calculated for each band according to SMITH et al. (1997), as follows: PIC 1 ∑Pi2, where Pi is the band frequency of the i-th allele. A marker index for each of the RAPD primers was obtained by multiplying PICvalue by number of polymorphic loci. A pairwise genetic similarity matrix was generated using Jaccard similarity coefficient (WEISING et al. 2005). Variation within accessions was estimated with four different parameters: (1) mean percentage polymorphic bands, (2) mean Jaccard similarity, (3) the expected heterozygosity which is equivalent to Nei’s unbiased gene diversity HS (NEI 1978) when calculations are based on polymorphic and biallelic loci, and when sample sizes are equal among populations, and the Shannon diversity index (WEISING et al. 2005). Variation among accessions was calculated as the coefficient of genetic differentiation GST (equivalent to the fixation index FST for biallelic loci) according to the formula GST (HT – HS)/HT where HT is the total genetic diversity and HS is the mean within-accession diversity (NEI 1977). Gene diversity parameters were obtained using POPGENE ver. 1.32 (YEH et al. 1997), assuming Hardy-Weinberg equilibrium since watermelon plants

145

RAPD

SSR

0.48 0.47*

0.12 0.11*

43.70%* 17.22%*

0.83%* 10.02%*

39.08%*

89.16%*

0.34 0.34*

0.11 0.10*

0.14 0.12*

0.11 0.10*

have mainly unisexual flowers and are expected to be outcrossing to a high degree. Analysis of molecular variance (AMOVA) using Arlequin ver. 3.0 (EXCOFFIER et al. 2005) was calculated to partition genetic variation at different levels; between sweet watermelons and cow-melons, and between and within accessions. Levels of similarity (relatedness) among and within accessions were quantified with an UPGMA (unweighted pair-group method using arithmetic averages) cluster analysis using NTSYS-pc, ver. 1.80 (ROHLF 1993). Distortion was estimated with a cophenetic correlation analysis between the Jaccard similarity matrix and a similarity matrix generated from the dendrogram. SSR data For single-locus evaluation of the SSR data, alleles at each locus were assigned letter codes. PIC values for all loci were calculated based on the allele frequencies. POPGENE was used to calculate percentage polymorphic alleles within accessions, expected heterozygosity HE, observed heterozygosity HO, and the Shannon index. GST values (weighted average of FST for all alleles) were calculated for differentiation among accessions. Several AMOVAs were calculated to estimate the partitioning of genetic variation at different levels. Alternate homozygotes were assigned as 1 or 0, and heterozygotes were given a value of 0.5 following

146

C. Mujaju et al.


Table 4. Description of SSR loci used in the study, and PIC values. Designation in ref*forward/ reverse

5’ to 3’ F- sequence3’ to 5’ R- sequence

MCPI-07-M13F MCPI-07-R MCPI-32-M13F MCPI-32-R MCPI-28-M13F MCPI-28-R MCPI-03-M13F MCPI-03-R MCPI-12-M13F MCPI-12-R MCPI-37-M13F MCPI-37-R MCPI-21-M13F MCPI-21-R MCPI-13-M13F MCPI-13-R MCPI-14-M13F /2MCPI-14-R

GGTTATGGCCATCTCTCTGC GAGAGTGGGCGTAAGGTGAG AAGGCTGCAGAGACCATGAC AATGATGAAGAACGGGCAAG AATGTTAAGCAGTAAGCACATGG ACACCGGAGAAGGTGAATTG GCATAAACCACCTGTGAGTGG ATGGCTTTGCGTTTCATTTC GGAGTAGTGGTGGAGACATGG TCCTTTCTCTTTCGCAAACTTC AATCTTCCCCATGCCAAAAC GACTTCCAAACCCTCCCTTC AAAGTTTTCATGCCAACGTATC TCAGCCAATATGGTCAAATAGC TTCCTGTTTCATGATTCTCCAC TCAGAATGGAGCCATTAACTTG TCAAATCCAACCAAATATTGC GAGAAGGAAACATCACCAACG

AT(oC)

AN

Fragment size

PIC

(AAG)9

55

3

236/253/255

0.39

(AAG)5(ATC)8

55

3

265/268/271

0.77

(AAG)9

55

3

273/282/283

0.79

(TG)8

55

4

195/200/215/220

0.80

(AAG)7N69(AT)26

55

5

154/170/230/233 /247

0.80

(AAT)9

55

5

123/165/177/192/220

0.87

(AG)11

55

5

181/184/192/195 /200

0.88

(AG)25

55

7

0.88

(AAT)15

55

8

208/210/214/216 /218 /220/222 227/242/255/257 /261 /274/282/285

SSR motif

0.97

NB: AT annealing temperature and AN allele number. *SSR markers described by JOOBEUR et al. (2006).

STAUB et al. (2000). SSR fragments were also scored phenotypically as multilocus profiles, and Jaccard similarity was used to produce a similarity matrix from which an UPGMA cluster analysis was constructed, and the distortion effect estimated with a cophenetic correlation analysis. Both marker types Correlation between the two separate Jaccard similarity matrices with RAPD and SSR data, respectively, was investigated with a Mantel test (MXCOMP in NTSYS-pc, using 9999 permutations to compute the significance of a given correlation). In addition, a multidimensional scaling analysis (MDS) was applied to another Jaccard similarity matrix containing the combined RAPD and SSR data. While clustering methods show a hierachical, categorical structure which is inherently incapable of describing gradients or multiple patterns in data (CRISP and WESTON 1993), ordinations are designed to reveal multiple, continuous, and overlapping patterns of variation (SNEATH and SOKAL 1973) and are most appropriate under a nonhierarchical model of infraspecific variation (SWOFFORD and BERLOCHER 1987).

RESULTS RAPD analyses The 10 RAPD primers used in this study produced 138 scorable RAPD markers of which 122 (88.4%) were

polymorphic (Table 2). These markers ranged in molecular weight from approx. 150 to approx. 2200 base pairs (bp), and PIC values for RAPD primers ranged from 0.47 (OPT-01) to 0.77 (OPJ-06). RAPD marker index values ranged from 3.18 (OPE-04) to 10.40 (OPK-14). Four different estimators of within-accession variation were calculated (Table 1), ranging from 39.3 to 63.9 for percentage polymorphic bands, 72.6 to 82.5% for mean Jaccard similarity, 0.13 to 0.22 for expected heterozygosity, and 0.20 to 0.33 for Shannon’s index. The three most diverse accessions according to all of these estimators were CM2645, SWM2879 and CM2643, whereas the order among the four less variable accessions varied between the different estimators. When calculations were performed across all cow-melon accessions, and all sweet watermelon accessions, respectively, variability was only slightly lower for sweet watermelon in spite of being represented by only two accessions as compared to five for cow-melon. Analysis of molecular variance (AMOVA) within and among seven accessions of watermelons divided into two major groups of cow-melons and sweet watermelons (Table 3) revealed that 43.7% of the total variation resides between these two groups, 17.2% between accessions within groups and 39.1% within accessions. The overall GST for estimating between-accession differentiation regardless of major grouping was 0.48, i.e. very similar to the AMOVA ΦST value of 0.47. GST and AMOVA ΦST values obtained in calculations carried out separately for the two major groups, showed more differentiation among

Hereditas 147 (2010) cow-melon (GST and ΦST 0.34) than sweet watermelon accessions (GST 0.14 and ΦST 0.12). Results of the cluster analysis were illustrated in a dendrogram (Fig. 2). The cophenetic correlation between the genetic similarity matrix and the dendrogram was 0.93, suggesting a very high goodness of fit (ROHLF 1993). Two major clusters were differentiated at 37% genetic similarity: one larger cluster containing the five cow-melon accessions and one smaller cluster with the three sweet watermelon accessions. Within the sweet watermelon cluster, one larger subcluster contained all samples of SWM2854 (Matabeleland South) together with some samples of SWM2879 (Midlands) and the single sample of SWM2839 (Midlands) while the smaller subcluster contained the remaining samples of SWM2879. The neighbouring provinces Midlands and Matebeleland South have a similar climate and trading between these areas is frequent. The cow-melon cluster contained samples of five different accessions, two from Masvingo (CM2645 and CM2650), two from Mashonaland Central (CM2643 and CM2746) and one from Mashonaland West (CM2768). While Mashonaland Central and Mashonaland West are adjacent, they are both more than 450 kilometers from Masvingo (Fig. 1). One of the accessions from Masvingo, CM2645, formed a distinct subcluster splitting off at 54% similarity from the remainder. The other four accessions clustered together showing weak differentiation except for accession CM2643 for which all but one of the samples formed a single subcluster.


147

variation, just as in the RAPD analysis, except for number of polymorphic alleles which was higher for sweet watermelons. Analysis of molecular variance (AMOVA) within and among seven accessions of watermelons divided into two major groups of cow-melons and sweet watermelons (Table 3) revealed that only 0.8% of the total variation resides between these two groups, 10% between accessions within groups and 89.2% within accessions. The overall GST for between-accession differentiation was 0.12, very similar to the AMOVA ΦST value of 0.11. Calculations carried out separately for differentiation among cow-melon and among sweet watermelon accessions, respectively, produced almost identical values (GST 0.11 and ΦST 0.10) for the two data sets. The cophenetic correlation between the genetic similarity matrix and the cluster analysis was 0.96, suggesting a very high goodness of fit. The same two groups as in the RAPD-based analysis (Fig. 4) were retrieved; sweet watermelons differentiated from cow-melons at 15% genetic similarity. Within the cow-melons, CM2645 formed a distinct subcluster just as in the RAPD-based analysis (Fig. 3). Similarly, CM2643 from Mashonaland Central was also relatively well differentiated from the others. Contrary to the RAPD analysis, two of the other three cow-melon accessions were also well-delimited, with only one overlapping sample. Both marker types

SSR analyses A total of 43 SSR alleles were observed with an average of 4.78 alleles per locus, and with a PIC index ranging from 0.39 (MPCI-07) to 0.97 (MPCI-14) (Table 4). In contrast to the RAPD analysis, several samples shared band profiles at all investigated SSR loci, producing 10 inseparable groups with 2 to 7 samples each (Fig. 4). Again, four different estimators of within-accession variation were calculated (Table 1). Values for percentage polymorphic alleles varied from 55.6 to 88.9 and for mean Jaccard similarity from 73.5 to 95.0%. Values for expected heterozygosity, 0.21 to 0.39, were lower than values for observed heterozygosity (0.36 to 0.56) for each accession except SWM2854. Finally, the Shannon index varied from 0.31 to 0.60. There was less coherence between the different estimators for SSR-based variation compared to in the RAPD analysis. Still, two accessions were indicated as more variable than the remainder, namely CM2645 and CM2650. CM2645 had the highest variability also according to RAPD analysis. Similarly, CM2746 instead had very low variability according to both RAPD and SSR. When the whole cow-melon group was compared to the sweet watermelons, the latter showed slightly lower

A Mantel test demonstrated a highly significant correlation between the RAPD and SSR datasets, r 0.848 (P 0.001). Multidimensional scaling, conducted on combined RAPD and SSR data, produced the same two major clusters as the dendrograms. The sweet watermelon accessions were completely intermingled, as also all the cow-melon accessions except for those belonging to CM2645 which had consistently lower values on the first component and thus took an intermediate position between the sweet watermelon samples at one side, and the remaining cow-melon samples at the other side. Morphological variation Watermelon variation was also observed in the greenhouse where three individuals per accession (only one in accesFig. 5). sion SWM2839) were allowed to grow to fruition (Fig In the cow-melon group, the three plants of accession CM2643 showed great uniformity in fruit shape and colour, consistent with the well-defined subcluster in both RAPDand SSR-based dendrograms. Accessions CM2746 and CM2768 were each divided into two separate subclusters in the RAPD dendrogram but the fruiting plants happened

148

C. Mujaju et al.


Fig. 2. UPGMA dendrogram of watermelon landraces from Zimbabwe using RAPD data, showing two major clusters, A cow-melons (CM) and B sweet watermelon (SWM). C is a sub-cluster of plants from accession 2645.



149

Fig. 3. UPGMA dendrogram of watermelon landraces from Zimbabwe using SSR data, showing two major clusters, A cow-melons (CM) and B sweet watermelon (SWM). C is a sub-cluster of plants from accession 2645.

to belong to the same subclusters and produced relatively uniform fruits. By contrast, plants of CM2645 (well differentiated according to the RAPD and SSR dendrograms but containing high levels of intra-accession variability) and CM2650 (heterogenous but with the three fruiting

plants belonging to the same subcluster) showed considerable intra-accession variation in rind colour and pattern. For the sweet watermelon group, all fruiting plants belonged to the larger of the two subclusters in both RAPD and SSR dendrograms. Nevertheless, the three plants of SWM2854

150

C. Mujaju et al.


Fig. 4. Three-dimensional plot of watermelon accessions using multi-dimensional scaling on combined RAPD and SSR data. CM refers to cow-melon and SWM to sweet watermelon.

and the single plant of SWM2839 exhibited considerable variation in fruit colour and rind pattern whereas the three plants of SWM2879 were quite similar. DISCUSSION Evaluation of molecular marker methods The number of markers produced by each RAPD primer in this study is relatively high (an average of 14 marker bands

per primer) possibly due both to the fact that the primers had a high GC content (60–80%) which has proven to yield polymorphic band patterns (LEVI et al. 2001a) and to the pronounced differentiation between sweet watermelons and cow-melons. LEVI et al. (2001a) reported an even higher average of 22 markers per band per primer, possibly because C. colocynthis was included in the analysis. The use of RAPD in estimating genetic diversity has been much debated due to potential problems with, e.g.

Fig. 5. Watermelon fruits in the greenhouse at Balsgård in 2008. Accessions represented are 2643 (A, one fruit), 2645 (B, two fruits), 2650 (C, two fruits), 2746 (D, two fruits) and 2768 (E, two fruits) from the cow-melon group, and 2839 (F, one fruit), 2854 (G, three fruits) and 2879 (H, two fruits) from the sweet watermelon group.

Hereditas 147 (2010) reproducibility, primer competition and the inability to distinguish heterozygotes from homozygotes (NYBOM 2004; WEISING et al. 2005). Co-dominantly inherited SSR markers were therefore used as a complement to the RAPD data. These SSR markers were only moderately polymorphic, with 3 to 8 alleles per locus, and low PIC-values indicating that they may not have sufficient discriminatory power. Contrary to RAPD, SSR markers were not able to provide each sample with an individual-specific DNA profile. In spite of spending approximately the same amount of resources (time and money) on the RAPD and SSR analyses, the RAPD-based data appear to yield more information suggesting that this method can be sufficient for assessing genetic diversity, especially in laboratories that have restricted access to technical facilities. Levels of within-accession diversity estimated as Jaccard similarity were rather similar between the two marker types. For all other parameters, values were generally higher for SSR-based data compared to RAPD. The discrepancy was especially large when values for expected heterozygosity and Shannon index were compared. Similar results have been noted in many other reports; SSRbased estimates of expected heterozygosity were 1.7 to 4.6 times as high as estimates obtained with the dominant marker methods RAPD and AFLP in the same plant material (reviewed by NYBOM 2004). The commonly used estimators of differentiation are mathematically tied to estimates of expected heterozygosity (these parameters are negatively correlated). Consequently, GST for differentiation among accessions was much higher for the RAPD-based data compared to the SSR-based data in our study. Similarly, the corresponding AMOVA ΦST estimate was much higher for RAPD compared to SSR (except for among the two sweet watermelon accessions). Previous studies have often reported that values for differentiation are similar between dominant and co-dominant markers when applied to the same plant material (reviewed by NYBOM 2004). However, according to JOST (2008), the interpretation of GST and ΦST as measures of differentiation produces nonsensical results when gene diversity is high. Therefore, the RAPDbased estimations are likely to be more sound than the SSR-based in our study. Differentiation between the two major forms Dendrograms derived from UPGMA cluster analysis and multidimensional scaling indicated strong differentiation between cow-melons and sweet watermelons. In addition, partitioning of variation with AMOVA exhibited significant variation (44% with RAPD and 0.8% with SSR, P 0.01) between these forms. Considerable differentiation between sweet watermelons and cow-melons has been reported also by LEVI et al. (2000, 2001a, 2001b) using


151

RAPD, LEVI et al. (2005) using RAPD and ISSR, JARRET et al. (1997) using SSR, and NAVOT and ZAMIR (1987) using isozymes. In both cluster analyses, CM2645 from Masvingo was distinct from the rest of the cow-melons. Moreover, this accession occupied a position in between of the cowmelons and the sweet watermelons in the MDS analysis, indicating that gene flow may have taken place between the two forms. Further evidence of putative hybridization is obtained from the fact that CM2645 had the highest amount of intra-accession variation according to most of the RAPD- and SSR-based parameters. Sampling of material for our study was carried out in such a way that this particular accession happens to be the one growing closest to the three sweet watermelon accessions. It should, however, be pointed out that both forms occur in all of the sampled provinces, and gene flow therefore could take place anywhere in the watermelon-growing area. Variation within and among accessions Values for expected heterozygosity within watermelon accessions ranged between 0.13 and 0.22 in the RAPDbased data. If seeds had been collected at random in the watermelon fields, even higher levels would probably have been obtained. However, in our study, most plants from the same accession are likely to be either full siblings or half-sibs since the seed batches were collected from fruits of a single plant. In spite of the resulting close relatedness among samples within accessions, our RAPDbased values for expected heterozygosity are relatively high. Mean values for within-population expected heterozygosity reported in a large compilation of studies on wild plant species were, e.g. 0.13 for annuals and 0.20 for short-lived perennials, and 0.12 for selfing, 0.18 for mixed breeding, and 0.27 for outcrossing species (NYBOM 2004). By contrast, the SSR-based values for expected heterozygosity in watermelon, 0.21–0.39, are considerably lower than those reported for wild species, e.g. 0.46 for annuals and 0.55 for short-lived perennials, and 0.41 for selfing, 0.60 for mixed breeding and 0.65 for outcrossing species (NYBOM 2004). Levels of observed heterozygosity in the watermelon accessions varied between 0.36 and 0.53, i.e. only slightly below the grand mean of 0.58 in the compilation of wild species (NYBOM 2004). Contrary to NYBOM (2004) where SSR-based HO values generally were lower than HE, SSR-analysis in watermelon revealed higher values for observed heterozygosity compared to expected heterozygosity in all accessions except SWM2854. Possibly this discrepancy is due to the fact that the watermelon seedlings were not obtained after random mating but instead from single mothers, resulting in decreased values for expected heterozygosity.

152

C. Mujaju et al.

Both marker types indicated significant differentiation between accessions, both when counted across all accessions and when calculated within each of the two major groups, cow-melons and sweet watermelons. The RAPDderived estimates of among-accession differentiation (ΦST 0.47, GST 0.48) were similar to those obtained for wild annual (ΦST 0.62, GST 0.47) or short-lived perennial species (ΦST 0.41, GST 0.32) reported by NYBOM (2004) but higher than values obtained for, e.g. mixed breeding (ΦST 0.40, GST 0.20) and outcrossing species (ΦST 0.27, GST 0.22). The much lower measures of differentiation revealed by SSR compared to RAPD may be artefactual; according to JOST (2008), GST necessarily approaches zero when gene diversity is high, even if subpopulations are completely differentiated. In previous studies, higher levels of genetic diversity have been reported within C. lanatus var. citroides compared to C. lanatus var. lanatus (NAVOT and ZAMIR 1987; JARRET et al. 1997). By contrast, our study indicates that sweet watermelon and cow-melon have similar levels of genetic diversity in Zimbabwe. This variation reflects the heterogeneous nature of landraces compared to uniform commercial varieties examined in previous studies. In southern Africa, watermelon landraces are often grown in the more marginal, risk-prone habitats and ethnological niches for which the improved varieties are not suitable. Here, geneflow mostly depends on informal seed exchanges and farmer practices. Local seed sources, other than the farmers’ own seed, have the advantage that the variety or mixture is usually known to be adapted to the agro-ecological and socio-economic conditions of a given area. Also, farmers continuously select for better watermelon features to mitigate against the effects of a harsh climatic environment. This in turn affects the distinction of particular accessions since the plants are open pollinated and there is rarely an isolation distance practiced onfarm. Moreover, hybridization with non-cultivated forms may also occur; edible watermelons often grow together with weedy forms of watermelons resulting from introgression between cultivated forms of both major groups and wild forms of var. citroides. The existence of watermelon weedy types was also reported in Namibia (MAGGS-KOLLING et al. 2000). Conclusion Both molecular markers confirmed significant differentiation between the two subspecies of Citrullus lanatus, and revealed considerable variation among and within watermelon accessions, for both cow-melon and sweet watermelon types. In domesticated crops, landraces have been, and still are, the primary source of genetic diversity for plant breeding. It is therefore prudent to further explore the organization of landrace diversity and the forces that

Hereditas 147 (2010) shape and maintain within- and among-landrace diversity. Thus, more research should be undertaken to further assess variability within each of the subspecies using more accessions, and investigate possible associations with utility values, geographical origin, and/or socio-economic patterns. Acknowledgements – We thank Åsa Gunnarson for technical help in the laboratory. Funding was received from Nordiska Ministerrådet.

REFERENCES Brush, S. B. 2000. The issues of in situ conservation of crop genetic resources. – In: Brush, S. B. (ed.), Genes in the field. On farm conservation of crop diversity. IPGRI, Rome, IDRC, Ottawa, and Lewis Publishers, Boca Raton, p. 3–28. Crisp, M. D. and Weston, P. H. 1993. Geographic and ontogenetic variation in morphology of Australian waratahs (Telopea: Proteaceae). – Syst. Biol. 42: 49–79. Dane, F. and Lang, P. 2004. Sequence variation at cpDNA regions of watermelon and related species: implications for the evolution of Citrullus haplotypes. – Am. J. Bot. 91: 1922–1929. Davis, A. R., Levi, A., Tetteh, A. et al. 2007. Evaluation of watermelon and related species for resistance to race 1W powdery mildew. – J. Am. Soc. Hort. Sci. 132: 790–795. Excoffier, L., Laval, G. and Schader, S. 2005. Arlequin ver. 3.0: an integrated software package for population genetics data analysis. – Evol. Bioinf. Online 1: 47–50. Jarret, R. L., Merrick, L. C., Holms, T. et al. 1997. Simple sequence repeats in watermelon [Citrullus lanatus (Thunb.) Matsum. & Nakai]. – Genome 40: 433–441. Jeffrey, C. 2001. Cucurbitaceae. – In: Hanelt, P. (ed.), Mansfeld’s encyclopedia of agricultural and horticultural crops. 3. Springer, p. 1510–1557. Jost, L. 2008. GST and its relatives do not measure differentiation. – Mol. Ecol. 17: 4015–4026. Joobeur, T., Gusmini, G., Zhang, X. et al. 2006. Construction of a watermelon BAC library and identification of SSRs anchored to melon or arabidopsis genomes. – Theor. Appl. Genet. 112: 1553–1562. Lee, S. J., Shin, J. S., Park, K. W. et al. 1996. Detection of genetic diversity using RAPD PCR and sugar analysis in watermelon [Citrullus lanatus (Thunb.) Mansf.] germplasm. – Theor. Appl. Genet. 92: 719–725. Levi, A., Thomas, C. E., Keinath, A. P. et al. 2000. Estimation of genetic diversity among Citrullus accessions using RAPD markers. – Acta Hort. 510:385–390. Levi, A., Thomas, C. E., Keinath, A. P. et al. 2001a. Genetic diversity among watermelon (Citrullus lanatus and Citrullus colocynthis) accessions. – Genet. Resour. Crop Evol. 48: 559–566. Levi, A., Thomas, C. E., Wehner, T. C. et al. 2001b. Low genetic diversity indicates the need to broaden the genetic base of cultivated watermelon. – HortScience 36: 1096–1101. Levi, A., Thomas, C. E., Simmons, A. M. et al. 2005. Analysis based on RAPD and ISSR markers reveals closer similarities among Citrullus and Cucumis species than with Praecitrullus fistulosus (Stocks) Pangalo. – Genet. Resour. Crop Evol. 52: 465–472.

Hereditas 147 (2010) Maggs-Kolling, G., Madsen, S. and Christiansen, J. L. 2000. A phenetic analysis of morphological variation in Citrullus lanatus in Namibia. – Genet. Resour. Crop Evol. 47: 385–393. Navot, N. and Zamir, D. 1987. Isozyme and seed protein phylogeny of the genus Citrullus (Cucurbitaceae). – Plant Syst. Evol. 156: 61– 67. Nei, M. 1977. F-statistics and analysis of gene diversity in subdivided populations. – Ann. Hum. Genet. 41: 225–233. Nei, M. 1978. Estimation of average heterozygosity and genetic distance from a small number of individuals. – Genetics 89: 583–590. Nybom, H. 2004. Comparison of different nuclear DNA markers for estimating intraspecific genetic diversity in plants. – Mol. Ecol. 13: 1143–1155. Rohlf, F. J. 1993. NTSYS-PC numerical taxonomy and multivariate analysis system, ver. 2.1. – Exeter Publishing, Ltd., Setauket, NY. Rubatzky, V. E. 2001. Origin, distribution and uses. – In: Maynard, D. N. (ed), Watermelons: characteristics, production and marketing. ASHS Press, Alexandria, Va., p. 21–26.


153

Shimotsuma, M. 1963. Cytogenetic and evolutionary studies in the genus Citrullus. – Seiken Ziho. 15: 24–34. Sneath, P. H. A. and Sokal, R. R. 1973. Numerical taxonomy. – W. H. Freeman. Smith, J. S. C., Chin, E. C. L., Shu, H. et al. 1997. An evaluation of the utility of SSR loci as molecular markers in maize (Zea mays L.): comparisons with data from RFLP and pedigree. – Theor. Appl. Genet. 95: 163–173. Staub, J. E., Danin-Poleg, Y., Fazio, G. et al. 2000. Comparative analysis of cultivated melon groups (Cucumis melo L.) using random amplified polymorphic DNA and simple sequence repeat markers. – Euphytica 115: 225–241. Swofford, D. L. and Berlocher, S. H. 1987. Inferring evolutionary trees from gene frequency data under the principle of maximum parsimony. – Syst. Zool. 36: 293–325. Weising, K., Nybom, H., Wolf, K. et al. 2005. DNA fingerprinting in plants: principles, methods and applications. – CRC PressTaylor & Francis Group, USA. Yeh, F. C., Yang, R. C., Boyle, T. et al. 1997. POPGENE (ver. 1.32): the user-friendly shareware for population genetic analysis. – Mol. Biol. and Biotechnol. Center. Univ. of Alberta, Canada.