| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
a USDA-ARS, Forage and Range Research Lab., Utah State Univ., Logan, UT 84322-6300
b Univ. of Missouri-Columbia, Columbia, MO 65211
c U.S. Army Corps of Engineers, Engineering Research and Development Center-Cold Regions Research and Engineering Lab., Hanover, NH 03755-1290
* Corresponding author (stlarson{at}cc.usu.edu)
 |
ABSTRACT |
|---|
 TOP ABSTRACT INTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONREFERENCES |
|---|
Abbreviations: AFLP, amplified fragment length polymorphism • ANTH, anthesis date • BOLT, reproductive bolting • CIRC, plant circumference • HGHT, plant height • IM, interval mapping • MQM, multiple-QTL model • QTL, quantitative trail locus (i) • RFLP, restriction fragment length polymorphism • rMQM, restricted MQM mapping • STS, sequence-tagged site • SSR, simple-sequence repeat
|
INTRODUCTION |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONREFERENCES |
|---|
Growth habit is a highly variable and ecologically important trait in perennial grasses. Aggressive rhizomes characterize quackgrass [Elymus repens (L.) Desv. ex Nevski] and johnsongrass [Sorghum bicolor (L.) Moench], which rank among the world's worst perennial grass weeds (Holm et al., 1977). In general, caespitose (i.e., growing in bunches or tufts) and rhizomatous grasses dominate semiarid and mesic grasslands, respectively (Sims et al., 1978). Nutrient islands beneath caespitose grasses may also contribute to clone fitness in this growth form in both mesic and semiarid communities, whereas the distribution of rhizomatous grasses may be restricted to microsites characterized by higher soil organic carbon and nitrogen concentrations (Derner and Briske, 2001). We have observed L. cinereus and L. triticoides growing in close proximity in mixed stands, at several disturbed sites, with no apparent difference in microhabitat. Conversely, we have observed L. triticoides in riparian or wet zones and L. cinereus inhabiting dry adjacent uplands, restricted to seemingly different natural microhabitats. In any case, L. cinereus and L. triticoides display profound differences in growth habit. Lateral branches of L. cinereus grow strictly upward, often within the lower leaf sheaths as tillers, whereas the lateral branches of L. triticoides frequently grow horizontally or downward under the soil surface as rhizomes. In addition to obvious morphological differences in growth habit and stature, L. cinereus and L. triticoides display differences in salt tolerance, seed dormancy, seed color, seed shattering, mineral content, tillering, texture, and other characteristics.
The F1 hybrids of L. cinereus and L. triticoides are robust and vigorous plants with relatively large biomass potential compared to other range grasses of western North America. Wu et al. (2003) recently constructed high-density molecular genetic linkage maps for two full-sib families, TTC1 and TTC2. The 164-sib TTC1 and 170-sib TTC2 families were derived from two different L. triticoides x L. cinereus F1 hybrids, TC1 and TC2, crossed to different clones of the same L. triticoides tester genotype (T–tester). The TC1 and TC2 F1 hybrids were derived from naturally heterogeneous L. triticoides and L. cinereus seed accessions from Oregon and Alberta, respectively. Together, the TTC1 and TTC2 families capture a snippet of the divergence among and heterogeneity within the L. triticoides and L. cinereus source populations. These two experimental populations provide unique system for gene discovery research and development of breeding markers in perennial grasses. A close relative of quackgrass, Leymus also provides a useful system to study weedy traits such as rhizomes. Our primary objective here was to identify, characterize, and compare QTLs controlling growth habit, plant height, and flowering traits. Another major objective was to compare the location of these Leymus QTLs with genomic regions controlling related traits in other cereal and grass species.
|
MATERIALS AND METHODS |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONREFERENCES |
|---|
|
Rhizome proliferation was measured as plant circumference (CIRC) in late spring or early summer. For caespitose plants, this could be simply measured by stretching a tape ruler around the tussock at the soil surface. The circumference of irregular sods, characteristic of the more rhizomatous plants, was approximated as the perimeter of a polygon where the corners represent the outermost rhizome branches (i.e., the shortest distance around the outermost rhizomes) at the soil surface. Anthesis date (ANTH) was measured as the number of days from 1 January until the first day of anther extrusion, which is most apparent on warm dry mornings, beginning mid-June. In practice, the first day of anther extrusion was interpolated between two or three observations per week. A priori, one unexpected phenomenon in the TTC1 and TTC2 families was variation in reproductive bolting (BOLT), where some genotypes flowered and some did not, first observed in 2002. Individual clones were retrospectively rated as 0 (not flowered) or one (flowered), after seed harvest and mowing, based on ANTH notes take in 2002. We subsequently rated individual plants for the proportion of bolting culms from 0 (no spikes produced) to 1 (virtually all major culms bolting), before seed harvests and mowing, in 2003 and 2004. Plant height (HGHT)as measured after anthesis using a 2-m ruler. Plant height (i.e., actual culm length from soil surface to the spike terminal) was not measured on plants that did not flower.
Data Analysis
QTL analyses were based on trait means from two replicants for each of the 164 TTC1 and 170 TTC2 segregates in 2002, 2003, and 2004. QTL analyses were also performed on averages over all 3 yr, using output from the LSMEANS procedure of SAS (Statistical Analysis Systems Institute Inc., Cary, NC). Coefficients of skewness and kurtosis, g1, and g2 were calculated as described by Snedecor and Cochran (1980). Departures from normality were deemed significant if they exceeded two standard errors of skewness (SES) and kurtosis (SEK) estimated by (6/N)–2 and (24/N)–2, respectively (Tabachnick and Fidell, 1996). Histograms of phenotypic values (clonal averages) from 2002, 2003, and 2004 were also prepared for CIRC, BOLT, ANTH, and HGHT (Supplementary Data).
Broad-sense heritabilities within years were obtained using a SAS program for estimating heritability from lines evaluated in an RCB design in a single environment (Holland et al., 2003). Likewise, heritabilities among years were obtained using another SAS program for estimating heritability from lines evaluated in RCB designs in multiple environments (Holland et al., 2003), modified to account for repeated measurements on perennial plants. All class variables (i.e., rep, clone, year) were treated as random effects. Genotypic and phenotypic correlations between traits evaluated using a SAS program for estimating correlations from RCB designs in multiple environments (Holland et al., 2003), also modified to account for repeated measurements. The original SAS programs for estimating heritabilities, genotypic correlations, and phenotypic correlations (Holland et al., 2003) are available at http://www4.ncsu.edu/
jholland/homepage_files/Page571.htm (verified 24 July 2006).
A sequential and reiterative procedure of QTL detection was performed using the MAPQTL five package (Van Ooijen, 2004). Genome-wide interval mapping (IM) (Lander and Botstein, 1989; Van Ooijen, 1992) was performed in 1-cM increments to identify putative QTLs and possible cofactors used in a multiple-QTL model (MQM) (Jansen, 1993; Jansen, 1994; Jansen and Stam, 1994). Markers with the highest log-likelihood ratios (i.e., LOD test statistics) for each QTL (no more than one per chromosome) were selected as the initial set of possible MQM cofactors. A backward elimination procedure was applied to this initial set of cofactors using a conservative significance level of 0.001 to ensure the independence of each cofactor. A reiterative process of restricted MQM (rMQM) mapping, which excludes any syntenous cofactors (i.e., cofactors located on the same linkage group that is being scanned), was used to refine the location of rMQM cofactors (QTLs) and identify new rMQM cofactors (QTLs). Moreover conventional MQM mapping was used to detect (or exclude) syntenous QTLs, where that possibility was apparent. Finally, all putative QTLs were also evaluated using the nonparametric Kruskal–Wallis rank sum test, equivalent to the two-sided Wilcoxon rank sum test for two genotype classes (this study), which means that no assumptions are being made for the probability distributions of the quantitative traits (Lehmann, 1975).
A threshold of 3.3 LOD was used throughout these QTL, MQM, and rMQM detection procedures as a close approximation for a genome-wide 5% significance level as determined from simulation tables based on genome size and family type (Van Ooijen, 1999) and empirical thresholds based on permutation analyses with 1000 replications (Churchill and Doerge, 1994) specific to each trait. Both methods (Van Ooijen, 1999; Churchill and Doerge, 1994) produced remarkably similar results for all four traits evaluated in this study.
Genome-wide IM and rMQM LOD scans based on 2002, 2003, 2004, and average phenotypic values were performed for CIRC, BOLT, ANTH, and HGHT using MapChart version 2.1 (Voorrips, 2002) and the updated Leymus TTC1 and TTC2 genetic maps as described in the results section below. Detailed graphs of each genome-wide LOD scan using the complete high-density linkage maps are provided in the Supplementary Data Online.
|
RESULTS |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONREFERENCES |
|---|
The thioredoxin h RFLP marker, Xbm2, maps near the self-incompatibility gene locus, S, on homeologous group 1R of Secale (Korzun et al., 2001), and thioredoxin h STS markers have been utilized in perennial ryegrass (Taylor et al., 2001) and Kentucky bluegrass (Patterson et al., 2005). The thioredoxin h STS primers used by Patterson et al. (2005) amplified two distinct sets of sequences, designated TRX1 (AY943821 and AY943822 from L. cinereus Acc:636; AY943823 and AY943824 from L. triticoides Acc:641) and TRX2 (AY943825-AY943827 from L. cinereus Acc:636; AY943828-AY943830 from L. triticoides Acc:641). We designed TRX1 and TRX2 primers that would amplify corresponding sequences from L. cinereus but not L. triticoides—thus providing informative polymorphisms for mapping in the TTC1 and TTC2 backcross families. The TRX2 amplicons mapped to homologous positions of the LG6b linkage group, evident by the coliniearity with homologous TTC1 and TTC2 AFLP markers (Fig. 1). The TRX1 primers amplified 289 bp sequences that mapped to LG1a in TTC1 and LG2b in TTC2 (Fig. 1). The TRX1 locus on LG1a is presumably homeologous to the Xbm2 locus on Secale 1R. The role of thioredoxin h genes in seed germination, self-incompatibility, grain quality, and characteristics has been evaluated in cereals, grasses, and other plants (Juttner et al., 2000; Langridge et al., 1999; Li et al., 1994, 1997; Sahrawy et al., 1996).
Of special interest with regard to flowering traits, the VRN1 and VRN2 genes are the two most potent genes controlling differences in vernalization requirement between winter annual and spring annual barley, wheat, and rye (Dubcovsky et al., 1998), and both of these genes have been cloned (Yan et al., 2003, 2004). The VRN2 gene is present on homoeologous regions of 4H in winter barleys (evidently absent in most spring barleys) and a 4A/5A translocation region of Triticum monococcum 5A (Dubcovsky et al., 1998; Yan et al., 2004). Primers designed from the wheat VRN2 gene amplified mixed sequences from Leymus that were very similar in size (supplemental data) and sequence (DQ486013) to the wheat VRN2 gene. These putative Leymus VRN2 sequences map to a locus on the distal long arm of Leymus 5Ns (Fig. 1), which evidently corresponds to the VRN2 locus on Triticum monococcum 5A. Although we have not yet mapped the Leymus VRN1 sequences, this gene is closely associated with the CDO504 marker in other Triticeae species and also maps to Leymus LG5Ns and LG5Xm (Fig. 1) (Wu et al., 2003).
In summary, seven SSR primer pairs and seven STS primer pairs (supplemental data) detected 15 additional anchor loci in the TTC1 family and nine additional anchor loci in the TTC2 family, not previously described by Wu et al. (2003). The updated TTC1 map includes 1069 AFLP markers and 53 anchor loci in 14 linkage groups spanning 2001 cM. The updated TTC2 map contained 1002 AFLP markers and 45 anchor loci in 14 linkage groups spanning 2066 cM. Some 488 homologous AFLP loci and 31 anchor markers have been mapped in both families, showing similar map order. Thus, 1583 AFLP markers and 67 different anchor loci have been mapped into 14 linkage groups, which evidently correspond to the 14 chromosome pairs of allotetraploid Leymus (2n = 4x = 28). The map locations of the 17 new anchor loci (24 total including seven homologous pairs) (Fig. 1) were largely consistent with the other anchor loci previously mapped in Leymus (Wu et al., 2003).
Growth Habit
The L. cinereus Acc:636 and L. triticoides Acc:641 progenitor accessions displayed very different levels of rhizomatous spreading, which became more pronounced each year (Table 1). Although CIRC measurements of the TC1 and TC2 F1 hybrids were significantly greater than L. cinereus Acc:636, these means were well-below the midparent, TTC1, and TTC2 averages especially in 2004 (Table 1).
|
|
|
|
Flowering Traits
The BOLT and ANTH traits showed relatively strong negative genotypic correlations in both TTC1 and TTC2 families (Table 3), evidently controlled by several common (probably pleiotropic) QTLs (Fig. 1). Thus, data from these two traits are presented together in one section.
The L. cinereus Acc:636 and L. triticoides Acc:641 progenitor accessions; F1 interspecific hybrids (TC1 and TC2); and TTC1 and TTC2 families displayed similar BOLT and ANTH phenotypic means (Table 1), except for the fact that L. cinereus showed relatively weak bolting in the first establishment year. Leymus cinereus has a relatively large stature, requiring several years to reach reproductive maturity. Yet, the TTC1 and TTC2 families displayed transgressive BOLT and ANTH segregation that persisted through 2002, 2003, and 2004 (see trait distributions in Supplemental Data). Unlike the F1 hybrids or parental source populations, some transgressive progeny failed to flower in 2002, 2003, 2004, and 2005.
The BOLT and ANTH traits both showed relatively weak plot-mean heritabilities over years (Table 2). The ratio of genotype x year interaction to phenotypic variance was substantially greater for ANTH (0.18) and BOLT (0.21) compared with CIRC (0.02) or HGHT (0.13), which explains the low plot-mean heritabilities over years for the ANTH and BOLT flowering traits. Although, BOLT and ANTH traits show significant genotype x year interaction and the significance of QTLs vary by year (Tables 4), several QTLs were detectable over years (Fig. 1). One notable exception was a highly significant TTC2 ANTH QTL on LG6a in 2002, which was virtually nonexistent in subsequent years. Bolting deficiencies were strongly correlated with late flowering and weakly correlated with short caespitose characteristics (Table 3).
A total of nine BOLT QTLs were detected in the TTC1 and/or TTC2 families (Table 4, Fig. 1). The TTC1 and TTC2 BOLT QTL peaks on the upper portion of LG4Xm overlap well enough (Fig. 1) that we counted these as one homologous QTL. Likewise, we believe that there is insufficient separation of the TTC1 and TTC2 BOLT QTLs on LG6a to declare these as different (Fig. 1). Thus, L. cinereus contributed five negative TTC1 and/or TTC2 BOLT QTL alleles and four positive TTC1 and/or TTC2 BOLT QTL alleles (Table 4, Fig. 1). Two of these nine BOLT QTLs were significant in both TTC1 and TTC2 families (i.e., homologous), whereas seven QTLs were unique to TTC1 or TTC2. Four QTLs explained up to 43.9% of the BOLT variation in the 2003 TTC2 data set, which was the most explanatory QTL model in this study (Table 4).
A total of five significant ANTH QTLs were detected in the TTC1 or TTC2 families (Table 4, Fig. 1). All five significant ANTH QTLs were unique to TTC1 or TTC2 (Table 4 and Fig. 1). However, the TTC1 family displayed ANTH LOD values that were nearly significant on a chromosome region homologous to the TTC2 ANTH QTL on LG4Xm (see ANTH LOD scan in Supplemental Data). Similarly, both TTC1 and TTC2 displayed nearly significant ANTH LOD values on homologous regions of LG3b. In any case, L. cinereus contributed three positive and two negative ANTH QTL alleles at the five significant QTLs (Table 4, Fig. 1).
The genome-wide LOD scans were similar but not identical for the BOLT and ANTH flowering traits (Fig. 1 and Supplemental Data). The TTC1 BOLT QTL on LG3a had no apparent effects on ANTH. The TTC2 BOLT QTLs on LG4Ns were associated with significant or nearly significant ANTH QTLs. Likewise, the BOLT QTLs on LG4Xm were associated with significant or nearly significant ANTH QTLs in the TTC1 and TTC2 families. The TTC1 ANTH QTL on LG 5Ns was associated with elevated BOLT LOD, albeit less than the 3.3 LOD threshold (Supplemental Data). Interestingly, the TTC1 and TTC2 BOLT QTLs on LG6a were associated with significant ANTH QTLs only in the TTC2 family. The TTC1 BOLT QTL on LG6a was not associated with elevated ANTH LOD values. The latter observation raises some doubt about the putative homology of the TTC1 and TTC2 BOLT QTLs on LG6a (Fig. 1). Finally the TTC1 BOLT QTLs on LG6b and LG7b were closely associated with a significant or nearly significant ANTH QTLs, respectively.
Significant skewness and kurtosis was detected for BOLT and ANTH; however, all QTLs (Table 4, Fig. 1) were highly significant (P < 0.0001) on the basis of the nonparametric Kruskal–Wallis rank sum test.
Plant Height
The L. cinereus Acc:636 plants were substantially taller than the L. triticoides Acc:641 plants in 2003 and 2004 (Table 1). However, the L. cinereus plants require several years to reach full-size; thus, interspecific differences in plant height were not apparent in 2002. Interestingly, the TC1 and TC2 F1 hybrid genotypes displayed greater plant height than the L. cinereus Acc:636 or L. triticoides Acc:641 reference individuals, especially in 2002 and 2003 (Table 1). The 2004 HGHT means were substantially greater in 2004, which may be attributable to better rainfall (i.e., 2002 and 2003 were severe drought years) and/or longer plant establishment.
Heritabilities over years (Table 2) for HGHT were intermediate between CIRC and the two flowering traits (ANTH and BOLT). Likewise, the ratio of genotype x year interaction to phenotypic variance for HGHT (0.13) was intermediate between CIRC (0.02) and the flowering traits, ANTH (0.18) and BOLT (0.21). Weak positive genotypic correlations between HGHT and BOLT were detected in TTC1 and TTC2 families, but no other trait correlations with HGHT were detected in both families.
We detected a total of seven HGHT QTLs in the TTC1 and/or TTC2 families (Table 4, Fig. 1). The TTC2 HGHT QTL on LG3a was split into two QTLs using a combination of rMQM and unrestricted MQM mapping (Table 4, Fig. 1). The single largest HGHT QTL was located in homologous regions of LG2a in both TTC1 and TTC2 families. Likewise, overlapping TTC1 and TTC2 HGHT QTLs on LG5Xm were too close to separate (Fig. 1), which we count as one homologous QTL. Moreover, LG5Xm was the only chromosome that contributed negative HGHT QTL alleles from the taller L. cinereus species, which also seems to support our interpretation that these are homologous QTLs in the TTC1 and TTC2 families. Thus, L. cinereus contributed six positive and one negative HGHT QTL alleles in the TTC1 and/or TTC2 families, which is consistent with divergent phenotypes of the parental species.
|
DISCUSSION |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONREFERENCES |
|---|
These are the first QTL analyses conducted using the high-density linkage maps published by Wu et al. (2003). Throughout much of the linkage map, relatively small irregularities in the LOD scans (Supplemental Data) resulted from imperfect genotyping, missing data, and ambiguous marker orders. Nevertheless, these QTL maps provide a useful starting point for high-resolution QTL mapping experiments using six advanced backcross populations that are currently being genotyped and evaluated in clonally replicated field trials.
Comparison of TTC1 and TTC2 QTLs
Two circumference, two bolting, and two height QTLs were evidently homologous in both TTC1 and TTC2 families. Conversely, two circumference, seven bolting, all five anthesis date, and five height QTLs were unique to TTC1 or TTC2 families. Differences between the TTC1 and TTC2 families can be attributed to differences between the TC1 and TC2 hybrids, which in turn can only be attributed to genetic variation within the L. cinereus Acc:636 and L. triticoides Acc:641 natural germplasm sources. Thus, functionally important QTL variation was apparent within and between natural source populations of these experimental families and species.
Growth Habit
The coincidence of QTL effects near the VP1 loci on LG3a and LG3b, in both TTC1 and TTC2, suggest that these QTLs may be controlled by homoeologous copies of the one gene. If so, this gene evidently controls much of the dramatic difference in growth habit between L. cinereus, L. triticoides, and perhaps other grasses. The only other CIRC QTLs, on LG5Xm and LG6a, were unique to the TTC1 and TTC2 families, respectively. A preponderance of mapped barley mutations (four of six loci) that affect vegetative axillary development have been localized to chromosome 3HL (minus arm), including absent lower laterals (als), low number of tillers1 (int1), and semi brachytic (usu), which produce fewer tillers, in addition to granum-a (gra-a), which produces significantly more tillers (Babb and Muehlbauer, 2003; Franckowiak, 1996). A recessive gravitropic lazy dwarf gene (lzd) was also mapped to the short arm of barley chromosome 3 (Franckowiak, 1996; Takahashi et al., 1975), but this mutation seems to be more proximal to the centromere than the Leymus LG3a and LG3b growth habit QTLs. Chromosome 3 is highly conserved within the Triticeae (Devos and Gale, 2000). Colinear from end to end with rice chromosome 1, Triticeae group three is also the most conserved of all chromosome groups when compared with rice (La Rota and Sorrells, 2004). Rice chromosome 1 contains putative transmembrane auxin efflux carrier and DNAJ-like genes (International Rice Genome Sequencing Project, 2005), near the Vp1 locus, originally identified in Arabidopsis pin-forming (Gälweiler et al., 1998) and arg1 (altered response to gravity) (Sedbrook et al., 1999) mutants, respectively. However, pin1- and arg1-like sequences are also present in other regions the rice genome. Rhizome (Hu et al., 2003) and tiller angle (Li et al., 1999) QTLs also map to rice chromosome 1, but these rice loci were not located near the rice Vp1 gene.
The TTC2 CIRC rMQM QTL on LG5Xm mapped near the TCP-like sequence (TCP2) amplified from L. triticoides rhizome cDNA using primers designed from the teosinte branched one gene (Doebley et al., 1997), BCD1130, BCD1707, and PSR128 loci. The PSR128 marker also maps in the centromere region of maize chromosome 4 near the recessive lazy1 (la1) gene (Lawrence et al., 2004). Lazy maize mutants, allelic to la1, actively grow downward (prostrate) under light (Firn et al., 2000). We found BCD1707 and BCD1130 sequences on rice chromosome 11, using a TBLASTX search (Ware et al., 2002) of the rice genome (International Rice Genome Project, 2005), near another lazy (la) gene (Miura et al., 2003). The centromere and short arm regions of Triticeae group 5 are homoeologous to rice chromosomes 9 and 12, respectively (Van Deynze et al., 1995; La Rota and Sorrells 2004). Rice chromosome 9 contains a major QTL (Ta) for tiller angle (Li et al., 1999).
The TTC1 CIRC QTL peak on LG6a is located approximately (minus) 40 cM from the MWG2264 locus. Interestingly, the uniculm (cul2) mutation and MWG2264 loci are both located (minus) 8.8 and 3 cM relative to the cMWG679 locus on barley 6H (Babb and Muehlbauer, 2003; Graner et al., 1991; Graner, 2004). Compared with other barley mutants, uniculm2 (cul2) is unique in that it inhibits formation of axillary meristems and does not produce tillers (Babb and Muehlbauer, 2003).
We were not able to identify rhizome QTLs in Leymus that were homeologous to rhizome QTLs of Sorghum and Oryza. Paterson et al. (1995) detected about 12 rhizome QTLs including a conspicuous cluster of QTLs affecting rhizome number, rhizome distance, and seedling tillers on Sorghum linkage group C, which corresponds to wheat group 4 (Draye et al., 2001). The latter Sorghum QTL also corresponds with sucker and stalk number QTL in Saccharum (Jordan et al., 2004), in addition to the Rhz2 gene and 11 other QTLs controlling rhizome proliferation differences between O. sativa L. and O. longistaminata A. Chev. & Roehr. (Hu et al., 2003). We did not detect any plant circumference QTLs on Leymus LG 4Ns or LG4Xm, which include the PRC140 rhizome QTL marker (Draye et al., 2001) derived from a Sorghum LG C and sequences orthologous to the maize teosinte branched one gene (Doebley et al., 1997). The Rhz3 gene and rhizome QTL on Oryza chromosome 4 correspond to rhizome QTLs on Sorghum LG D (Hu et al., 2003) and Saccharum (Jordan et al., 2004). This region of Oryza chromosome 4 and Sorghum LG D evidently corresponds to Triticeae group 2 (Paterson et al., 1995; Van Deynze et al., 1995). We did not detect any significant rhizome QTL effects on Triticeae group 2.
Bolting and Anthesis Date
The genetic control of flowering time is complex, involving three major groups of genes on all seven Triticeae chromosome groups (Snape et al., 2001). Two of these groups interact with the environment, namely those controlling vernalization (Vrn) and photoperiod (Ppd). The third set of genes controls developmental rate independent of vernalization and photoperiod, so-called "earliness per se" (Eps) genes. A relatively large number of Eps genes have been mapped to all seven homoeologous groups of wheat and/or barley (Laurie et al., 1995; Snape et al., 2001), several of which may correspond to BOLT and/or ANTH QTLs detected in Leymus.
The Vrn genes map to Triticeae groups 1, 4, 5, and 7 (Dubcovsky et al., 1998; Snape et al., 2001). Moreover, the two most potent genes, on groups 4 and 5, have been cloned and characterized (Yan et al., 2003; Yan et al., 2004). Surprisingly, no significant bolting or anthesis date QTLs were associated with the Leymus CDO504 markers on LG5Ns and LG5Xm, which are presumably closely linked with genes that are orthologous to VRN1. We have amplified VRN1 sequences from Leymus but have not yet found polymorphisms needed for mapping. Likewise, no significant bolting or anthesis date QTLs were associated with the VRN2 gene on LG5Ns. A second yet undetected VRN2 gene may or may not be present on the Xm genome. The most conspicuous wheat and barley Ppd genes are located on homoeologous regions of the 2A, 2B, 2D, and 2H short (plus) chromosome arms. Other Ppd genes have also been reported on barley 1H (Laurie et al., 1995) and 6H (Strake and Börner, 1998). We did not detect ANTH or BOLT QTLs on short (minus) arm of Leymus group two chromosomes.
The TTC1 and TTC2 families displayed substantial genetic variation for anthesis date and bolting, including genotypes that have not flowered in the field, which was not apparent among the parental species accessions. This transgressive segregation can be explained by a mixture of antagonistic (i.e., positive and negative) QTL alleles from each parental species. Leymus cinereus contributed five positive and four negative BOLT QTLs and three positive and two negative ANTH QTLs. Thus, correspondence of bolting and anthesis date QTLs (and relatively strong negative genotypic correlations) suggest that outbreeding depression (i.e., transgressive genotypes that to flower) was associated with flowering (i.e., lateness) genes under balancing selection. Although population means and standard deviations may not demonstrate significant bolting depression, the fact that a substantial number of progeny failed to flower year after year indicates that some outbreeding depression is occurring. We have not observed this phenomenon within the L. cinereus Acc:636 or L. triticoides Acc:641 natural source populations.
Plant Height
The HGHT QTLs on LG2a displayed remarkably strong effects and similar map locations in both TTC1 and TTC2 families, suggesting that these QTLs are homologous. The LG2a HGHT QTL was the only major positive HGHT effect associated with chromatin of the much taller L. cinereus species and detected in both TTC1 and TTC2 families. We interpret these results to mean that this is one of the key QTLs controlling relatively large differences in plant height between L. cinereus and L. triticoides. Börner et al. (1999) mapped gibberellin-sensitive gal (GA-less) and gibberellin-insensitive gai (GA-ins) mutations about 55 cM apart on barley 2H. The Leymus LG2a HGHT QTL spans a 40 cM interval from the XANTHA locus down below the CNL045 locus (Fig. 1). The XANTHA–CNL045 LG2b interval also includes the cWMG763 locus (Fig. 1), which was mapped to the short (plus) arm of barley 2H about 36 cm distal from the MWG2287 marker in the barley IGRI x FRANKA population (Graner et al., 1991; Graner, 2004). The MWG2287 locus is closely linked to the barley gai mutation in the centromeric region of barley chromosome 2H (Börner et al., 1999). We speculate that the gai (sdw3) locus, described by Börner et al. (1999) and Gottwald et al. (2004), may correspond with a major plant height QTL associated with the cluster of DNA markers near the CNL045 locus of LG2a. The CNL045 loci are associated with high-density clusters of markers on Leymus LG2a and LG2b linkage groups, probably caused by reduced recombination near the centromere. Yang et al. (1995) also described a partially dominant GA-ins dwarfing gene on the short arm of chromosome 2A of wheat (Rht21), which could be homeologous to the gai mutation mapped by Börner et al. (1999). In any case, the barley, wheat, and Leymus plant height genes on homeologous group two are evidently different from the so-called "green revolution" GA-ins dwarfing genes of wheat, barley, and rice encode gibberellin response modulators that map to Triticeae group four (Peng et al., 1999).
Description of additional SSR (CNL and KSU) and STS markers added to the Leymus TTC1 and TTC2 maps originally published by WU et al. (2003)
|
Amplicon size based on comparisons with internal size standards in capillary electrophoresis.
FAM 5' labeled primer.
Amplicon size based on sequence data.
|
ACKNOWLEDGMENTS |
|---|
Received for publication December 13, 2005.
|
REFERENCES |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONREFERENCES |
|---|
ák, C.N. Chinoy, and M.D. Gale. 1995. Structural evolution of wheat chromosome 4A, 5A, and 7B and its impact on recombination. Theor. Appl. Genet. 91:282–288.ák. 1996. Genetic map of diploid wheat, Triticum monococcum L., and its comparison with maps of Hordeum vulgare L. Genetics 143:983–999.[Abstract]
0.05 signficance level, detected using a restricted multiple QTL model (rMQM) are indicated by black or white bars for each trait x year. The updated molecular genetic linkage maps include 488 homologous AFLP markers and mapped in both TTC1 and TTC2 families (