Introduction

Lolium perenne L. (perennial ryegrass) is an outcrossing, wind-pollinated species. Selfing is largely prevented by a gametophytic, two-locus incompatibility system (SZ) acting at the stigmatic surface (Cornish et al, 1979). The system was first elucidated in rye (Secale cereale L.) (Lundqvist, 1954). All eight grass species in which the incompatibility system has been examined in detail (see Baumann et al, 2000) exhibit the same two locus system in which the two loci are complementary in action. Both S and Z alleles must be matched in pollen and style for an incompatible reaction to occur. Both S and Z loci are highly polyallelic as shown by estimates of allele numbers in a synthetic cultivar (Devey et al, 1994) and a native population (Fearon et al, 1994).

The S and Z loci have been mapped in rye (Wricke and Wehling, 1985, Gertz and Wricke, 1989; Fuong et al, 1993; Phillip et al, 1994; Voylokov et al, 1994, 1998). Li et al (1994) reported the cloning of a putative S gene with a thioredoxin like sequence in the grass species, Phalaris coerulescens. This finding was later revised as being a gene linked to S, mapping about 2 cM away (Langridge et al, 1999). For L. perenne the only known linkages are between the S locus and the isozyme locus PGI/2 (Cornish et al, 1980) and there is also evidence that the Z locus is linked to the isozyme locus GOT/3 (Thorogood and Hayward, 1992).

This paper reports the genotyping, based on the functioning of a two locus, gametopytic incompatibility system, and subsequent mapping of the incompatibility loci with a family of 139 plants extensively mapped with RFLP and isozyme markers (Jones et al, 2002).

Materials and methods

Plant material

The L. perenne mapping family P150/112 is the core mapping family for the International Lolium Genome Initiative (ILGI). It was formed from an anther culture-derived doubled haploid plant (female) crossed with an unrelated heterozygous plant donated from Dr Pete Wilkins’ forage breeding programme (IGER) (Jones et al, 2002). One hundred and thirty-nine individuals from this family were genotyped for S and Z. The incompatibility genotype of the female parent was designated S11Z11. As the male parent was unrelated, and knowing that the S and Z loci are highly polyallelic (Devey et al, 1994; Fearon et al, 1994), it was assumed that it was heterozygous at both loci for two further alleles and was assigned the genotype S23Z23.

Plant pollination and incompatibility genotyping strategy

According to the two locus model of self-incompatibility, the F1 family derived from this cross segregates into a maximum of four possible incompatibility genotypes (S12Z12, S12Z13, S13Z12, S13Z13). Inter-pollination of these genotypes results in the incompatibility reactions shown in Table 1. There are no reciprocal differences in reaction type. There are also no fully compatible reactions as all plants produce 25% S1Z1 gametes that never effect successful pollination. At first genotypes needed to be identified that could be used to pollinate each plant of the mapping family. In theory, two pollinator genotypes crossed to the complete set of plants in the mapping family could provide enough information to assign genotypes to every plant in four out of the six possible pair-wise combinations of pollinator genotypes. In the two cases where pollinator genotypes have no S or Z allele pairs in common (ie, S12Z12 and S13Z13 or S12Z13 and S13Z12), it would be impossible to distinguish between the two stylar genotypes, which are not the same as the pollinator genotypes (ie, S12Z13 and S13Z12 or S12Z12 and Z13Z13) as both would produce a 50%-compatible reaction with both pollinators. If three different genotypes are used as pollinators it is possible to determine which of the four stylar genotypes each of the plants possessed: if the stylar genotype was the same as one of the pollinators an incompatible reaction would occur and the other two reactions would be partially compatible. If all pollinators give a partial reaction with the stylar plant, that plant must belong to the genotype not present in the set of pollinators. It is, however, much more preferable to have at least one representative of each of the four genotypes to use as pollinators. We are then certain to obtain a fully incompatible reaction where the pollinator and stylar parent have the same SZ genotype. The incompatible reaction is then confirmed by the 75% compatible reaction of the same stylar plant with the pollinator genotype with no allele pairs in common, and the 50% compatible reaction with the two genotypes that have either the S or the Z allele pair in common.

Table 1 Compatibility reactions obtained when inter-crossing the four SZ genotypes expected in the F1 population
Full size table

In our pollination scheme we initially chose six plants at random which we crossed in diallel fashion. Choosing six plants at random gave us a 96.9% chance of obtaining at least two pollinators with appropriate genotypes that would enable us to classify all stylar plants pollinated with these two plants (ie, two pollinator genotypes with either S or Z allele pairs in common). Further informative pollinations were determined by the results of this initial diallel.

Pollination methods

Plants were grown in 15-cm diameter pots in Humax John Innes No 3 with wetting agent. Plants were vernalised (short days, low temperature) naturally in an unheated, unlit glasshouse over winter in readiness for pollinations to be made the following spring. Pollination methods were adapted from those of Cornish (1979). Plants to be used as pollinators were chosen, which possessed inflorescences that were about to anthese or had been anthesing for a few days. Plants were still capable of producing pollen 2 or more weeks after the first inflorescences had started anthesis. These plants were placed in a growth room running at 23°C, with a light level at plant height of approximately 300 μmol/m2/s, on their sides with flowering stems enclosed in a cellophane bag. The plants anthesed around midday. Mature ovaries with stigmas attached were collected from basal florets of spikelets that were close to anthesis. Spikelets were removed from the plants and ovaries were harvested at some distance from flowering plants to avoid contamination from stray pollen. Ovaries were placed on 50-mm diameter petri-dishes plated with agar (2.5% Agar agar, 25% sucrose and 25 ppm Boric acid) which provided a semi-solid medium on which the ovaries could be supported allowing the feathery stigmas to stand proud of the Agar surface in readiness for pollen reception (Lundqvist, 1961). Two ovaries from each of up to 16 stylar plants were placed on a single petri dish. Pollen, which collected in the cellophane bags, was shaken onto the surface of the agar, ensuring that only free-flowing, non-clumped pollen was used. Anthers in the bag were removed prior to pollination to ensure smooth pollen flow onto the plates.

A minimum of 16 h post-pollination, styles and stigmas were removed from the ovaries with a scalpel on microscope slides. Stigmas were temporarily mounted under a coverslip in a drop of decolourised Aniline Blue (0.1% water-soluble aniline blue in 0.1 M K3PO4) (Martin, 1959) in readiness for microscope examination. Pollen-tube growth was observed microscopically at low power (×10) under UV fluorescent light. Incompatible grains produced pollen tubes that were soon arrested at or near the stigma surface and became occluded with callose that fluoresced brightly under UV light. There was also considerable build-up of callose within the grain itself. Compatible grains, on the other hand, were able to germinate successfully, tubes often reaching the base of the style. Sometimes these tubes fluoresced brightly but at other times were relatively difficult to see.

Pollinations were scored by estimating the proportions of compatible and incompatible grains. The fully incompatible pollinations were easiest to score. Mostly it was possible to distinguish 50% and 75% compatible reactions but at other times it was only possible to ascribe a partially compatible reaction where the number of grains was small or large numbers of inviable non-fluorescent, non-germinating grains obscured observations of many of the viable grains. Such problems with scoring were also encountered by Cornish (1979).

Molecular marker generation and linkage analysis

RFLP and isozyme segregation data were collected on the mapping family as described by Jones et al (2002). The S and Z loci were assigned to linkage groups using Joinmapâ„¢ v. 2.0. (Stam and Van Ooijen, 1995) using the linkage group numbering system established by Jones et al (2002).

Results

Incompatibility genotyping

Pollinations between our six randomly selected plants were classified as in Table 2. Plant 31 was arbitrarily assigned the genotype S12Z12. From the table it can be seen that plants 169 and 181 gave the same pollination reactions and the three plants are intra-incompatible. Therefore these two plants must have had the same genotype as plant 31. Plant 59 had a unique pollination phenotype and it was 75% compatible with the three S12Z12 genotypes. This plant must therefore have had the genotype S13Z13. Plants 63 and 69 had the same reactions as each other and are intra-incompatible. They therefore must have had the same genotype as each other. They gave a 50% compatible reaction with both S12Z12 and S13Z13 genotypes and could therefore be assigned to either of the remaining genotypes. We assigned plants 63 and 69 arbitrarily to genotype S13Z12. The fourth genotype segregating in the family, S12Z13, was not found in this round of pollinations.

Table 2 6 × 6 diallel of pollinations of six randomly selected plants
Full size table

A second round of pollinations was made using the six pollinators genotyped above as pollen parents and a further 48 plants as stylar parents. The reactions obtained were of two types: either only one pollen genotype giving an incompatible reaction and the other two genotypes giving either a 50% or 75% compatible reaction: or all pollen genotypes giving partially compatible reactions, in accordance with expectations given in Table 1. Those plants, which gave partial reactions with all three genotypes so far ascribed, must belong to the remaining genotype, S12Z13.

A total of 54 plants had been genotyped: 11 S12Z12, 14 S13Z12, 13 S12Z13 and 16 S13Z13 genotypes were ascribed. Ultimately, all 139 plants in the family that were analysed for incompatibility reaction were pollinated with at least two plants of each of the four genotypes. It was possible to unequivocally assign an incompatibility genotype to each plant in the mapping family resulting in the genotyping of all 139 plants.

The number of plants of each S and Z genotype is given in Table 3. Segregation of the four genotypes was not significantly different from a 1:1:1:1 ratio (χ2 = 0.34, P[3df] > 0.95).

Table 3 Segregation of S and Z alleles for 139 plants of the mapping family
Full size table

Mapping the S and Z loci

The Z locus was assigned to linkage group 2 using a LOD threshold of 5 and mapped as indicated in Figure 1. A similar LOD threshold assigned the S locus to linkage groups 1 or 3, but failed to split group1 from group 3. The S locus could be associated only with linkage group 1 using a LOD threshold >13; it was subsequently mapped as indicated in Figure 1. The ‘linkage’ between these two groups was associated with significantly distorted segregation ratios for the majority of the markers on linkage group 3 (Jones et al, 2002) and LOD scores as high as 13.2 associating markers in the region of the final map position of the S locus on linkage group 1 and some of the markers on linkage group 3 (see Table 4). A more detailed analysis of the S and CDO920 allele frequencies (Table 5) revealed that the distorted segregation ratio of CDO920 alleles (χ2 = 14.7, P[2df] < 0.001) could be accounted for by the excessive number of plants with the S2/CDO920a alleles and the low number of plants with the S2/CDO920b alleles which resulted in a significant deviation from the expected 1:1 ratio (χ2 = 49.6, P[2df] < 0.001). The segregation ratio for plants with the S3/CDO920a and S3/CDO920b allele combinations did not depart significantly from a 1:1 (χ2 = 2.9, P[2df] > 0.05).

Figure 1
figure 1

Map positions of the S and Z loci and closely linked markers. (a) Linkage group 1, (b) Linkage group 2. Complete marker maps for these linkage groups are given by Jones et al (2002).

Full size image
Table 4 LOD scores for the two-point comparison of each marker on linkage group 1 (L1) with each marker on linkage group 3 (L3). LOD scores ≥10 are indicated in bold. The relative genetic distances (cM) of the markers on L1 and L3 and the number of genotypes scored (n) for each marker are indicated
Full size table
Table 5 Numbers of individuals recovered for the four genotype classes associated with the segregation of the S alleles (2 and 3) on linkage group 1 and the CDO920 alleles (a and b) on linkage group 3
Full size table

The Z locus had no such association with linkage group 3 or any other linkage group.

Discussion

The results from the pollinations in this mapping family confirm the two-locus control of self-incompatibility. Disturbance of expected S and Z allele ratios and linked markers has been suggested in Lolium (Hayward et al, 1998) and in rye (Philipp et al, 1994) based on the hypothesis that some S or Z alleles, or SZ allele combinations, are less effective than others at preventing self-pollination. However, in the present study, both male and female gametes that form progeny from the original fully compatible cross are equally likely to contain either of the two segregating S and Z alleles. There is no evidence in this family that any S or Z incompatibility allele is more or less likely to effect pollination. The most likely cause of segregation distortion of S, Z, or linked markers would be in crosses resulting in partially compatible pollination where some SZ allele combinations produce an incompatibility reaction and are excluded from effecting pollination; or in progeny of selfed plants which are heterozygous for a self-fertility mutation at one of the incompatibility loci or other loci as demonstrated by Thorogood (1991) in Lolium and Voylokov et al, (1994) in rye.

The position of S on linkage group 1 is endorsed by the observation of Cornish et al (1980) who found that S and PGI/2 are linked with a recombination frequency of 15.8%. Although PGI/2 did not segregate in the ILGI mapping family, a F2 L. perenne mapping family positions PGI/2 on linkage group 1 (Armstead, Turner and Humphreys, personal communication). In addition, the map positions of S and four common RFLP markers on linkage group 1 in L. perenne and rye indicate a high degree of conserved synteny for this region of the genome between the two species (Voylokov et al, 1998; Korzun et al, 2001) (Figure 2). The positions of markers flanking the S locus are not entirely co-linear but this may be as a result of sampling effects, rather than genuine chromosomal re-arrangements. The RFLP marker, CDO98, mapped within 0.6 cM of the S locus and is probably closer than the Bm2 gene isolated by Langridge et al (1999) which was estimated to lie up to 2-cM away from the S locus.

Figure 2
figure 2

Syntenic relationships between rye and ryegrass. (a) Voylokov et al, (1998) vs ILGI population (b) Korzun et al, (2001) vs ILGI population.

Full size image

Markers from the linkage groups of the mapping family used in this study have enabled alignment of the Lolium map with those of other members of the Poaceae family including rye and members of the Triticeae tribe (Jones et al, 2002). The Z locus was found to be on the linkage group 2 coinciding with the location of Z in rye (Fuong et al, 1993). This again indicates conserved synteny between the species though none of the markers in the ILGI map were used in the construction of the rye map.

The association of the S locus with some of the markers on linkage group 3 (Table 4) indicates that there is likely to be a gene(s) on linkage group 3 that interacts with the S locus (or a gene tightly linked to it). The result of this interaction is that pollen or plants formed after pollination which possess the S2 allele and allele(s) from one of the linkage phases on group 3 are virtually inviable, whereas pollen or plants possessing the S2 allele and allele(s) from the other linkage phase on group 3 are super-viable (ie, more progeny with these genotypes are recovered than expected). The inviable/super-viable balance results in the restoration of the expected 1:1 ratio for the segregation of the S alleles. The action of this locus had no effect on the classification of incompatibility genotypes according to the classic two locus, gametophytic incompatibility model and therefore must act downstream from the initial pollen-stigma surface interaction. Whether it acts gametophytically or effects viability of post fertilisation products is unknown. The isozyme locus, GOT/3, is positioned on linkage group 3 (see Table 4) and a previous study (Thorogood and Hayward, 1992) found an association between a self-fertility locus segregating in L. perenne/L. multiflorum × L. temulentum first generation back-crosses to L. perenne/L. multiflorum and the GOT/3 locus with a recombination frequency of 18%. Furthermore, progeny of selfed plants heterozygous for self-fertility (50% self-compatible) were found to reveal significantly skewed ratios at the GOT/3 locus (Thorogood, 1991). Controlled pollinations confirmed that the self-fertility locus was either S or Z (or a locus closely linked to either of these loci (Thorogood and Hayward, 1992). As it was already known that S was linked to the isozyme locus PGI/2 (Cornish et al, 1979), and as PGI/2 and GOT/3 were also known to segregate independently (Hayward and McAdam, unpublished) it was concluded that the self-fertility trait from L. temulentum was likely to be a mutation of the Z locus linked to GOT/3. In the current paper, GOT/3 maps to linkage group 3 and therefore is not linked to either S or Z. The significant co-segregation between self-fertility and GOT/3, and the skewed ratios of GOT/3 allele segregations observed on selfing plants that were heterozygous for self-fertility (Thorogood, 1991) was probably due to interaction between the S locus and the locus identified on linkage group 3 in the present study. However, the action of this locus on S or Z is in contrast to other loci, unlinked to S and Z, identified in L. perenne (Thorogood and Hayward, 1991), Phalaris coerulescens (Hayman and Richter, 1992) and rye (Voylokov et al, 1993) which in all cases override the S and Z incompatibility response to produce self-fertile plants. In rye, a self-fertility locus has been mapped to chromosome 5R (Fuong et al, 1993).

Commercial turf and forage ryegrass cultivars are produced, exploiting the self-incompatibility system, as synthetics made from polycrossing a number of selected mother plants. The production of synthetics is an effective way of exploiting heterosis yet, inevitably, selection for important traits can lead to homozygosity of regions of the genome and homogeneity of these regions across the breeding population. Clearly, if selection for traits that are linked to S and Z occurs, seed production ability of these populations could be seriously compromised.

Genome mapping in L. perenne is well enough advanced (Jones et al, 2002; Armstead, Turner and Humphreys, personal communication) to be used as a tool for identifying and mapping traits of interest to forage and turfgrass breeders.

Quantitative trait loci (QTL) have been identified for forage quality traits such as water-soluble carbohydrate (Turner et al, 2000), crown rust (Puccinia coronata Corda) resistance (Thorogood et al, 2001) and delayed leaf senescence (Thorogood et al, 1999). Molecular markers associated with QTL can be used to aid selection for such complex, difficult to measure traits. Identifying the map positions of S and Z now means that marker-assisted selection strategies can be devised to select for favourable combinations of incompatibility alleles and linked QTL. Ultimately, the aim will be to produce allele specific probes for both S and Z loci in order to monitor and maintain heterogeneity at these loci in any future marker-assisted selection programmes.