A Consensus Linkage Map for Sugi (Cryptomeria japonica) From Two Pedigrees, Based on Microsatellites and Expressed Sequence Tags

Corresponding author: Yoshihiko Tsumura, Forestry and Forest Products Research Institute, Tsukuba, Ibaraki 305-8687, Japan., ytsumu{at}ffpri.affrc.go.jp (E-mail)

Communicating editor: S. MCCOUCH

	ABSTRACT

TOP ABSTRACT MATERIALS AND METHODS RESULTS DISCUSSION LITERATURE CITED

A consensus map for sugi (Cryptomeria japonica) was constructed by integrating linkage data from two unrelated third-generation pedigrees, one derived from a full-sib cross and the other by self-pollination of F₁ individuals. The progeny segregation data of the first pedigree were derived from cleaved amplified polymorphic sequences, microsatellites, restriction fragment length polymorphisms, and single nucleotide polymorphisms. The data of the second pedigree were derived from cleaved amplified polymorphic sequences, isozyme markers, morphological traits, random amplified polymorphic DNA markers, and restriction fragment length polymorphisms. Linkage analyses were done for the first pedigree with JoinMap 3.0, using its parameter set for progeny derived by cross-pollination, and for the second pedigree with the parameter set for progeny derived from selfing of F₁ individuals. The 11 chromosomes of C. japonica are represented in the consensus map. A total of 438 markers were assigned to 11 large linkage groups, 1 small linkage group, and 1 nonintegrated linkage group from the second pedigree; their total length was 1372.2 cM. On average, the consensus map showed 1 marker every 3.0 cM. PCR-based codominant DNA markers such as cleaved amplified polymorphic sequences and microsatellite markers were distributed in all linkage groups and occupied about half of mapped loci. These markers are very useful for integration of different linkage maps, QTL mapping, and comparative mapping for evolutional study, especially for species with a large genome size such as conifers.

TREE breeding is a time-consuming process, mainly because of the long intervals between generations, which has prevented tree breeders from using crossbreeding effectively. However, the presence of many molecular markers and use of quantitative trait locus (QTL) analysis make it possible to construct genetic maps, to detect QTL, and subsequently to perform marker-assisted selection for molecular breeding (STAUB et al. 1996 ). A double pseudo-testcross strategy has generally been adopted for constructing genetic maps in conifers having allogamous characteristics (GRATTAPAGLIA and SEDEROFF 1994 ). Although this strategy exploits one of the characteristics of conifers, namely high heterozygosity within species, the average estimate of gene diversity within species in gymnosperms does not exceed 28.1% (HAMRICK et al. 1992 ). Therefore, QTL analyses using multiple pedigrees should be important for understanding QTL within conifer species. However, difficulties of usage may be encountered when some types of genetic markers, such as randomly amplified polymorphic DNA (RAPD) or amplified fragment length polymorphism (AFLP), are used to function as a "bridge" marker to merge linkage groups or QTL derived from different pedigrees. For map comparisons to be meaningful, a detection of the orthologous locus in each pedigree can be achieved by DNA sequence homology and conserved map location (BROWN et al. 2001 ). Thus, markers based on expressed sequences, such as cDNA-based restriction fragment length polymorphisms (RFLPs) and cleaved amplified polymorphic sequences (CAPS) derived from expressed sequence tags (ESTs), should be used as bridge markers for integrating maps from different pedigrees. Especially in species with large genomes, such as conifers, the signal of a single-copy gene in RFLP analysis is generally weak, and the DNA is not usually well digested because it is methylated (IWATA et al. 2001 ). Therefore, a large number of CAPS markers derived from ESTs are especially valuable as bridge markers between multiple pedigrees. In addition, mapping with multiple populations provides several advantages over mapping based on a single population. In particular, many candidates for bridge loci derived from ESTs can be placed on a single map. Therefore, dense consensus maps including numerous EST-based markers become fundamental tools for comparing linkage groups and QTL derived from different pedigrees.

Sugi (Japanese cedar), Cryptomeria japonica D. Don, is an important forest tree, because of its excellent characteristics, including rapid growth, straight bole, ready regeneration, and soft wood with a pleasant color and scent. Several projects to map the sugi genome have been undertaken on the basis of different marker systems and types of segregating populations, such as a full-sib F₁ population (KURAMOTO et al. 2000 ; NIKAIDO et al. 2000 ) and a three-generation pedigree derived from self-pollination of F₁ individuals (MUKAI et al. 1995 ; IWATA et al. 2001 ). Some loci were identified relating to quantitative traits such as juvenile growth rate, profusion of flowering, rooting ability of cuttings (YOSHIMARU et al. 1998 ), and modulus of elasticity of the wood (KURAMOTO et al. 2000 ). The information contained in these maps is sufficient for integration to correlate the loci identified on them.

In the Pinaceae, intensive genome studies have been conducted on Pinus teada (e.g., SEWELL et al. 1999 ; BROWN et al. 2001 ; TEMESGEN et al. 2001 ). Genome studies have been extended to other pine species, such as P. radiata (DEVEY et al. 1999 ) and P. elliottii (BROWN et al. 2001 ), and have resulted in the partial construction of comparative maps. However, pine genomes are large (e.g., the estimated C-value for loblolly pine is 21–23 pg; WAKAMIYA et al. 1993 ) and contain complex gene families (KINLAW and NEALE 1997 ). The large genome has caused some problems, such as nonidentical allelic association of RFLP patterns (e.g., JERMSTAD et al. 1994 ; DEVEY et al. 1996 ; SEWELL et al. 1999 ). In contrast, C. japonica has several advantages in genome studies relative to the Pinaceae. First, TSUMURA et al. 1997 developed EST markers derived from C. japonica cDNA, used them as markers in PCR amplification in related families, and showed that more than half of these EST markers, including single and multiple fragments, were also targetable under low-stringency conditions in DNA from other members of the Taxodiaceae, Cupressaceae, Sciadopitaceae, and Pinaceae. Second, the estimated C-value for C. japonica is only 11 pg (HIZUME et al. 2001 ), approximately half the genome size of pine. Third, gibberellin treatment promotes flower-bud formation, which accelerates production of the next generation. C. japonica can also be rapidly propagated by cuttings.

This article presents the results of integrating the linkage data from two independent pedigrees into a single consensus map. This consensus map will serve as a fundamental tool for molecular breeding in C. japonica and related species and a basis for studies of genome organization and evolution in conifers.

	MATERIALS AND METHODS

TOP ABSTRACT MATERIALS AND METHODS RESULTS DISCUSSION LITERATURE CITED

Mapping populations:
Two unrelated native cultivars, Yabukuguri and Iwao, were used for the first generation of the first pedigree (referred to as YI). These two cultivars were selected for QTL analysis as to growth patterns. Growth patterns of Yabukuguri and Iwao were slow and quick growers, respectively. Furthermore, Yabukuguri shows a trait for poor male flower fertility. Two unrelated native cultivars, Kumotooshi and Okinoyama, were used for the first generation of the second pedigree (referred to as KO). These two cultivars were selected by Ohba and co-workers (KAWASAKI et al. 1984 ; OHBA et al. 1988 ) as part of an effort to clarify heritable traits of heartwood color (red and black). In the KO pedigree, 73 self-pollinated progeny of the third generation were derived from self-fertilization of an F₁ plant from Kumotooshi x Okinoyama. The KO was previously used to construct a linkage map based on RFLP, RAPD, and isozyme markers and a morphological trait (MUKAI et al. 1995 ); later it was used to add CAPS markers to the linkage map (IWATA et al. 2001 ). Because high segregation distortion ratios were detected when these former linkage maps were constructed (MUKAI et al. 1995 ; IWATA et al. 2001 ), sib-cross strategy of the two F₁ plants of Yabukuguri x Iwao was adapted to obtain 150 full-sib progeny of the third generation in the YI pedigree (Fig 1).

View larger version (28K): [in this window] [in a new window] [Download PPT slide]   Figure 1. Two three-generation pedigrees used for Cryptomeria genetic mapping. The third-generation progeny of YI and of KO were derived from sib-crosses and self-fertilization in the second generation, respectively.

Genetic markers:
For the YI pedigree, four kinds of genetic markers were used to construct the linkage map: 146 CAPS markers, 133 RFLP markers obtained with 119 cDNA probes, 42 microsatellites, and five single nucleotide polymorphisms (SNPs) in three genes. For the KO pedigree, we used 96 CAPS markers, 122 RFLP markers (117 probes derived from cDNA and 3 probes derived from genomic DNA libraries), 33 RAPD markers with dominant manner, one isozyme, and one morphological trait. Of the 96 CAPS markers in KO, segregation data for 68 were obtained, and 46 have already been assigned to positions on the KO linkage map (IWATA et al. 2001 ). We investigated the segregation patterns of 28 additional CAPS markers in the KO pedigree (APPENDIX AA). In all, the segregation patterns of 326 genetic markers in the YI pedigree and 253 in the KO pedigree were determined (Table 1).

The primary source of RFLP probes was the C. japonica cDNA libraries constructed by MUKAI et al. 1995 and UJINO-IHARA et al. (2000). These probes were used in combination with six restriction enzymes (BglII, DraI, EcoRI, EcoRV, HaeIII, and HindIII). The primer pairs for CAPS included those used by TSUMURA et al. 1997 and IWATA et al. 2001 and the continuously developed primer pairs listed in TABLE AA, which were derived from the sequence information of two cDNA libraries (MUKAI et al. 1995 ; UJINO-IHARA et al. 2000 ). Microsatellite markers were derived from three different microsatellite-enriched genomic libraries developed by MORIGUCHI et al. 2003 and N. TANI, T. TAKAHASHI, T. UJINO-IHARA, H. IWATA, K. YOSHIMURA and Y. TSUMURA (unpublished data; listed in APPENDIX ABB). Genotypes of 42 microsatellite loci were determined by electrophoresis on 7.5% polyacrylamide gels (ethidium bromide stain). To determine SNP genotypes, PCR products were purified and sequenced using Big Dye terminator cycle sequencing kits (Perkin-Elmer, Foster City, CA), following the manufacturer's recommendations and the corresponding primer for each gene on an Applied Biosystems (Foster City, CA) model 3100 automated sequencer.

Identification of orthologous markers:
We detected 45 orthologous CAPS markers between the KO and YI pedigrees, which were found by the coincidental existence of polymorphisms between the pedigrees in the second generation. To increase the number of orthologous markers, we screened for polymorphisms in the second-generation individuals of the YI pedigree by using probes for all RFLP markers found on the KO linkage map. We found that 29 RFLP loci yielded polymorphisms between the two full-sib individuals of the second generation of the YI pedigree. In all, 70 orthologous markers were used to integrate the two independent linkage maps.

Genetic linkage analysis and map construction:
Because the segregating generation in the YI pedigree was produced by sib crossing in the second generation, a double pseudo-testcross strategy was adopted for linkage analysis (GRATTAPAGLIA and SEDEROFF 1994 ). Each segregating marker was scored individually for all configuration types defined in Fig 2 (RITTER et al. 1990 ). The segregation ratio of each marker was tested with a chi-square test for goodness of fit to the expected 1:1 ratio when the marker was present in one of the two parents or to the expected 3:1 ratio when the marker was present in both parents. In the first round of analysis, segregation data were used to construct two maps based on meiosis in both F₁ parents, YI96 and YI38. Two data matrices, therefore, were designed to construct the two parental linkage maps. Markers belonging to configuration type b (Fig 2) were ignored in this round of analysis. For markers belonging to configuration type c, the data from heterozygous individuals were ignored, because the parental origin could not be deduced. Parental maps were then constructed with MAPMAKER 3.0 (LANDER et al. 1987 ) using the backcross option. The linkage phase was deduced statistically from two-point linkage data. The highest two-point linkage LOD value indicates the putative correct linkage phase in the reciprocal data set (given phase and reverse phase). Markers were initially associated by using the "Group" command (two-point comparison). For each linkage group, marker orders were then defined by using the "Order" command. Three different orders were compared for three different information values in MAPMAKER (1, 2, and 5 cM) at LOD = 2.0. Other markers were then added with the "Try" command. The "Ripple" command was employed at LOD 2.0 to assess the robustness of the marker order.

View larger version (19K): [in this window] [in a new window] [Download PPT slide]   Figure 2. Informative patterns for mapping as defined by RITTER et al. 1990 . The five configurations presented correspond to segregating loci in the progeny. All the segregating-locus configurations found in an F2 generation match one of these five configurations. Two kinds of loci are shown: loci defined by a single band, for which only one segregating allele is observed, and loci defined by allelic bands, for which the different segregating alleles are observed.

In the second round of analysis, linkage analysis for the YI pedigree was done with JoinMap 3.0 (VAN OOIJEN and VOORRIPS 2001 ), using the parameter set for progeny derived by cross-pollination (CP). In this analysis, all the configuration patterns shown in Fig 2 were used. Two recombination data sets derived from the parental meiosis were analyzed together. Linkage groups were assigned with a minimum LOD threshold of 3.8. Loci that were completely linked were identified and removed from the data set before the marker order within groups was determined. Map distances were calculated with KOSAMBI's (1944) mapping function.

The two data sets were merged for linkage groups that retained markers orthologous to each other. The YI linkage map was also integrated with the previously constructed KO linkage map by using JoinMap 3.0. Highly skewed marker-segregation ratios (P < 0.001) were removed when the integrated map was constructed. The integrated map was constructed on the basis of the mean recombination frequency and the combined LOD scores. The images of the linkage groups were drawn with MAPCHART (VOORRIPS 2002 ).

Estimation of genome length and map coverage:
The estimated genome length G_e was determined from the partial linkage data according to G_e = N(N - 1)X_e/K with a confidence interval of G_e/(1 ± 1.96/), where N is the number of markers and thus N(N - 1) is the number of pairwise comparisons. X_e is the maximum distance between two adjacent markers in centimorgans at a certain minimum LOD score, and K is the number of marker pairs with the same minimum LOD score (HULBERT et al. 1988 ; CHAKRAVARTI et al. 1991 ). A minimum LOD score of 3.8 was chosen to estimate the genome.

To calculate the observed genome length, the total length of the map G_t was calculated. In addition, the observed genome length G_o was calculated by the formula of NELSON et al. 1994 , which takes into account all markers, linked and unlinked: G_o = G_t + X_o(L – R), where X_o is the observed maximum distance between two framework markers; L is the total number of linkage groups, triplets, doublets, and unlinked markers; and R is the haploid number of chromosomes.

The expected genome map coverage C_e was calculated from the equation C_e = 1 - e^-XeN/1.25Ge (LANGE and BOEHNKE 1982 ), adjusted for chromosome ends. In this equation, N is the number of framework markers; X_e is the maximum distance between two adjacent framework markers in centimorgans at a certain minimum LOD score; and G_e is the estimated genome length. Only framework markers were considered, because these equations refer to randomly distributed markers. The observed map coverage C_o is defined as the ratio of the observed genome length G_o to the estimated genome length G_e.

	RESULTS

TOP ABSTRACT MATERIALS AND METHODS RESULTS DISCUSSION LITERATURE CITED

Genetic markers:
From 187 CAPS primer pairs, 210 probes derived from three cDNA libraries, 42 microsatellite markers, five SNPs in three genes, 26 RAPD primers, one isozyme stain, and one morphological trait, a total of 505 genetic markers that segregated among the progeny of the segregating generation in the two independent pedigrees of C. japonica were identified. Of these 505 markers, a total of 444 markers (176 CAPSs, 197 RFLPs, 42 microsatellites, five SNPs, 22 RAPDs, one isozyme, and one morphological trait) were identified as unique markers. The remaining 61 markers yielded 2–4 loci per marker and were restricted to the categories of gene-based markers (CAPS and RFLP) and RAPD. On average, between the two pedigrees, each marker type yielded 1.15 (YI) and 1.08 (KO) unique segregation loci, except for RAPD. The maximum numbers of scorable segregation loci per marker type were 4 (CAPS), 3 (RFLP), and 3 (RAPD). Although some previous studies in conifers reported that allelic associations among RFLP fragments could not be identified for some loci because of too many bands per single gel image (DEVEY et al. 1996 ; JERMSTAD et al. 1998 ; SEWELL et al. 1999 ), our RFLP image allowed allelic association among RFLP fragments to be deduced, owing to fewer bands per single image.

Segregation distortion:
A chi-square test was performed to test the null hypotheses of segregation ratios of 1:1, 1:2:1, and 3:1 for markers in the YI pedigree and of 1:2:1 and 3:1 for markers in the KO pedigree. The segregation ratios of 58 (17.8%) and 61 (25.1%) markers were significantly distorted (P 0.05) from the expected Mendelian ratios in the YI and KO pedigrees, respectively. For 37 (11.3%) and 32 (13.2%) markers in the YI and KO pedigrees, respectively, the differences from the expected Mendelian ratios were even more significant (P 0.01; Table 2). When we ignored the results for SNP, morphological trait, and isozyme markers (owing to their small numbers), the CAPS markers had the highest percentages of distorted segregation in both pedigrees (19.9% in YI, 31.3% in KO). The percentage of RFLP markers with distorted segregation ratios was slightly lower than that of the CAPS markers. However, microsatellite markers, on the basis of noncoding regions of the genome, represented only a small percentage of markers with distorted segregation, compared with the CAPS and RFLP markers (7.1% of microsatellite markers in YI). The RAPD markers, on the basis of bands randomly extracted from the entire genome, indicated that 19.4% of markers were distorted in the KO pedigree.

These markers with highly distorted segregation ratios at the 0.1% level were excluded from linkage analysis. However, we included the KO markers showing distorted segregation ratios in the linkage analysis, because MUKAI et al. 1995 and IWATA et al. 2001 used them in linkage analyses and obtained plausible maps for the KO pedigree. They observed that most of these markers were clustered on the linkage maps, which led them to speculate that the main cause of the segregation-ratio distortions was linkage with deleterious or lethal alleles (MUKAI et al. 1995 ; IWATA et al. 2001 ). If we exclude these markers with distorted segregation ratios, it would be difficult to obtain long enough linkage groups, because the cluster of ignored markers would hamper making connections between linkage groups on both sides of the cluster. We therefore included these markers with distorted segregation ratios in the linkage analysis of the KO pedigree.

First round of linkage analysis:
Linkage analysis in the YI pedigree was based on 130 CAPS markers, 125 RFLP markers, 38 microsatellite markers, and 5 SNP markers. When we did a first-round analysis with MAPMAKER software, we split the data set into separate subsets of data for constructing linkage maps corresponding to parental meiosis. Seventy-seven CAPS markers, 86 RFLP markers, 21 microsatellites, and 4 SNP markers segregated in the gametes of the YI96 parent. A scaffold map was obtained at a LOD of 3.8 and a distance-linkage criterion, , value of 0.3. Twelve major linkage groups and 1 unlinked marker were found. During marker ordering, 141 markers were placed in the linkage groups, but 46 other markers could not be placed. The observed and estimated map lengths were estimated to be 1650.9 and 2168.5 cM, respectively, at a LOD score of 3.8, and 95.9% of the genome was estimated to be covered by the linkage map of YI96 (Table 3).

In the YI38 parent's meiosis, 83 CAPS markers, 26 microsatellite markers, 64 RFLP markers, and 4 SNP markers segregated in the gametes. A scaffold map was also obtained at a LOD of 3.8 and a of 0.3. Sixteen major linkage groups and 4 unlinked markers were obtained in the first-round analysis. During marker ordering, 137 markers were placed in linkage groups, but 32 other markers could not be placed. The observed and expected map length estimates were 1584.8 and 1810.1 cM, respectively. The expected map coverage estimate indicated that 96.1% of the genome was covered by the linkage map based on YI38 meiosis (Table 3).

We obtained one more linkage map, based on F₁ hybrid meiosis in the KO pedigree. Twenty-eight additional CAPS markers were added to the data set for linkage analysis. A scaffold map was obtained at a LOD of 4.0 and a of 0.3. Ninety-seven CAPS markers, 123 RFLP markers, 31 RAPD markers, one isozyme, and one morphological trait segregated in the gametes of these F₁ hybrids. Twelve major linkage groups were recognized; no unlinked markers were observed. Upon ordering of these markers, locations of 193 markers were determined on the KO linkage map; 60 other markers, however, could not be placed. The observed and expected map length estimates were 1165.0 and 1395.5 cM, respectively. The estimated map coverage rate of the KO linkage map was 96.5% of the genome (Table 3). Clustering of markers resulted in overestimation of the genome size. Therefore, we first evaluated whether the genetic markers were randomly distributed or not; all linkage groups were divided into 5-, 10-, and 20-cM intervals, respectively, following the method of CERVERA et al. 2001 . We detected statistically significant clustering of markers in the YI96 map calculated by MAPMAKER at all intervals. The genetic markers on the YI38 map calculated by JoinMap also were significantly clustered only at 20-cM intervals. It is possible that the genome length estimates of these maps were overestimated.

Second round of linkage analysis:
A total of 146 CAPS markers, 133 RFLP markers, 42 microsatellites, and 5 SNP markers segregated in the gametes of the YI96 and YI38 parents, and these were used for synthetic map construction with the CP of JoinMap 3.0 (VAN OOIJEN and VOORRIPS 2001 ). In the second-round analysis, the CP-type data set was used to calculate recombination rates between markers belonging to five configuration types (Fig 2; MALIEPAARD et al. 1997 ; VAN OOIJEN and VOORRIPS 2001 ): The segregation pattern of 232 markers was configuration type a, 6 was type b, 59 was type c, 23 was type d, and 6 was type e. Twenty-five loci were removed from the data set because of high (P < 0.1) distortion in their segregation ratios. The YI map constructed with JoinMap included 301 genetic markers, making a total of 291 loci that were found to be linked, with a LOD of 3.8. The 291 markers were assigned to 12 linkage groups and covered 1294.4 cM. On average, the linkage map of the YI pedigree presented one marker every 4.3 cM (Fig 3).

View larger version (286K): [in this window] [in a new window] [Download PPT slide]   Figure 3. Linkage maps for C. japonica. The linkage groups on the left were derived from segregation data for the YI pedigree and, on the right, from segregation data for the KO pedigree. The linkage groups in the center were derived from integration analysis of both sets of segregation data with JoinMap 3.0. Markers that are orthologous between the two pedigrees are indicated by allelic bridges. Markers not suitable for integration are indicated by dotted bridges. Loci showing distorted segregation ratios are marked with one (0.01 < P < 0.05), two (0.001 < P < 0.01), or three (P < 0.001) asterisks. The first one to three letters of the locus names indicate the origin of the genetic markers: CC, cDNA library derived from cambium; CD, cDNA library derived from seedlings; CP, cDNA library derived from pollen grains; GD, random genomic library; CS, CJG, and CJS, the three microsatellite-enriched genomic libraries; single letter, RAPD markers derived with the Operon 10mer kit; LAP, leucine aminopeptidase isozyme marker; and MT, morphological trait. Marker types are indicated by the last letter in the locus names: C, CAPS; M, microsatellite; R, RFLP; and S, SNP. Numbers at the end of locus names mean that locus duplication has occurred for that marker.

For the KO pedigree, second-round linkage analysis was also done with JoinMap 3.0 using the F₂ population type code. For linkage analysis, 243 markers were used and 237 markers were found to be linked with a LOD of 4.0 and were assigned to 14 linkage groups, which covered 817.2 cM. On average, the linkage map of the KO pedigree presented 1 marker every 3.0 cM (Fig 3).

Construction of the consensus map:
A total of 180 CAPS markers, 213 RFLP markers, 38 microsatellites, 5 SNP markers, 33 RAPD markers, one isozyme, and one morphological trait were used to construct the consensus map. The segregation data from the two independent pedigrees contained 70 orthologous markers. We observed good correlation of the two-point distances between orthologous markers in the KO and YI pedigrees (Fig 4). We used the "Combine Groups for Map Integration" command of JoinMap 3.0. After the multiple linkages containing the same orthologous markers were associated, a consensus map was constructed. We observed 6 markers in which each probe or primer set derived from a single cDNA source belonged to unrelated linkage groups in the consensus map. In these cases, we refer to these markers as putative paralogous markers and omitted them from the list of orthologous markers. The consensus map produced from 65 orthologous markers included 172 CAPS markers, 200 RFLP markers, 37 microsatellites, 5 SNP markers, 22 RAPD markers, one isozyme, and one morphological trait. A total of 438 markers from the KO pedigree spanning 1372.2 cM were assigned to 11 large linkage groups, 1 small linkage group, and 1 unintegrated linkage group. On average, the consensus map presented 1 marker every 3.0 cM (Fig 3).

View larger version (28K): [in this window] [in a new window] [Download PPT slide]   Figure 4. Comparison of two-point distances (in cM) between orthologous markers in YI and KO linkage data.

The KO5 linkage group contained four markers that were orthologous with the YI9&KO3 linkage group in the consensus map. When we included three linkage groups together in calculating a consensus linkage map, the marker ordering in the YI9&KO3&KO5 linkage group was, however, largely contradictory to the marker ordering in YI9. Therefore, we stopped adding the segregation data of KO5 to those of YI9 and KO3. Furthermore, we observed 10 contradictions in orthologous marker ordering between the consensus map and the YI map, and between the consensus map and the KO map, as indicated by crossing of lines connecting the YI, KO, and consensus maps (Fig 3).

	DISCUSSION

TOP ABSTRACT MATERIALS AND METHODS RESULTS DISCUSSION LITERATURE CITED

Segregation ratio distortion:
Several reasons for distortion of segregation ratios in plants have been put forth, including such factors as chromosome loss (KASHA and KAO 1970 ), genetic isolation mechanisms (ZAMIR and TADMOR 1986 ), and the presence of viability genes (e.g., HENDRICK and MUONA 1990 ; BEAVIS and GRANT 1991 ; LIEDL and ANDERSON 1993 ; BRADSHAW and STETTLER 1994 ). Nonbiological factors such as scoring errors (DEVEY et al. 1994 ; XU et al. 1997 ; NIKAIDO et al. 1999 ) and sampling errors (PLOMION et al. 1995 ; ECHT and NELSON 1997 ) can also lead to distortion in segregation ratios. When the KO pedigree was used to construct the first genetic map for C. japonica, the segregation ratios of 35 loci were distorted (<5% level of significance) among the 164 segregating loci on the linkage map (MUKAI et al. 1995 ). When the CAPS markers were added to the KO linkage map, IWATA et al. 2001 found that 15 out of 60 CAPS markers showed a significant deviation (5% level of significance) from the expected segregation ratio; 11 CAPS markers were distorted from the expected segregation ratio out of 26 additional CAPS markers newly added to the KO linkage map in this study. In total, 25.1% of all markers used for segregation analysis in the KO pedigree showed distorted segregation ratios (Table 2).

MUKAI et al. 1995 speculated that distorted segregation ratios of these markers were caused by putative "embryonic lethal gene(s)" (or viability genes), because clustering of genetic markers showing distorted segregation ratios is consistent with the idea that they may be closely linked to a viability gene. If so, the probability of embryonic lethal genes becoming homozygous would be increased, because the third generation in the KO pedigree was formed by self-pollination of F₁ hybrid individuals. On the other hand, we found that 17.8% of genetic markers of the YI pedigree had segregation ratios that deviated from the expected ratios (i.e., those corresponding to the configuration types in the YI pedigree) and were smaller than those in the KO pedigree. We found six conspicuous clusters of markers having distorted segregation ratios in the KO1, KO2, KO6, KO8, KO10, and KO13 linkage groups in the KO linkage map (Fig 3). CD1712R was the only genetic marker showing a distorted segregation ratio in the YI linkage map within regions equivalent to those containing the clusters on the KO linkage map. The sib-cross in the second generation of the YI pedigree could decrease the number of genetic markers whose segregation was skewed from the expected ratios. TSUMURA et al. 1989 found that 25% of isozyme loci had distorted segregation ratios when segregation analyses were conducted on progeny of self-pollination, but that no isozyme loci had distorted segregation ratios when segregation analyses were conducted on progeny of sib-crosses. KUANG et al. 1999 also detected a high proportion of markers with segregation distortion (34% at 5% significance level) when they analyzed megagametophytes of selfed seeds, except for seeds that died within 1 month of germination. These were derived from a single radiata pine tree. On the other hand, it was reported that only 15% of markers showed segregation distortion (10% significance level) on a genetic linkage map of willow from a full-sib cross of Salix viminalis (HANLEY et al. 2002 ). These results supported that sib-crosses could reduce the number of markers with segregation distortion relative to selfed progeny.

The difference in segregation distortion ratios between the two pedigrees should affect map length. REMINGTON and O'MALLEY 2000 analyzed effects of lethal or semilethal loci due to inbreeding on their genetic maps. However, their model was restricted to the segregation progeny derived from selfed progeny of a single tree. In future analysis, we will extend their model to segregation progeny derived from sib-crosses, like the YI pedigree, and compare the effects of lethal or semilethal loci in the KO and YI pedigrees.

Genome length and coverage:
Previous studies have used various computer programs for generating genetic maps of forest trees (e.g., BARRENECHE et al. 1998 ; DEVEY et al. 1999 ; SEWELL et al. 1999 ; LESPINASSE et al. 2000 ). In general, maps constructed with JoinMap are shorter than those constructed with a multilocus-likelihood package such as MAPMAKER or OUTMAP (SEWELL et al. 1999 ; BUTCHER et al. 2002 ; GOSSELIN et al. 2002 ). Our results also showed that all three maps (YI96, YI38, and KO) constructed with JoinMap were shorter than those constructed with MAPMAKER (Table 3). The multilocus-likelihood method used by MAPMAKER assumes an absence of crossover interference; so when interference is present, JoinMap correctly produces shorter maps, even though both programs use the Kosambi mapping function (STAM 1993 ). This difference was also observed in barley and was attributed to how each program calculates map distance when the actual interference differs from that assumed (QI et al. 1996 ).

In C. japonica, four studies concerned with the construction of genetic maps have been reported (MUKAI et al. 1995 ; KURAMOTO et al. 2000 ; NIKAIDO et al. 2000 ; IWATA et al. 2001 ). Using F₁ progeny of unrelated parents, KURAMOTO et al. 2000 constructed two linkage maps by the two-way pseudo-testcross strategy and estimated the genome length and map coverage statistics by the methods of HULBERT et al. 1988 . The estimated expected genome length of Iwao-sugi was 2868.0 cM at a LOD score of 4.0, and that of Boka-sugi was 2790.7 cM at a LOD score of 4.0 (KURAMOTO et al. 2000 ). On the other hand, our estimates of genome length were between 2168.5 (YI96) and 1395.5 cM (KO) according to MAPMAKER and between 1632.4 (YI38) and 1121.8 cM (KO) according to JoinMap. The reported genome lengths above were outside the confidence intervals (95% criteria) of our data. However, KURAMOTO et al. 2000 used only RAPD markers to construct genetic maps, in contrast with our linkage map, which was composed mostly of EST markers. RAPD markers could be dispersed throughout the genome more randomly than EST markers. If a majority of EST markers used in our study are clustered, our genome length estimate would be underestimated.

Our expected map coverage estimates ranged from 95.9% (YI96) to 96.5% (KO) according to MAPMAKER. Although the linkage maps for Iwao-sugi based on RAPD markers covered only 62% of the genome (KURAMOTO et al. 2000 ), the total length of the linkage map for Iwao-sugi (1756.4 cM) was longer than those we found (1567.3 cM for YI96 and 1138.6 cM for KO, MAPMAKER analysis). Although many factors can affect marker coverage and genome map density, such as genome length, number of markers, distribution of marker polymorphism, distribution of markers on the genome, crossover distribution on the genome, mapping population size and type, and mapping strategy (LIU 1998 ), a nonrandom distribution of EST markers might be the main cause of the discrepancy. Our estimates suggest that our linkage maps covered the entire genome of C. japonica. However, our linkage map probably does not cover nondense regions of genes in the genome. Extensive microsatellite markers or random genetic markers, such as AFLP and RAPD, would be helpful tools for filling in the nondense regions of genes in the genome.

Construction of the consensus map:
The 11 chromosomes of C. japonica are represented in the YI map generated with the CP mode of JoinMap and also in the consensus map. The smallest linkage group (YI12&KO8) in the consensus map could belong to any of the 11 linkage groups, but additional genetic markers are needed to make this assignment. It was impossible to find a clear correspondence of the KO15 linkage group with the other consensus map because of a lack of orthologous markers. However, the KO15 linkage group is part of the LG1 linkage group (IWATA et al. 2001 ). The LG1 linkage group was divided into two parts between the CD133 and CD344 markers at the second-round analysis. The interval between these markers was 21.8 cM in the linkage map of IWATA et al. 2001 . One part of LG1 corresponded to KO13; the other part corresponded to KO15. Therefore, KO15 would be placed downstream of KO13 in the YI3&KO13 consensus linkage group if orthologous markers existed (Fig 3).

Southern blot analyses using cDNA and gene probes have revealed genes that are found in double, and occasionally multiple, copies in many plant species (e.g., BERNATZKY and TANKSLEY 1986 ; HELENTJARIS et al. 1988 ), including forest tree species (KINLAW and NEALE 1997 ; DEVEY et al. 1999 ; SEWELL et al. 1999 ). We found 12 tightly linked clusters of EST markers that came from a single cDNA clone, out of 17 EST markers revealing multiple loci derived from single cDNA clones within one linkage group. Five EST markers from a single cDNA clone were dispersed throughout the genome. Thus, our data demonstrate that >50% of EST markers derived from multigene families were tightly linked or located on the same chromosome.

Some changes in marker order (other than those due to translocation) were observed during construction of consensus maps (SEWELL et al. 1999 ; LESPINASSE et al. 2000 ; SEBASTIAN et al. 2000 ; CERVERA et al. 2001 ; JEUKEN et al. 2001 ; LOMBARD and DELOURME 2001 ). Small discrepancies in marker ordering may be due to mapping imprecision rather than to real rearrangements (LOMBARD and DELOURME 2001 ). We observed four large and six small differences in marker order between the YI or KO map and the consensus map. One of the reasons for the discrepancies might be due to chance, because LOD score criteria decided arbitrarily were not stringent. Distorted segregation ratios were observed at three loci of the KO pedigree at which we detected large discrepancies in marker ordering. The KO pedigree showed many loci with distorted segregation ratios, which might affect marker order in the linkage map.

One of the main goals of constructing consensus maps is to compare QTL between different genetic backgrounds, especially in allogamous species. We can determine how many and where QTL exist in such species by using multiple pedigrees with different genetic backgrounds. In C. japonica, QTL relating to juvenile growth, flower bearing, and rooting ability of cuttings have been identified in the KO pedigree (YOSHIMARU et al. 1998 ), and QTL for the modulus of elasticity of wood have been identified in a different pedigree (KURAMOTO et al. 2000 ). We will be able to summarize QTL from different pedigrees on the consensus map and obtain knowledge of how many QTL exist and which QTL are expressed throughout the pedigrees in future analysis.

Our markers and genetic maps should be valuable for researchers studying related species, such as the Taxodiaceae and Cupressaceae, because TSUMURA et al. 1997 showed that a high proportion of the EST markers could be useful in other species of the Taxodiaceae and Cupressaceae. Therefore, these EST markers will allow studies of genome evolution and comparative mapping between species within the Taxodiaceae and Cupressaceae. The consensus map developed in this study will become the basis of genome studies of the Taxodiaceae and Cupressaceae. Our data are available on our web site (http://www.ffpri.affrc.go.jp/labs/cjgenome/database/cjdatae.html).

	FOOTNOTES

¹ Present address: National Agricultural Research Center, Tsukuba, Ibaraki 305-8666, Japan.

	ACKNOWLEDGMENTS

The authors thank K. Mikuni, M. Koshiba, Y. Kawamata, Y. Taguchi, K. Aoyagi, and J. Kobayashi for excellent assistance and H. Tachida and T. Kado for supplying segregation data for SNP markers. We also thank S. Kanetani, S. Ueno, Y. Moriguchi, T. Sugaya, I. Karube, and T. Ochiai for their assistance with progeny maintenance in the nursery and also thank Q. Han, S. Katahata, S. Yamada, and T. Yokota for helpful advice and support throughout this investigation. We appreciate K. Ohba for establishing foundation of the sugi genome project and supplying the KO pedigree and two anonymous reviewers for their helpful comments in the latest version of the manuscript. This study was supported by grants from the Program for Promotion of Basic Research Activities for Innovative Biosciences (PROBRAIN) and the Pioneer Special Study of the Ministry of Agriculture, Forestry and Fisheries in Japan.

Manuscript received February 17, 2003; Accepted for publication July 28, 2003.

APPENDIX A Description of additional CAPS markers in sugi Restriction enzyme Locus Forward primer: 5' to 3' Reverse primer: 5' to 3' Anneal temp. No. of PCR cycles Putative size (bp)a YI KO CC2052C TGTTGCCGGTAGGGTTTCTA TTACCGTATTGCTTGCCATTG 55 36 >2000 HinfI CC2081C GCATGGCAGAAGCAGAAG TTCACATATGCGATGACACAA 60 36 1200 StyI CC2123C CGGCGCTTACCTCATCGTT CCCTGCTACCGACGGACTCTA 60 36 2000 NciI CC2188C AGCTGTGCGATCAAGTTTCTG ATGGGCGTGCCTCCTAA 60 40 1400 BstOI CC2286C ATAATGCCACCTCCAGGAC AGGCCAGTTTAACAAATGTCA 60 36 1800 AluI CC2333C GGTGGACCTTCGTTCTG AACCCAACTGCACTACTCTT 60 40 1500 TaqI CC2340C TACAGGAGGCGGAGGAC CTCAAACTGCCAAACAACAA 60 35 1000 AluI CC2377C GAAGGAGCTGAAGGAGG CTAAGCGTTGAAACTGAGAA 55 35 1500 HaeIII CC2419C CAATGAGGAGGTCTGTATG AAATTTGGAGGATCTCAAC 60 40 1500 NdeII CC2435C GCAGGCAGTTCAGAGTTTT TCCCGAAGAGAGTTTTATGG 60 40 900 HaeIII, RsaI CC2448C ATCCTAAGTCCCCAGAAAGT GAATTGGAATGGCATAAAGA 60 40 2000 RsaI, HhaI CC2467C CGGAGGAGGCGGCTGAGAGT CGACCCTGAAGATTGTTTGA 60 40 800 EcoO109I EcoO109I CC2469C TCGACTTCGGTAGCAGCACA TCATCCGCCTCGTCCTCCTC 62 35 600 AluI CC2522C CGACGAAGAGGATGATGAAC GCCAGCTGTGATATGATTGT 60 40 2000 HinfI, NdeII CC2541C CGCAAGAGAGCTCGTCGTCA CAAACTTGGAGGATGTGTCA 60 35 2000 DdeI, HinfI CC2577C AGGTCTGTAAGGTGTGAGGG ATAGAAAGGCAACAGTAGCA 60 40 1100 DdeI DdeI CC2583C AATTATGGGAGAGAACTGGA ATTAAACCGTACATGGAACT 60 40 1500 DdeI DdeI CC2588C CTGCCGCTGCCGTTTATTCC TTATCCACGACGTACACACC 60 40 900 SspI CC2621C GTTGCTGTGGGAGGACTTTG AGCCCACCTAATAGATGAGA 52 36 700 HaeIII HaeIII CC2631C GCATTTGCTCCCATTAGTTC TTTCTTCCTCGCCATTCTTC 60 36 1300 BstOI CC2643C CACGGTGGCATTGACATCTT ACCTACGCTACAACCCTCCC 62 36 >2000 MspI CC2645C TGTCGGTGTGTTGCCTCTTC GTGGGCTTCTGCATAATCAT 62 36 1100 BglII CC2657C ACCTGCCCTCCTTTCCATTC CAACTGTTACACCGCCCTCC 60 36 2000 ScrFI CC2674C CCGACTCACCCTTTCTTCAC TGCCATATCTCAACAATCTC 52 36 1000 AluI CC2676C CAAGGGTTTGGGAAAGGGAG CCGATTGAGGAGACTGCTAA 60 36 500 BstOI CC2683C TGCGAAATGTTAGCCCTCTG CCCTCTGTATCATCCCTGTC 60 36 500 HaeIII CC2700C ATTTGTGCAGGTTATTTGTC TATTCGGTGGAGGAGGTGGT 60 36 700 ScrFI CC2702C TTCGCCAAGCCACCATAGAC CTGCCACCACAACACCCTCC 60 36 500 RsaI CC2713C ATCATAGCTGCGAAGAACAC GTCCCGTCATTGCCACACCA 60 36 350 MspI MspI CC2716C GTTGACATGATCCGAAAGAG CAAACGCAAATACTGAAAGG 60 36 1000 AluI CC2731C CAAGCCCAAGCCCAGGTCGT TGCAGGGATAGGATAGGTAG 62 36 >2000 TaqI CC2746C TAGAAATTGCTCATGTGGGT CCTCTTCTTTCCGCTGCTGT 60 36 2000 DdeI, RsaI DdeI CC2750C GGCAGCACACAGACAACACA GATACTTCTCAGGCCCAACT 62 36 1700 SinI CC2752C CCGCACTGCCATCTACGACT AACCTCTCCTCCAACTCACC 62 36 1000 HaeIII ALP (900, 1100) CC2781C CAGAGAAACCCAGCGAGGAA GCAACAATGGCATACAAACT 60 36 1200 DraI CC2795C ATCCAGGAGCAAAGAAAGGT ATAGCAGCAGAATGGTCAGG 60 36 800 DdeI DdeI CC2831C GGCGATGGCAGCAAACGAAG CACGCACCACTCCACCCTAC 62 36 500 DraI, HhaI CC2846C AAGTAAGTTGGTCGGTAGGT AAGAAGGCATTTTGGTGAGG 60 36 1400 DdeI MspI CC2856C GACGAAGGCTGAAAAAGGTG GCATCTAGGCATACGCTGAA 62 36 2000 MboI CC2860C CTAAAGGGAAACAAATCAGG TACTCGTCTTCTAACCGTCA 60 36 1100 DraI, HincII CC2895C TCATGGCATTGCGGAGAGGG CGGCCTGTAAGACCACCTGA 60 36 1200 TaqI CC2909C GCAGCAATCTTTCCTCCTCC GCATGCATTTAGCCTTCACC 62 36 >2000 AluI CC2918C TTGGCTTCTATGGACCTATG ACTGGACTTTTGCGATGCTT 60 36 >2000 AluI, MboI NdeII CC2921C TTTTGGCGGTGGGAGGAATG CAAGAATCGGTGAAGAACAG 60 36 1400 RsaI, TaqI TaqI CC2939C CTCGCTGAGCAAGACTAGGG CATGACGAAAATGCCCTGTA 60 40 1000 NdeII, TaqI TaqI CC2946C GTATCCAGGGATGCTCGAAA AAATTGCCATCCTTCCTCCT 55 40 1200 TaqI CC2989C ATTTGGAACTTCGGAAGCCT TTGATGCATATCCTGTCCCA 55 40 700 HaeIII CC3098C ACAATGCCTTCCCATGAAGT ATCAGGCTGTTGGGAATCAG 60 40 1100 AvaI, TaqI TaqI CC3133C AAGGTTCATCGCCCTATGTG AGCTCCAACCTCAAAGACCA 60 40 900 TaqI CC3145C TCCACTTAGCGTCAATTCCC CACACTTCCATGTTAGGGGC 60 40 2000 AluI, HinfI CC3336C TGGTCATGATGTGCTTGGTT AGTTGCTACAATGTTCCCGC 60 40 >2000 TaqI CC3367C AGAGATGGCGCTCACTCATT TACTACACCACCGCTTGCAG 60 40 900 AluI AluI CC3393C TCTCCTGAATGGGATGAAGC CACATGCTTGCCGAAATAAA 60 40 700 HinfI CC3413C GGAAAACAGTGTGAGGGTGC TGGCATGGTCTCGTTTGTTA 60 40 1300 BglI, BglII CC3416C CCCTCAACTCCTCCAATGAA CATGACCTGTCGTGCTTGAT 60 40 1400 HaeIII HaeIII CC3430C GACGAGGGACGACCTGTTTA ACTCAACACCAGCATCTCCC 60 40 2000 HindIII CC3816C AGTCAGAGCTGCCTGGAAAG GCCACGAAGGGATTCATTTA 60 40 2000 MspI RsaI CC3823C CCCCACAGGACATCAAAACT ACGCATTCTCCATCACTTCC 55 40 900 HinfI CC3839C CTGCATTTCCTCTGGAATCG TTGGGATAAACCTTTTTGCG 60 40 2000 TaqI CC3872C AGCGGAAGTACCCTTTGGAT GGTTCCCAGTGATTTCCTGA 60 40 1600 TaqI a Putative PCR fragment sizes were deduced by agarose gel electrophoresis (ethidium bromide staining).

APPENDIX B Description of microsatellite markers in sugi Locus Forward primer: 3' to 3' Reverse primer: 5' to 3' Anneal temp. PCR cycle Motif Putative size (bp)a CJS0002M CTTTTTTCAAATTTAGTGATGT CCCATGCCCCACTGTCCACC 55 30 (TC)12(TC)17 237 CJS0091M GAGAGATAAGAGGGTAGAGGT CAATGCCAACTTAGAAGAC 60 30 (GA)43 298 CJS0268M CCTTAGAAAGCTATGCCAC GCAACGCATCCATAATACC 60 30 (AC)53 352 CJS0331M GGAGAGATAGACGACAAAAGAG CCATCTTGCTAATCTGTCC 60 30 (GA)6 245 CJS0333M AGGAGATTAGGATGGTGGG GGTTTGCCTCTTCTATGAG 60 30 (GA)26 264 CJS0356M CTAAAGAATAGATGACTCCAC TATAACGCTTTTGCCCTCA 60 30 (GA)64 337 CJS0401M GATCTAAACTTGAGCATAAC CAATCCTGTCTCCATACCC 55 30 (CG)8(GA)54 222 CJS0455M GTTACTTTGAAAAATGAGCC AACATCAAGATTAAAGGGAC 58 30 (CT)20 166 CJS0485M CATATCTAATATCTAATACCTTG TCTCCCTATCTAGCCCTCTG 50 35 (GA)9(GA)30(GA)27 331 CJS0520M TCCCTTTTGGTATTTTACAC ACTCAAATTGCGATAATCTC 55 30 (TG)18 196 CJS0584M TGGTTTGCCTTTGGTTGCTC GGACTTTCTATTTACCTCTTGG 60 30 (AG)80 329 CJS0665M CCAAGCATAGGGAAAAAGAG GGGGAGTAAGGATGACATTT 60 30 (GA)45(GA)29 367 CJS0686M CAATGCAAATATAAGTTCACCC TCCACCTCTTTTTCATTCTC 55 30 (GA)52 275 CJS0838M TATGTAGAAGCGTGTGATGT GATAATTGCCTTTGTTGTCC 58 30 (GT)23 170 CJS0955M CACACTCCCCGTCTCCGACAG ACCCTGATTCCCCATACACC 58 30 (TCT)4(GA)29 137 CS1226M CTCTAGTCCTCAATGGTGGT TATTAAGCATTTTCCCTCTC 60 35 (CA)14 139 CS1281M CCCCCTCTCATTAGTTACCA CAAAAATCAACAAGCCAACC 60 30 (CT)15 233 CS1413M GGAAAGGATGTTATGGGTGT CGGTTGATTTTGTCGGCACT 60 35 (TG)11(GT)15 285 CS1522M AAAGTTTGATTAGGGCAGGG AAACGTGGGTGCTATCCTTC 62 30 (AC)16 222 CS1737M TACCCTCAACCCTTCACCCT TTACCCACCTCTCTTTCCTC 60 30 (AG)40 248 CS1895M TGAGAGAGGGAGGGAGGGTT GAGTCCTTGTCCCGTTTTGT 60 30 (TG)10 405 CS2024M AGTAATACAAGATAAGGGAG TCCACCTCTATACCTCTACA 55 30 (AG)15(AG)4(AG)10 314 CS2056M GAGAGACATGGGGGAAGAGG GGTTCTAACACATGAATGGC 60 30 (GA)20(GA)7 295 CS2169M GTAGAGGAGGGATATAGAGT TCCTTGTCCATCTCTCTTTA 55 30 (GA)9 141 CS2484M TGAGAAAGGGAGAGAGGGAT CCCCCTTCTCTTTTTCACTC 60 30 (GA)13 158 a Putative PCR fragment sizes were deduced from sequences of genomic clones between forward to reverse primers.

	LITERATURE CITED

TOP ABSTRACT MATERIALS AND METHODS RESULTS DISCUSSION LITERATURE CITED

BARRENECHE, T., C. BODENES, C. LEXER, J.-F. TRONTIN, and S. FLUCH et al., 1998  A genetic linkage map of Quercus robur L. (pedunculate oak) based on RAPD, SCAR, microsatellite, minisatellite, isozyme and 5S rDNA markers. Theor. Appl. Genet. 97:1091-1103.

BEAVIS, W. D. and D. GRANT, 1991  A linkage map based on information from four F₂ populations of maize (Zea mays L.). Theor. Appl. Genet. 82:636-644.

BERNATZKY, R. and S. D. TANKSLEY, 1986  Majority of random cDNA clones correspond to single loci in the tomato genome. Mol. Gen. Genet.