RPViewDoc.pdf

354

A framework linkage map of perennial ryegrass

based on SSR markers

G.P. Gill, P.L. Wilcox, D.J. Whittaker, R.A. Winz, P. Bickerstaff, C.E. Echt, J. Kent,

M.O. Humphreys, K.M. Elborough, and R.C. Gardner

Abstract: A moderate-density linkage map for Lolium perenne L. has been constructed based on 376 simple sequence

repeat (SSR) markers. Approximately one third (124) of the SSR markers were developed from GeneThresher ® librar-

ies that preferentially select genomic DNA clones from the gene-rich unmethylated portion of the genome. The remain-

ing SSR marker loci were generated from either SSR-enriched genomic libraries (247) or ESTs (5). Forty-five percent

of the GeneThresher SSRs were associated with an expressed gene. Unlike EST-derived SSR markers, GeneThresher

SSRs were often associated with genes expressed at a low level, such as transcription factors. The map constructed

here fulfills 2 definitions of a “framework map”. Firstly, it is composed of codominant markers to ensure map transfer-

ability either within or among species. Secondly, it was constructed to achieve a level of statistical confidence in the

support-for-order of marker loci. The map consists of 81 framework SSR markers spread over 7 linkage groups, the

same as the haploid chromosome number. Most of the remaining 295 SSR markers have been placed into their most

likely interval on the framework map. Nine RFLP markers and 1 SSR marker from another map constructed using the

same pedigree were also incorporated to extend genome coverage at the terminal ends of 5 linkage groups. The final

map provides a robust framework with which to conduct investigations into the genetic architecture of trait variation in

this commercially important grass species.

Key words: Framework map, perennial ryegrass, SSR, simple sequence repeat, GeneThresher, Lolium perenne .

364

Résumé : Une carte génétique de densité moyenne a été produite pour le Lolium perenne L. à l’aide de 376 microsa-

tellites (SSR). Environ un tiers (124) des microsatellites ont été développés à partir de banques GeneThresher qui sont

enrichies en clones d’ADN génomique provenant des régions hypométhylées et riches en gènes au sein du génome. Les

autres microsatellites ont été obtenus soit de banques génomiques enrichies en SSR (247) ou d’EST (5). Quarante-cinq

pour cent des microsatellites GeneThresher étaient associés à un gène qui s’exprime. Contrairement aux microsatellites

dérivés d’EST, ceux provenant de la banque GeneThresher étaient souvent associés à des gènes exprimés à un faible

niveau tel que des facteurs de transcription. La carte produite répond à 2 critères définissant une carte de référence.

D’abord, elle est composée de marqueurs codominants ce qui assure la transportabilité intra- et interspécifique. Deuxiè-

mement, la carte a été assemblée de façon telle que l’ordre des gènes est appuyé statistiquement. La carte compte 81

marqueurs de référence étalés sur sept groupes de liaison ce qui correspond au nombre de chromosomes. La plupart

des 285 autres marqueurs ont été assignés à l’intervalle le plus vraisemblable au sein de la carte de référence. Neuf

marqueurs RFLP et 1 microsatellite d’une autre carte produite avec le même pedigree ont été incorporés de manière à

accroître la couverture des extrémités de 5 groupes de liaison. La carte finale offre une carte de référence robuste avec

laquelle il sera possible de réaliser des études de l’architecture génétique de la variation phénotypique chez cette gra-

minée d’une grande importance commerciale.

Received 12 July 2005. Accepted 28 November 2005. Published on the NRC Research Press Web site at http://genome.nrc.ca on

28 April 2006.

Corresponding Editor: G.J. Scoles.

G.P. Gill, 1 D.J. Whittaker, R.A. Winz, 2 P. Bickerstaff, 3 and K.M. Elborough. 4 ViaLactia Biosciences, PO Box 109-185,

Newmarket, Auckland, New Zealand.

P.L. Wilcox 1 and C.E. Echt. 5 Cell Wall Biotechnology Center, New Zealand Forest Research Institute Ltd., Private Bag 3020,

Rotorua, New Zealand.

J. Kent. 6 SignaGen, New Zealand Forest Research Ltd., Private Bag 3020, Rotorua, New Zealand.

M.O. Humphreys. Institute of Grassland and Environmental Research, Plas Gogerddan, Aberystwyth, Wales SY23 3DA, United

Kingdom.

R.C. Gardner. 1 School of Biological Sciences, University of Auckland, Private Bag 92019, Auckland, New Zealand.

1 Authors contributed equally to this work.

2 Present address: Department of Chemistry, Wine Science, University of Auckland, Private Bag 92019, Auckland 1020, New Zealand.

3 Present address: Infoics Limited, PO Box 83153, Edmonton, Auckland 1230, New Zealand.

4 Corresponding author (e-mail: kieran.elborough@vialactia.com).

5 Present address: USDA Forest Service, Southern Institute of Forest Genetics, 23332 Highway 67, Saucier, MS 39574-9344, USA.

6 Present address: Applied Biosystems, 52 Rocco Drive, Scoresby, Victoria 3179, Australia.

Genome 49 : 354–364 (2006)

doi:10.1139/G05-120

Gill et al.

355

Mots clés : carte génétique de référence, ivraie vivace, SSR, répétition de séquences simples, GeneThresher, Lolium pe-

renne .

[Traduit par la Rédaction] Gill et al.

Introduction

Perennial ryegrass ( Lolium perenne L.) is the most impor-

tant temperate pasture species for providing forage to the

dairy, wool, and meat industries and is considered a valuable

turf and amenity grass. Lolium perenne is a diploid obligate

outbreeder that maintains a high degree of genetic diversity

in natural and agricultural populations (Roldan-Ruiz et al.

2000). Hybridization with other related species, such as

Lolium multiflorum (Italian ryegrass), Festuca arundinacea

(tall fescue), and Festuca pratensis (meadow fescue), pro-

vides a valuable resource for the introgression of commer-

cially favourable traits.

As with other crops, there has been considerable interest in

applying molecular marker technology for marker-assisted se-

lection (MAS) to improve selection efficiency compared with

conventional breeding methods. Perennial ryegrass displays

continuous phenotypic variation for most target traits that are

controlled by quantitative trait loci (QTL). The genome posi-

tion and the number of QTL accounting for the majority of

genetic variation for a trait may be determined through the

use of genetic mapping to identify linked marker loci. An

important prerequisite is a genetic map that provides ade-

quate genome coverage and ideally uses markers that are

transportable across different pedigrees. Several “compre-

hensive” genetic maps exist for perennial ryegrass that are

predominantly based on a combination of restriction frag-

ment length and amplified fragment length polymorphism

(RFLP and AFLP) markers (Bert et al. 1999; Armstead et al.

2002; Jones et al. 2002 a ). Simple sequence repeats (SSRs)

are favoured over other DNA marker types for the genera-

tion of genetic maps as they are easy to use, codominant,

highly polymorphic, and are often transportable to other

pedigrees. A more recent perennial ryegrass genetic map

contains SSR markers derived from ESTs (Faville et al.

2004) that can potentially provide markers that are also

functionally associated with trait variation.

The term “framework map” has been used in at least 2

contexts. One context is to describe maps constructed with

codominant markers to ensure map transferability either

within or among species. Another context that has been used

is statistical: here, the order of marker loci on the final map

achieves a predetermined level of statistical confidence in

the final order of selected marker loci. This concept has

been advocated for linkage mapping in humans by Keats et

al. (1991), who used the term “support-for-order” to quantify

the level of confidence in the final gene order relative to the

next most-likely order for the same set of marker loci. A

generally accepted figure is a log-likelihood of odds (LOD)

difference of 3.0, which corresponds to a selected gene order

being 1000 times more likely than the next most-likely order

on a group-wise basis. The framework map concept has also

been widely applied in the construction of linkage maps in

forest tree species (Grattapaglia and Sederoff 1994; Echt and

Nelson 1997; Remington et al. 1999; Wilcox et al. 2001). In

contrast, comprehensive maps are those that include all loci

without applying any support criteria. Since the quality of

marker data can vary from one locus to another, inclusion of

poorer-quality data can lead to artificial map expansion, and

the overestimation of linkage group length and genome size.

The ryegrass genetic maps described above are of the com-

prehensive type and do not meet all the statistical criteria of

a framework map concept (Keats et al. 1991; Liu 1998).

Perennial ryegrass has an estimated genome size of 5.66

pg/2C (Arumuganathan et al. 1999). The repetitive DNA

content is expected to be in the order of 75%, similar to that

reported for maize (San Miguel et al. 1996), which has a

comparable genome size. Repetitive DNA poses a challenge

to complete genomic sequencing and the alternative EST-

based approaches typically miss 40%–50% of genes as a

result of their low expression level or cell-type specificity

(Bonaldo et al. 1996). A targeted-sequencing strategy for gene-

rich regions using methyl-filtration technology of plant

genomic libraries (termed GeneThresher ® by Orion Genomics,

St. Louis, Mo.) removes methylated “junk DNA”, leaving

only the small unmethylated portion of the genome containing

genes (Rabinowicz et al. 1999). GeneThresher libraries are

slightly enriched for SSRs (Whitelaw et al. 2003; Bedell et al.

2005), thereby providing an excellent source of genetic mark-

ers that are often associated with genes.

This paper describes a robust genetic-linkage map for pe-

rennial ryegrass based on 376 SSR markers and meets the

criteria of the framework map concept. Density and cover-

age of this map is such that it should be immediately useful

for applications in molecular breeding, as well as for more

basic applications such as comparative genomics.

Materials and methods

Mapping pedigree and SSR marker development

The F 2 -mapping population used in this experiment was

derived by selfing an F 1 hybrid between 2 partially inbred

parental lines (produced from 4 generations of selfing). Pa-

rental lines were subject to opposing selection for carbohy-

drate content (Turner et al. 2001) and the same pedigree

(including some of the same genotypes used in this study)

was used to construct an RFLP-based map (Armstead et al.

2002).

SSR markers were developed from 3 sources. Genetic

Identification Services (Chatsworth, Calif.) provided en-

riched genomic libraries for CA, GA, TAGA, AAG, and

TACA repeats using the method described by Jones et al.

(2000). Secondly, a perennial ryegrass genomic sequencing

project using methyl-filtrated GeneThresher libraries (sup-

plied by Orion Genomics) was mined for appropriate SSRs.

Finally, an in-house EST database comprising 12 different

libraries prepared from a range of tissue and treatment types

was searched for SSRs that were suitable for marker devel-

opment. SSRs from each library source were evaluated for

redundancy and primer design using an in-house-developed

bioinformatics pipeline. The cut-off threshold for the num-

ber of repeats was at least 8 for dinucleotides, 7 for tri-

356

Genome Vol. 49, 2006

nucleotides, 6 for tetranucleotides, and 5 for penta-

nucleotides and hexanucleotides. Academic research licenses

are available for primer sequences of the 81 framework SSR

markers (ViaLactia Biosciences, New Market, Auckland,

New Zealand).

PCR amplification and product electrophoresis were per-

formed by a contract genotyping provider (SignaGen,

Rotorua, New Zealand) as outlined below. SSR primers were

labelled using either 6-FAM, VIC, NED, or PET fluoro-

chrome moieties (Applied Biosystems, Foster City, Calif.).

PCR was performed in a total volume of 10

re-evaluated and the RIPPLE analysis repeated until final

support-for-order exceeded 3.0. Criteria for dropping mark-

ers from the framework included map contraction (deter-

mined using the DROP MARKER command) and

consistency with triangular equality. The most-likely inter-

vals where the dropped markers (denoted “accessory” mark-

ers) were located was estimated using the TRY command,

and the nearest framework marker identified using the

NEAR command.

Markers that amplified only 1 band and segregated in

a 3:1 manner with the other allele having no discernable

amplicon are referred to as “pseudointercross” (PI) markers.

In these cases, the F 1 was assumed to be a heterozygote for

null alleles. Pseudointercross (PI) markers were mapped by

recoding the F 2 data to phase-unknown PI data (i.e., scored

twice, by assigning different homozygous classes as domi-

nant and recessive classes), and merging this with the 3:1 PI

data.

Markers that were linked to framework markers, but that

could not be placed into specific intervals, have been de-

noted as “floating” markers and were assigned to the chro-

mosome showing strongest linkage. Map distances were

calculated by Mapmaker using the Kosambi mapping func-

tion (Kosambi 1944). Segregation distortion of markers was

checked using standard

mol/L of each primer, 0.5 U of Ampli Taq Gold

DNA polymerase (Applied Biosystems) and 1× GeneAmp

PCR buffer I. A touchdown thermocycling protocol was

used as follows: initial denaturation at 92 °C for 9 mins; 2

cycles of 94 °C for 1 min, 65 °C for 1 min, and 70 °C for

35 s; 18 cycles of 93 °C for 45 s and 64 °C for 45 sec (tem-

perature was reduced by 0.5 °C/cycle until it reached

55.5 °C); 70 °C for 45 s; 20 cycles of 92 °C for 30 s, 55 °C

for 30 s, and 70 °C for 60 s; and a final extension at 70 °C

for 20 min. Products were separated using an ABI3100

capillary sequencer (Applied Biosystems) and sized using

either the GS-500 ROX or GS-500 LIZ ladder (Applied

Biosystems). Allele scoring was performed using Genotyper

v. 3.7 (Applied Biosystems).

SSR markers were evaluated on a panel consisting of a

single genotype from 6 New Zealand cultivars (data not

shown) and DNA from the sibs of each grandparent of the

mapping pedigree (grandparent DNA was not available).

Data from Armstead et al. (2002) was used to extend map

coverage (described below).

2 goodness-of-fit measures for both

the F 2 and PI data. A genome-wide threshold corresponding

to a comparison-wise p value of 0.01 was used to approxi-

mate an experiment-wise p value of 0.05.

Once a map was constructed according to the above crite-

ria, marker coverage was compared with that of an existing

F 2 RFLP-based map constructed by Armstead et al. (2002).

Gaps in genome coverage of the SSR framework map were

identified, and markers from the RFLP-based map that were

located in these gaps were incorporated. To do this, we used

the TRY command to locate the most-likely interval the new

markers should fall within, and re-ordered the group accord-

ing to the framework map criteria described above. Marker

placements were accepted where ( i ) map coverage was

extended, ( ii ) the final order incorporating the additional

marker achieved a minimum support-for-order of LOD 3.0,

and ( iii ) the addition of the new marker did not change the

pre-existing framework SSR marker order. A total of 20

markers from the F 2 map were initially selected for this

analysis.

Framework map construction

To construct a suitably robust map of markers with a com-

bination of segregation patterns, we chose a strategy where

F 2 markers only were used to construct a preliminary frame-

work. Since linkage phase was uncertain in a portion of the

F 2 data (denoted “F 2 phase-unknown”), any unlinked mark-

ers (after the first analyses were conducted) were inverted by

reversing homozygote categories and analysing the data with

both phases. Here we assumed that the markers with the in-

correct phase assignments would not be linked to any of the

markers where the correct phases were known (“F 2 phase-

known”). Marker groups were determined using a minimum

LOD ratio of 6.0 and a maximum recombination rate of

0.20. For groups with more than 7 markers, optimal order

was estimated using a matrix correlation method as imple-

mented in Mapmaker (Lander et al. 1987) Macintosh v. 2.0

via the FIRST ORDER command. For groups with 7 mark-

ers or fewer, the COMPARE function was used to determine

the best possible order based on a group-wise multipoint

log-likelihood value. The most likely order for the group

was next estimated by permuting or shuffling positions 3

markers at a time using the RIPPLE command. Orders with

log-likelihood differences of 3.0 or more compared with the

next most-likely order (support-for-order) were accepted for

framework loci. Where the log-likelihood difference for a

group of 3 markers was less than 3.0, markers were dropped

one at a time from regions where support-for-orders were

less than 3.0. After each marker was dropped, orders were

Results

Marker yields and sequence analysis

The proportion of markers developed and mapped from

each library source is given in Table 1. The frequency of

non-redundant SSRs present in the GeneThresher, enriched,

and EST libraries was 0.83%, 25.4%, and 1.8%, respec-

tively, when using the repeat-length criteria described in the

Materials and methods section. The number of suitable SSR

clones for marker development was severely reduced once

primer-design criteria were met. The EST library source was

most affected, with only 20.3% of non-redundant SSR

clones passing primer-design criteria. Efficiency loss owing

to primer design was mainly attributed to shortage of flank-

ing sequence (truncated clones) adjacent to the SSR. All

SSRs identified from the EST library source were confined

L containing

10 ng of genomic DNA, 0.2 mmol/L dNTPs, 1.5 mmol/L

MgCl, 0.25

Gill et al.

357

Table 1. Proportion of SSR markers developed from GeneThresher, EST, and SSR-enriched libraries.

Library

Clones

sequenced

Non-redundant

SSR yield

Primers

designed a

Library

yield b

(%)

Polymorphic in

screening panel c

Mapped d

GeneThresher 155 084

1287 (0.83%)

563 (43.7%) 0.36

258 (45.8%)

122 (21.7%)

SSR-enriched

6528

1658 (25.4%)

931 (56.2%) 14.3

355 (38.1%)

209 (22.4%)

EST

6596

118 (1.8%)

24 (20.3%) 0.36

9 (37.5%)

5 (20.8%)

a The proportion of non-redundant SSRs for which primers were designed (i.e., passed primer-design criteria).

b Percentage of SSR clones for which primers were designed out of the total number of clones sequenced in the library.

c Proportion of SSRs for which primers were designed that displayed polymorphism in a screening panel consisting of 8 L. perenne

cultivars.

d Proportion of SSRs for which primers were designed that mapped in the F 2 population. An additional 39 loci contributed by the

same SSRs amplifying more than one product are not included in the total.

untranscribed region (UTR), which often lim-

ited the amount of potential sequence available for primer

design. GeneThresher and EST libraries were of similar effi-

ciency for yielding SSRs that passed primer-design criteria

(0.36%) when the total number of clones sequenced in each

library type was considered. As expected, SSR-enriched

genomic libraries were the most efficient source (14.3%)

based on the total number of clones sequenced in each li-

brary type. The proportion of SSRs that had primers de-

signed and could be mapped in the population used for this

study did not differ greatly depending on the library source

(21–23%).

Our latest GeneThresher data set consists of 446 960 se-

quence reads, assembled into 80 162 contigs and 189 697

singletons. BLASTn analysis of the mapped GeneThresher

SSR singleton or contig sequence against the GenBank nr

and EST_other nucleotide databases determined that 44.6%

of GeneThresher SSRs were most-likely associated with an

expressed gene ( E <7×10 –7 ). Thirty-eight percent of the

GeneThresher SSR sequences mapped gave very high gene

similarity BLASTn hits ( E <2×10 –20 ). Given that

GeneThresher sequences often contained non-coding se-

quences such as introns and promoters, BLASTn scores of

E <7×10 –7 against the GenBank databases were signifi-

cant. A portion of the total GeneThresher SSRs mapped

were associated with a putative rice gene ortholog with clear

annotation (18.5%) once aligned to the rice genome at

Gramene (http:\\www.gramene.org) using BLASTn (Ta-

ble 2). Regions of alignment can be viewed by searching for

the rice TIGR locus ID in the genome browser and selecting

the ryegrass_methylfilter option from the GSS feature menu.

From the subset of mapped markers giving clear annotation,

approximately 35% of these belonged to the transcription

factor class of genes. Perennial ryegrass SSRs developed

from enriched genomic library sources corresponded to

genes only 6% of the time (BLASTn E <1×10 –7 ).

or 3

′

same markers. These were considered phase-unknown F 2

loci. Finally, 146 loci segregated in a 3:1 manner and are re-

ferred to as phase-unknown PI markers. Some of the primer

pairs (31) amplified more than one segregating locus provid-

ing a total of 70 marker loci. A total of 52 F 2 and 17 PI

markers showed segregation distortion (comparison-wise

< 0.01). Of these markers, all but 10 mapped to either

linkage group 5 or 7 (Fig. 1).

Using the procedure for framework map construction, as

described in the Materials and methods section, all F 2 mark-

ers were tested for suitability as framework markers. Of

these F 2 SSR markers, 81 met the criteria for acceptance as

framework markers. For the phase-unknown F 2 markers,

linkages were found for only 1 phase at the LOD threshold

used to assign markers to groups. Properties of a subset of

the framework SSR markers are described in Table 3. An ad-

ditional 233 F 2 and PI markers were placed as accessory

markers. A number of both F 2 and PI markers (57) were

linked to framework markers, but could not be placed into

specific intervals. These markers typically showed closest

linkage to non-terminal framework markers, but showed

more recombination to these markers than the immediately

adjacent flanking framework markers and were hence de-

noted as “floating” markers. Ten markers were unlinked af-

ter this initial analysis that provided 7 linkage groups — the

same number as the haploid chromosome number in peren-

nial ryegrass.

To check genome coverage of the SSR markers, the

framework map was aligned with a predominantly RFLP-

based map constructed with 180 progenies (Armstead et al.

2002), which included the same 94 F 2 individuals that were

used for this study. Alignment revealed a few regions located

at the terminal ends of linkage groups that were not covered

by the SSR-only framework map. To increase map coverage,

we then added 10 markers (9 RFLPs and 1 SSR) from re-

gions not covered, and incorporated these using the same

analyses as the SSR markers. The addition of these markers

to the framework map allowed 6 markers that were origi-

nally unlinked or floating to be placed as accessory markers.

A final summary listing the proportion of markers derived

from each category and the subsequent linkage group length

is provided in Table 4. The final map (Fig. 1) shows frame-

work loci, as well as accessory markers, with the interval

that they are most likely to be placed within. Floating mark-

ers are shown below the group to which they show the stron-

gest linkage. Markers showing segregation distortion are

also denoted. The 7 linkage groups initially covered a total

map length of 484.9 cM when only SSR markers were con-

Data analysis and map construction

No DNA was available for genotyping from either grand-

parents (F 0 ), although DNA from a sib of each of the grand-

parents was available. A total of 94 F 2 progenies were

genotyped for 381 SSR loci. Almost half (172) segregated in

a 1:2:1 ratio consistent with a phase-known F 2 population,

i.e., where the genotype of the grandparents’ sib was consis-

tent with those of the parents. In addition, another 63 loci

were segregating in a 1:2:1 ratio consistent with an F 2 segre-

gation pattern, but one or both of the grandparental sib geno-

types were not consistent with the alleles segregating at the

to the 5

′

358

Genome Vol. 49, 2006

Table 2. Putative gene function and map location of perennial ryegrass GeneThresher-SSR markers.

Map location

GeneThresher

SSR marker Putative gene function

Linkage

group

Map distance

(cM)

Rice TIGR locus

Transcription factor class

rv0904

Knotted homeodomain protein

6.7–20.2

LOC_Os06g43860

rv0922

Teosinte-branching 1 gene (TB1)

61.3–71.8

LOC_Os03g49880

rv1087

MADs box

53.6–82.4

LOC_Os03g54170

rv1159

Zinc-finger homeobox protein

68.6–71.9

LOC_Os11g03420

rv1190

YABBY transcription factor

50.3

LOC_Os03g11600

rv1316

NAM (no apical meristem) protein 3

39–43.6

LOC_Os07g37920

rv1385

GL2-type homeodomain protein

66.1

LOC_Os02g45250

rv1412

GT-2 transcription factor

0–20.3

LOC_Os03g02240

Other

rv0983 Cytochrome P450 6 62.2 LOC_Os05g12040

rv1063 Aluminium tolerance S222 gene 3 49.6 LOC_Os02g37590

rv1117 BEACH-domain-containing protein 2 — LOC_Os04g46900

rv1139 Microtubule-associated protein 5 59–64.8 LOC_Os09g27700

rv1149 Integral membrane protein 2 57.6 LOC_Os04g58760

rv1131 Fasiclin-like arabinogalactan 3 57.3 LOC_Os05g38500

rv1175 Transducin protein 7 44.8–49.9 LOC_Os08g07960

rv1242 Glycosyl hydrolase 1 27.1 LOC_Os05g31140

rv1278 Pectin acetylesterase 6 — LOC_Os10g40290

rv1282 Xyloglucan endotransglycosylase 2 — LOC_Os04g51460

rv1307 DMC1 meiotic recombination 5 68.6–71.9 LOC_Os12g04980

rv1317 Ubiquitin carrier protein 4 — LOC_Os01g16540

rv1338 Receptor-like protein kinase 7 50.2–52.3 LOC_Os04g44910

rv1384 Copper chaperone protein 2 — LOC_Os03g26650

rv1438 Sugar transporter protein 2 54.9 LOC_Os04g41460

Note: Only GeneThresher-SSR markers giving clear annotation at BLASTn scores of E <7×10 –7 are listed. Map

location for several accessory markers is given as an interval range, whereas floating markers were assigned only to

the appropriate linkage group. The TIGR locus ID corresponds to the highest BLASTn match from rice obtained dur-

ing a genome alignment of the GeneThresher perennial ryegrass sequence at Gramene (http:\\gramene.org). Regions of

alignment can be viewed using the genome browser and selecting the ryegrass_methylfilter option from the GSS fea-

ture menu.

sidered and 675.6 cM once markers for extending terminal

ends of linkage groups were incorporated.

framework maps constructed with independent progeny sets

descended from the same parent in maritime pine (Plomion

et al. 1995) using analytical methods similar to those de-

scribed here showed that some differences in framework or-

der do occur, but these were relatively infrequent and similar

to that expected owing to chance. Indeed, while the support-

for-order criterion of 3.0 is based on local orders, the multi-

ple number of chromosomes effectively increases the proba-

bility of a spurious order, but not by a large amount for a

genome with so few linkage groups. We calculate that the

genome-wise support-for-order corresponding to group-wise

LOD = 3 for 7 linkage groups is approximately LOD 2.15

(based on Bonferroni corrections for the number of linkage

groups). This indicates that, on a genome-wide basis, the de-

picted orders are more than 100 times more likely than the

next most-likely order, based on the genotypic data supplied.

Slight changes in marker order that could result, albeit with

low probability, may nonetheless be relatively trivial, de-

pending of course on the specific application of the map.

Preliminary marker data obtained in a second perennial

ryegrass mapping population (data not shown) was sufficient

to compare framework marker order of 2 linkage groups. All

framework markers in common retained their order (8 from

Discussion

We have constructed a moderate-density map of the

ryegrass genome using 376 SSR markers, complemented by

9 RFLP markers. The analytical approach we have used to

construct linkage groups differs from that of other (pub-

lished) ryegrass maps. Our approach is based on ordering

markers with a given minimum level of confidence in gene

order, and retaining in the resulting framework only those

markers where order achieves a preset level of confidence in

order (support-for-order > LOD 3.0). In so doing, we were

able to identify markers thatwere problematic in their place-

ment on the framework for a variety of reasons. This map

should therefore provide a reasonably robust basis with

which to apply to other pedigrees of this species and for

comparing genomic rearrangements with closely related spe-

cies. Such approaches have been used extensively in other

species, mostly in constructing linkage maps of forest trees

(Grattapaglia and Sederoff 1994; Echt and Nelson 1997;

Remington et al. 1999; Wilcox et al. 2001). Comparison of

Plik z chomika:

Inne pliki z tego folderu:

Inne foldery tego chomika: