A linkage map of spring turnip rape based on RFLP and HARD markers

A linkage map of spring turnip rape (Brassica rapa ssp. oleifera) was conslructed from an F 2 population of a cross J04002 x 5v3402. The map contained 22 RFLP loci, 144 RAPDs, one microsatellite, and one morphological marker (seed colour). All ten B.rapa linkage groups could be identified and the total map distance was 519 cM. A proportion of the markers (13%), most of which were located in two linkage groups, showed segregation distortion.

ntroduction The development of highly polymorphic DNA markers has facilitated the construction of genetic linkage maps. During the last few years linkage maps have been developed for many plant species, e.g. in the genus Brassica for B.oleracea (Slocum et al. 1990, Kianian and Quiros 1992,Landry et al. 1992), B.napus (Landry et al. 1991, Ferreira et al. 1994, Uzunova et al. 1995, and B.rapa (Song et al. 1991, Chyi et al. 1992, Teutonico and Osborn 1994. The most commonly used type of DNA marker in linkage studies has been restriction frag-ment length polymorphism (RFLP). Recently developed marker types based on use of the polymerase chain reaction (PCR) such as random amplified polymorphic DNA (RAPD), have several advantages over RFLPs. RAPD analysis is easy to perform and rapid, and does not require the use of radioactivity. In addition, because only minute amounts of crude template DNA are needed, it is possible to use rapid smallscale DNA extraction methods. A disadvantage is that the dominant nature of RAPD markers can cause problems if an F 2 intercross population is used. In such cases, estimation of recombination frequency is very inefficient between repulsion phase markers (Ott 1985) and, there-fore, two maps including only coupling phase markers have to be constructed.
Existing B.rapa linkage maps are mostly composed of RFLP markers. Our aim here was to construct a linkage map of spring turnip rape ( B.rapa ssp. oleifera) consisting mainly of RAPD markers. RFLP markers were used to integrate our map with the existing B.rapa map (Teutonico and Osborn 1994).

Plant material
The F 2 mapping population was derived by selfpollinating five F, individuals from a cross between two individuals of repeatedly selfed spring turnip rape lines J04002 and 5v3402. The linkage data are mostly based on 77 F 2 individuals; 28 additional plants were scored to confirm linkages between some markers.
DNA of the plants was extracted by a method slightly modified from that of Dellaporta et al. (1983), as described by Tanhuanpää et al. (1993).
RAPD primers were either synthesised on an Applied Biosystems 392 DNA/RNA Synthesizer (Table 1) or purchased from Operon Technologies (Alameda, California, USA). RAPD analysis was performed as described in Tanhuanpää et al. (1995a) with minor modifications. Putat- ive allelism of two RAPD markers was investigated by hybridisation using one of the RAPD bands as a probe.  Microsatellites are simple DNA sequences consisting of repeated nucleotide motifs, and show extensive polymorphism due to the occurrence of different numbers of repeat units. The microsatellites (Table 2) were amplified in PCR using a pair of flanking primers, one primer of each pair labelled with fluorescein. The amplified products were visualised with ALF DNA Sequencer (Pharmacia).
One morphological marker, seed colour, which exhibits dominant inheritance ('brown' dominant over 'yellow'), was scored visually in the F 2 population. Nomenclature RFLP probes and the respective loci ( Fig. 1) were named according to Teutonico  RAPD loci (Fig.l) were named by the primer: self-synthesised primers with plain numbers, and Operon primers with a letter and a number. Different polymorphic markers produced by the same primer were assigned with a small letter following the number of the primer ( Table 3).
The microsatellite marker on the map has the prefix MS.
The nomenclature of ten B.rapa linkage groups (LGI-LG10) follows that on the previous map (Teutonico and Osborn 1994), the groups being identified by the common RFLP loci. Unassigned groups were named with capital letters (A-C, Fig. 1).

Statistical analysis
Because the inbred lines J04002 and 5v3402 contained residual heterozygosity, the F, seed was not uniform. Some marker loci were homozygous in some of the five F, individuals, leading to genetically uniform (with respect to these loci) F 2 progeny which had to be omitted in the linkage analysis. Therefore, the number of segregating individuals within the pooled F 2 population varied from locus to locus.
Goodness-of-fit to the expected F 2 segregation at marker loci was tested by chi-square ana-lysis. Linkage relationships were evaluated by the MAPMAKER 3.0 computer program (Lander et al. 1987). Markers were grouped with a LOD score of 4.0 and a maximumrecombination fraction of 0.4 as linkage criteria. On a few occasions, the LOD score threshold for linkage was decreased to 2.0 to include additional RFLP loci (indicated with a dashed line in Fig. 1 The map was built in two phases. First, a framework map was constructed from data set A, using only those markers that could be ordered with a LOD score difference > 3.0 (in some cases 2.0) in favour of the best map. To build up the final linkage map, all the other markers linked to each group with a LOD score > 4.0 were placed to the side of the closest framework locus (markers from data set A and codominant markers to the left and markers from data set B to the right).

Results
A high level of DNA polymorphism was observed in the mapping population: 67% of the Fig.l. Linkage map of B.rapa ssp. oleifera constructed from the F 2 population of a cross, J04002 x 5v3402. For grouping markers, a LOD score threshold of 4.0 was used, except for TGIHI2 and WGIG6, which were attached to the framework using a LOD score of 2.0 (indicated with a dashed line). For ordering, a LOD score difference >3.0 (wider line) or >2.0 (LGs 4,6, 7, slim line) in favour of the best map was used. Dominant RAPD markers on the framework and on its left side are derived from J04002 (data set A), on the right side from 5v3402 (data set B). Marker distances are shown in centimorgans; for markers not included in the framework, two point map distances between the marker and the nearest framework locus are shown (LG9 includes four markers, 65a, 93a, 147 b, 140d, which did not show linkage to any framework markers but only to markers from data set B). Linkage groups are named after the previous B.rapa RFLP map (Teutonico and Osbom 1994); the orientation of groups LGI, 6, 7 and 8, where only one locus is common with the previous map, is arbitrary. Codominant markers are underlined, loci common with the previous map printed in italics. The nomenclature of markers is described in Material and methods. Loci exhibiting aberrant segregation are indicated with *(P<O.O5), **(P<o.ol) or ***(P<o.ool).
LGlOis split into two parts, which probably represent distal segments of the same chromosome, because in data set B the codominant markers in these segments map to opposite ends of the same linkage group. Groups B and C contain markers from data set B only. 212 AGRICULTURAL AND FOOD SCIENCE IN FINLAND Vol. 5 (1996)   underlined; lower case letters are not used in their name on the map (Fig. I).
-81 RFLP probes and 79% of the 340 RAPD primers tested detected polymorphism between the parents of the cross, J04002 and 5v3402.
Only one (35D) of the 14 microsatellites tested could be used as a marker; the others either detected no polymorphism, could not be inter- Vol. 5 (1996): 209-217. preted, or the primers failed to amplify detectable products ( Table 2). The F 2 population was scored with a total of 26 RFLP probes, 90 RAPD primers, one microsatellite and one morphological marker. The 90 RAPD primers amplified 176 reproducible polymorphic loci, of which 15 exhibited codominant inheritance.
The 114loci in data set A were arranged into twelve linkage groups, 3-16 markers each. In data set B (132 loci) 11 linkage groups with 3-20 markers each were found. Twenty markers in data set A and 27 markers in data set B remained unlinked.
Data set A was used for building the framework map, because all ten major linkage groups identified on the previous map (Teutonico and Osborn 1994) could be found. The framework map consisted of48 markers, 32 showing codominant inheritance. The length of the linkage groups ranged from 6.9 cM to 98 cM, the total map distance being 519 cM.
The final linkage map, with markers from both data sets, was composed of 58 dominant markers from J 04002, 71 dominant markers from 5v3402, 38 codominant markers and one morphological marker (Fig. 1). A total of 18 markers (printed in italics) were common with those of the previous map of Teutonico and Osborn (1994). Three triplets of linked markers were unassigned (groups A-C) and 32 individual loci remained unlinked.
Twenty markers (5 RFLPs and 15 RAPDs) on the final linkage map and seven unlinked markers exhibited distorted segregation (13% in total). Most of the mapped markers with skewed segregation clustered to linkage groups LG2 and LG3 and were distorted towards the J04002 allele. All except one of the distorted RAPD markers in LG2 and LG3 were derived from data set B (dominant allele from 5v3402).

Discussion
In this study, a linkage map of B.rapa ssp. oleifera was built from an F 2 population of the cross J04002 x 5v3402. Mainly RAPD markers were used, and all ten linkage groups of B.rapa could be identified.
Although repulsion phase markers were not used, it was impossible to order all markers accurately; the best order was usually only slightly more probable than the alternatives. There are a couple of explanations for this. First, estimation ofrecombination frequencies (and thus ordering of loci) between dominant markers is more inefficient than between codominant ones (Ott 1985). This holds true especially when the recombination fraction is small, which was the case in some chromosomal segments where markers appeared to cluster.
Second, the residual heterozygosity in the parents resulted in a reduced size of the F 2 progeny for some loci. This sometimes led to situations where the number of common informative loci between individuals was too low for a reliable estimation of recombination frequency. Finally, errors in genotyping may have caused ambiguity in the placement of loci. The inability to order all the loci reliably resulted in a total map length of only 519 cM; the total length of the map of Teutonico and Osborn (1994) was 1785 cM. The clustering of loci to some map positions may reflect suppressed recombination in heterochromatic regions (Roberts 1965). It may, however, also be due to limited resolution of the map. Clustering ofloci has been reported in maps of various different species (e.g. sugar beet, Barzen et al. 1995;Arabidopsis, Reiter et al. 1992;and Lactuca sotiva, Kesseli et al. 1994).
Interestingly, loci with distorted segregation ratios mapped primarily to LG2 and LG3, and were skewed towards J04002 alleles. The clustering of skewed loci may indicate the existence of gametic or zygotic lethal alleles or gametophytic selection, i.e. gametes containing these regions of the J04002 genome were more competitive. Similar findings of skewed clusters have been reported in various plant species, e.g. B.rapa (Chyi et al. 1992, Teutonico and, B.napus (Landry et al. 1991), Hordeum vulgare (Giese et al. 1994), Lactuca saliva (Kes-seli et al. 1994), Beta vulgaris (Barzen et al. 1995) and Medicago saliva (Echt et al. 1993). Our results agreed with those of Teutonico and Osborn (1994) in having a cluster of skewed loci in LG2. This is the first reported linkage map mostly consisting of RAPD loci in B.rapa. Not all the loci could be ordered unambiguously, which, however, does not diminish the value of the map. The loci can later be mapped more precisely in regions of particular interest by analysing more F 2 individuals. The map has already been used to find a QTL for palmitic acid (in LG9, Tanhuanpää et al. 1995 b) and for oleic acid (in LG6, Tanhuanpää et al. 1996), and will be used in future studies.
Our previous work (Tanhuanpää et al. 1996) demonstrates the possibility of transferring RAPD marker information from one cross to another, and thus, the map can provide information for other researchers, too. In that work (Tanhuanpää et al. 1996), we studied the occurrence of a total of 20 markers in two different F 2 populations; one was derived from a cross between one individual from the line J04002 and another individual from the line J 04072; the other population was the same as that used here, i.e. derived from the cross J04002 x 5v3402. Ten of the markers studied were derived from the parent, which differed in the two populations, and in those cases the probability of finding the same marker in the two populations was 40%.