Construction of a Genetic Map of Melon with
Molecular Markers and Horticultural Traits

C. Perin, L. Hagen, and C. Dogimont

INRA, Station de Génétique et d'Amélioration des Fruits et Légumes, BP 94, 84143­Montfavet, France

V. De Conto

Seminis Vegetable Seeds, Mas de Rouzel, Chemin des Canaux, 30900­Nîmes, France

M. Pitrat

INRA, Station de Génétique et d'Amélioration des Fruits et Légumes, BP 94, 84143­Montfavet, France

Additional index words. AFLP, inter-SSR, ISSR

Abstract. We present our strategy for building a saturated reference map of melon (Cucumis melo L.) based upon recombinant inbred lines (RILs) obtained by single-seed descent (SSD) from a 'Védrantais' x PI 161375 cross. The percentage of heterozygosity appears to be relatively close to the expected, and the percentage of skewed markers is not very high in this RIL population. Two types of molecular markers have been used: amplified fragment length polymorphism (AFLP) and inter-simple sequence repeat (ISSR). Both types of markers seem to be randomly located throughout this map, even though ISSR markers appear more frequently at the ends of some linkage groups. The map is based at this time on marker data from 122 RILs, with 301 AFLP, 49 ISSR, and 14 other markers, and is not yet saturated. A surprising result is that this map is relatively short.


Molecular biologists are working incollaboration with plant breedersin order to develop a genetic map saturated with molecular markers. In the case of melon, two maps have been published, but they are not saturated (Baudracco-Arnas and Pitrat, 1996; Wang et al., 1997). Moreover, few horticultural traits have been placed on these maps. Marker-assisted selection (MAS) for monogenic or polygenic characters is not yet very well developed in melon. MAS is of great interest for use in breeding multiple disease-resistant inbreds, and for selection of a large number of fruit characters that are expressed when fruits are mature.

Although different strategies may be used to construct a saturated genetic map, the main factors that must be considered are 1) the type and the number of populations segregating for the horticultural traits and pest resistances of interest and 2) the types of molecular markers available. The second factor includes the uniformity of distribution of markers across the genome, their cost, and their usefulness for merging maps from different parents. We present here the strategy we are following for building a saturated genetic map based on molecular markers.

Populations and horticultural traits

Our main mapping population was obtained from a cross between 'Védrantais', a 'Charentais' type, and PI 161375 (= Songwhan charmi), a line from Korea that possesses multiple disease and insect resistances, and fruit characters markedly different from 'Védrantais'. This cross has been used for a first partial map based on their F2/F3 progeny (Baudracco-Arnas and Pitrat, 1996). Recombinant inbred lines (RILs) were obtained by single-seed descent from the same F2 family used in construction of the partial map. A population of 200 RILs at the F6 to F8 generation was derived by selfing without selection. Heterozygosity is expected to be 2­5 = 3.1% at the F6 generation, i.e., 6 RILs per locus in the population.

The use of RILs as a mapping population has two main advantages. 1) RILs are sufficiently homogeneous to be propagated indefinitely without segregation. They can be scored for different traits such as disease resistance or fruit characters. Their homogeneity over generations allows opportunity for DNA extraction whenever desired so that they may be scored with new molecular markers. This is particularly important for quan

Cucurbitaceae '98

different from the theoretical one due to the fact that the markers are not independent. Where one marker is heterozygous, a linked marker has a higher probability of also being heterozygous. This leads to an excess of observed RILs with four, or more than four heterozygous loci.

Heterozygosity of RILs can also be estimated by the frequency of RILs heterozygous for every marker (codominant AFLP) (Figure 2). There is an excess of homozygosity compared with the theoretical distribution. This suggests the possibility that some regions of the genome are homozygous across the species and have perhaps some means of preventing crossover events.

Our set of RILs from the 'Védrantais' x PI 161375 cross is not greatly biased, is adequately representative of the population of RILs from this cross, and may be used to develop a reliable reference map saturated with different types of markers.

The horticultural and pest resistance traits segregating in this population include the following monogenically and oligogenically controlled traits:

· Monogenically controlled: Fusarium wilt resistance (genes Fom-1 and Fom-2), Aphis gossypii resistance (gene Vat), melon necrotic spot virus resistance (gene nsv), five carpels (gene p), green flesh color (gene gf), spots on the fruit epidermis, sutures on the fruit.

· Oligo- or polygenically controlled: cucumber mosaic virus resistance, and various fruit characters (sugars, organic acids, etc.).

Figure 1. Percentage of RIL, in 'Védrantais' x PI 161375 population, heterozygous for 0, 1, 2, x.. loci (codominant AFLP) compared with the theoretical distribution.

titative traits that are often sensitive to environmental factors. Estimation of phenotypic and environmental variability can be achieved with replications in different locations or different years. 2) There have been several or more meiotic cycles in the development of a RIL. It is, therefore, expected that there are 2¥ more recombination events between two loci in a RIL than in an F2, a backcross (BC), or a doubled haploid population that has only one meiotic event.

A RIL population can in addition be exploited for any residual heterogeneity of interest. For example, when a RIL is heterozygous for a character at the F5 or F6 generation, plants belonging to each type can be selfed for one or two more generations thus enabling one to obtain near isogenic lines for this character (Tuinstra et al., 1997).

RILs do have several disadvantages for mapping: 1) The standard error of the estimation of the recombination fraction (r) is higher than in F2 or BC populations when r is higher than 15 cM. RILs are, therefore, not as precise or useful for unsaturated maps. 2) Dominance effects cannot be estimated where (almost) all loci are homozygous. Only additive and epistatic effects can be estimated. 3) It takes time and often a considerable amount of work to produce a RIL population.

In the case of our 'Védrantais' x PI 161375 population, the percentages of RILs heterozygous for 1, 2, x.. codominant AFLP markers are shown in Figure 1. The observed distribution is slightly

Figure 2. Percentage of markers (codominant AFLP) for which RILs are heterozygous in the 'Védrantais' x PI 161375 population.

Cucurbitaceae '98

Molecular markers

Studies on polymorphism of the melon genome indicate that different types of markers can be used (Katzir et al., 1996; Kovalski et al., 1995; Neuhausen, 1992). It is not yet well known whether these types of markers are evenly (randomly) distributed. There are some instances where different types of markers are not located in the same areas; for instance in wheat, genome D is less polymorphic with Simple Sequence Repeat (SSR) than genomes A and B (Bryan et al., 1997). In melon, restriction fragment length polymorphism (RFLP) and randomly amplified polymorphic DNA (RAPD), AFLP, and SSR markers appear to be randomly distributed (Baudracco-Arnas and Pitrat, 1996; Danin-Poleg et al., 1998a Wang et al., 1997).

We have mainly used two types of markers in this study: AFLP (Vos et al., 1995) and ISSR (Danin-Poleg et al., 1998b; Zietkiewicz et al., 1994). DNA was extracted as previously reported from pools of 10 F6 individuals (Baudracco-Arnas and Pitrat, 1996). Primers for ISSR markers were purchased from British Columbia University. For AFLP, DNA was digested with MseI (frequent cutter) and either HindIII or EcoRI (rare cutters) and ligated to the appropriate adapters. A preamplification step was performed using primers for each adapter plus one 3'-selective nucleotide. For primers complementary to the EcoRI or HindIII adapters, this 3'-selective nucleotide was A (EO1 and HO1, respectively). The primer complementary to the MseI adapter had either a C (MO2) or A (MO1) as the selective nucleotide. The preamplification product was diluted and used as a template for a second amplification using primers with three selective nucleotides each.

Results for AFLP are presented in Table 1. The frequency of allelic markers is very similar for primer combinations of the three preamplifications (average 13.6%). The most efficient combinations were those using the EcoRI­MseI preamplification EO1-MO1, giving an average of 25.7 markers per primer pair.


The first molecular melon map was based on the F2/F3 progeny from the 'Védrantais' x PI 161375 cross using mainly RAPD and RFLP markers (Baudracco-Arnas and Pitrat, 1996). More recently, SSR markers have been added to this map and they seem to be randomly localized (Danin-Poleg et al., 1998a). Another map based on AFLP markers on a different cross has been published (Wang et al., 1997).

Using 122 RILs from the 'Védrantais' x PI 161375 cross, 488 markers were analyzed to generate a map: 411 AFLP, 63 ISSR, 5 RAPD, 1 RFLP, 4 fruit characters genes, and 4 disease resistance genes. Of the 488 molecular markers, 9.2% were skewed at P < 0.01 and 6.4% at P < 0.05. Allelic markers and skewed markers at P < 0.01 were discarded before mapping, which left 364 markers total (301 AFLP, 49 ISSR and 14 other markers) for mapping with MAPMAKER software (Lander et al., 1987).

A map with 17 linkage groups (12 major groups with >10 markers) and 10 unlinked markers was constructed with a LOD score of 4.0 (Figure 3). Linkage groups IX and XIV were merged when the LOD score was reset to 3.0. The total length of the map is 1366 cM. The average distance between the linked markers is 3.9 cM with a distribution as shown in Figure 4. The correlation between the length of each group and the number of markers

Table 1. Summary of AFLP primer pairs (three selective nucleotides each) grouped according to their respective restriction sites and preamplifications (first selective nucleotide each).

AFLP Allelic Markers/

primer Combinations Markers markers combination

pair (no.) (no.) (no.) (%) (mean no.)

HindIII­MseI(H01-MO1) 6 124 18 14.5 20.7

EcoRI­MseI(EO1-MO2) 6 67 9 13.4 11.2

EcoRI­MseI(EO1-MO1) 10 257 34 13.2 25.7

Total 22 448 61 13.6 20.4

Cucurbitaceae '98

Figure 3. Genetic map of melon with AFLP and ISSR markers.
Figure 3 (continued). Genetic map of melon with AFLP and ISSR markers.

groups V, VI, VII, IX, XII) even if most of them are localized within a group.

Discussion and conclusions

This map is not saturated as there are still some small groups and unlinked markers. One of the surprising results is that the total length of the map is low compared with values previously published for more unsaturated maps: 1390 cM (Baudracco-Arnas and Pitrat, 1996) or 1942 cM (Wang et al., 1997). It is expected that the number of recombination events on a RIL population should be twice as high as those of an F2 progeny (Paran et al., 1995). We observe that the distances are lower, reduced almost by half. In the case of group XIII (RILs) and group 7 (F2/F3), the total distance between two phenotypic characters (nsv and p) and two RAPD markers (D08 and R01)(Figure 6) in the RIL population is half that in the F2/F3 population.

We are now locating on the RIL map the RAPD, RFLP and SSR markers that were located on the F2/F3 map. This will allow us to verify these first results on the length of the map and the random distribution of the different types of markers as has been shown in rice and wheat (Chen et al., 1997; Kojima et al., 1998; Maheswaran et al., 1997; Nagaoka and Ogihara, 1997).

To locate horticultural traits on this reference map, F2 and BC populations lines segregating for other monogenic characters not present in the RIL population have been developed from crosses of 'Védrantais' (common parent) with a number of breeding lines:

Figure 4. Distribution of the distances between two markers.

Figure 5. Correlation between the number of markers in each of the main 13 groups and the length of this group (in cM).

(for the main 13 groups) is not very high (r = 0.39) (Figure 5), which indicates that all groups are not marked with the same density. Evenness of the distances between markers can be estimated by the standard error of the distances within each group. For instance, group XII has a standard error of 2.5 and group IV of 4.3.

The correspondence between this map and the previously published ones can be done by some common markers (Baudracco-Arnas and Pitrat, 1996). For instance, Vat in groups V (this map) and 2 (F3/F3 map), Fom-1 in groups XVII and 5, Fom-2 in groups XII and 6, nsv and p in groups XIII and 7. Groups II, VII, IX, XIV, XV, and XVI cannot be assigned to the previous maps. Conversely groups F and H cannot be assigned to the present map.

It seems that the ISSR markers are more localized at the end of each group (for instance in

Figure 6. Comparison of the distances between markers in the RIL map (group XIII) and in the F2/F3 map (group 7).

Cucurbitaceae '98

· disease resistance (powdery mildew, zucchini yellow mosaic virus, papaya ringspot virus);

· sex characters (male sterility, monoecy, gynoecy);

· fruit characters (epidermis color, etc.)

Using bulk segregant analysis (BSA) (Michelmore et al., 1991), PCR-based markers (RAPD or AFLP) close to the horticultural trait will be selected, and they will be mapped on the segregating population. Markers that are polymorphic between 'Védrantais' and PI 161375 will be mapped on the reference map, which will allow us to map the horticultural trait. Conversely, markers present on the reference map in this region of the genome will be tested on the other population in order to find markers more closely linked to the other characters.

Literature cited

Baudracco-Arnas, S. and M. Pitrat. 1996. A genetic map of melon (Cucumis melo L.) with RFLP, RAPD, isozyme, disease resistance and morphological markers. Theor. Appl. Genet. 93:57­64.

Bryan, G.J., A.J. Collins, P. Stephenson, A. Orry, J.B. Smith, and M.D. Gale. 1997. Isolation and characterisation of microsatellites from hexaploid bread wheat. Theor. Appl. Genet. 94:557­563.

Chen, X., S. Temnykh, Y. Xu, Y.G. Cho, and S.R. McCouch. 1997. Development of a microsatellite framework map providing genome-wide coverage in rice (Oryza sativa L.). Theor. Appl. Genet. 95:553­567.

Danin-Poleg, Y., N. Reis, S. Baudracco-Arnas, M. Pitrat, and N. Katzir. 1998a. Integration of simple sequence repeats into melon (Cucumis melo L.) linkage map. Genome (in press).

Danin-Poleg, Y., G. Tzuri, N. Reis, and N. Katzir. 1998b. Application of inter-SSR markers in melons (Cucumis melo L.). Cucurbit Genet. Coop. Rpt. 21:25­28.

Katzir, N., Y. Danin Poleg, G. Tzuri, Z. Karchi, U. Lavi, and P.B. Cregan. 1996. Length polymorphism and homologies of microsatellites in several Cucurbitaceae species. Theor. Appl. Genet. 93:1282­1290.

Kojima, T., T. Nagaoka, K. Noda, and Y. Ogohara. 1998. Genetic linkage map of ISSR and RAPD markers in Einkorn wheat in relation to that of RFLP markers.

Theor. Appl. Genet. 96:37­45.

Kovalski, I., Silberstein, R. Huang, and R. Perl-Treves. 1995. Development of molecular markers for genome analysis in Cucumis melo, p. 191­195. In: G.E. Lester and J.R. Dunlap (eds.). Cucurbitaceae '94: Evaluation and enhancement of cucurbit germplasm. 1­4 Nov. 1994, South Padre Island.

Lander, E.S., P. Green, J. Abrahamson, A. Barlow, M.J. Daly, S.E. Lincoln, and L. Newburg. 1987. MAPMAKER: an interactive computer package for constructing primary genetic linkage maps of experimental and natural populations. Genomics 1:174­181.

Maheswaran, M., P.K. Subudhi, S. Nandi, J.C. Xu, A. Parco, D.C. Yang, and N. Huang. 1997. Polymorphism, distribution, and segregation of AFLP markers in a doubled haploid rice population. Theor. Appl. Genet. 94:39­45.

Michelmore, R.W., I. Paran, and R.V. Kesseli. 1991. Identification of markers linked to disease-resistance genes by bulked segregant analysis: A rapid method to detect markers in specific genomic regions by using segregating populations. Proc. Nat. Acad. Sci. USA 88:9828­9832.

Nagaoka, T. and Y. Ogihara. 1997. Applicability of inter-simple sequence repeat polymorphisms in wheat for use as DNA markers in comparison to RFLP and RAPD markers. Theor. Appl. Genet. 94:597­602.

Neuhausen, S.L. 1992. Evaluation of restriction fragment length polymorphism in Cucumis melo. Theor. Appl. Genet. 83:379­384.

Paran, I., I. Goldman, S.D. Tanksley, and D. Zamir. 1995. Recombinant inbred lines for genetic mapping in tomato. Theor. Appl. Genet. 90:542­548.

Tuinstra, M.R., G. Ejeta, and P.B. Goldsbrough. 1997. Heteregenous inbred family (HIF) analysis: A method for developing near-isogenic lines that differ at quantitative trait loci. Theor. Appl. Genet. 95:1005­1011.

Vos, P., R. Hogers, M. Bleeker, M. Reijans, T. Lee, M. Hornes, A. Frijters, J. Pot, J. Peleman, M. Kuiper, and M. Zabeau. 1995. AFLP: A new technique for DNA fingerprinting. Nucleic Acid Res. 23:4407­4414.

Wang, Y.H., C.E. Thomas, and R.A. Dean. 1997. A genetic map of melon (Cucumis melo L.) based on amplified fragment length polymorphism (AFLP) markers. Theor. Appl. Genet. 95:791­797.

Zietkiewicz, E., A. Rafalski, and D. Labuda. 1994. Genome fingerprinting by simple sequence repeat (SSR)-anchored polymerase chain reaction amplification. Genomics 20:176­183.

Cucurbitaceae '98