This study focuses on the variability of chromosomal location and number of ribosomal DNA (rDNA) sites in some diploid and autotetraploid Festuca pratensis and Lolium perenne cultivars, as well as on identification of rDNA-bearing chromosomes in their triploid and tetraploid F. pratensis × L. perenne hybrids. The rDNA loci were mapped using fluorescence in situ hybridization (FISH) with 5S and 25S rDNA probes, and the origin of parental genomes was verified by genomic in situ hybridization (GISH) with L. perenne genomic DNA as a probe, and F. pratensis genomic DNA as a block. FISH detected variation in the number and chromosomal location of both 5S and 45S rDNA sites. In F. pratensis mostly additional signals of 5S rDNA loci occurred, as compared with standard F. pratensis karyotypes. Losses of 45S rDNA loci were more frequent in L. perenne cultivars and intergeneric hybrids. Comparison of the F. pratensis and L. perenne genomes approved a higher number of rDNA sites as well as variation in chromosomal rDNA location in L. perenne. A greater instability of F. pratensis-genome-like and L. perenne-genome-like chromosomes in tetraploid hybrids was revealed, indicating gains and losses of rDNA loci, respectively. Our data indicate that the rDNA loci physically mapped on chromosomes 2 and 3 in F. pratensis and on chromosome 3 in L. perenne are useful markers for these chromosomes in intergeneric Festuca × Lolium hybrids.

Genomic in situ hybridization (GISH) was used to make a detailed study of chromosome pairing at metaphase I (MI) of meiosis in six F(1) hybrid plants of the allotetraploid Festuca pratensis x Lolium perenne (2n = 4x = 28; genomic constitution FpFpLpLp). The mean chromosome configurations for all hybrids analysed were 1.13 univalents + 11.51 bivalents + 0.32 trivalents + 0.72 quadrivalents, and the mean chiasma frequency was 21.96 per cell. GISH showed that pairing was predominantly intragenomic, with mean numbers of L. perenne (Lp/Lp) and F. pratensis (Fp/Fp) bivalents being virtually equal at 5.41 and 5.48 per cell, respectively. Intergenomic pairing between Lolium and Festuca chromosomes was observed in 33.3% of Lp/Fp bivalents (0.62 per cell), in 79.7% of trivalents - Lp/Lp/Fp and Lp/Fp/Fp (0.25 per cell), and in 98.4% of quadrivalents - Lp/Lp/Fp/Fp and Lp/Lp/Lp/Fp (0.71 per cell). About 4.0% of the total chromosome complement analysed remained as univalents, an average 0.68 Lp and 0.45 Fp univalents per cell. It is evident that in these hybrids there is opportunity for recombination to take place between the two component genomes, albeit at a low level, and this is discussed in the context of compromising the stability of Festulolium hybrid cultivars and accounting for the drift in the balance of the genomes over generations. We speculate that genotypic differences between hybrids could permit selection for pairing control, and that preferences for homologous versus homoeologous centromeres in their spindle attachments and movement to the poles at anaphase I could form the basis of a mechanism underlying genome drift.

The development of grass cultivars derived from Festulolium hybrids is mostly limited because of their genetic instability and low fertility. The objective of this was to assess the genomic structure and fertility in two advanced breeding populations derived from an interspecific hybrid between Festuca pratensis Huds. (2n = 4· = 28) and Lolium perenne L. (2n = 4· = 28). Examination based on genomic in situ hybridization analyses of randomly chosen plants in the F 7 and F 8 generations, as well as those from earlier generations, indicated progressive changes in genome balance towards that of Lolium. The ÔdominanceÕ of Lolium chromatin over Festuca likely resulted from extensive recombination between chromosomes of the parental genomes, together with substitutions of whole Festuca chromosomes by whole Lolium chromosomes. The number of recombinant chromosomes and recombination breakpoints per genotype increased in successive generations, but their number was higher for Festuca than for Lolium. Because the mean seed set was only 38.2% in the F 8 generation, it was concluded that the F. pratensis · L. perenne amphiploids cannot be included successfully in a Festulolium breeding programme.