The genus Lolium : taxonomy and genetic resources

Several aspects of variation within the genus <em>Lolium,</em> and more in detail within <em>Lolium perenne</em> (perennial ryegrass) have been highlighted. As the results are extensively discussed in each chapter, the general discussion is focused on two aspects of the research.<u>Speciation</u>It is clear that the genus <em>Lolium is</em> a very variable genus. The variation within the species reduces the clarity of separation of the species. Stebbins (1956) found the differences between <em>Lolium</em> and <em>Festuca</em> not sufficient to justify two separate genera. He also states that the family of <em>Poaceae is</em> a phylogenetically derived family, and therefore of comparatively recent origin. Differences between species and between genera are still developing. In older families, intermediate forms or species have become extinct, therefore genera and species delimitations are clearer in these families, e.g. the <em>Papaveraceae</em> (Stebbins, 1956). Producing additional information besides morphological data, such as cytogenetic studies and biochemical studies, is of little use for these families as genera and species delimitations are easily made with simple morphological characters. In families of more recent origin, such as the <em>Poaceae,</em> as many characteristics as possible should be used to establish the relationships between the species. In Chapters 2, 3 and 4 several types of characters have been analyzed for the genus <em>Lolium. No</em> phylogenetic analysis was performed with these data, only phenetic analysis due to the nature of the characters used. Results indicate that all species, as mentioned in the general introduction, can be recognized, although species delimitations are not unambiguous. Only for <em>L. persicum</em> and <em>L. temulentum</em> the results indicate that these species could possibly be two varieties of one species. All species show diagnostic characters for one ore more of the different type of characters. Although the overlap of species is often unsatisfactory, joining of species would be even more artificial. As speciation is a continuous process it cannot be predicted at which point in time species are going to be sufficiently delimited.Man has had large influence on the speciation within <em>Lolium.</em> This is illustrated by the three weedy species within the genus. <em>L. remotum is</em> known as a weed in flax (Hjelmqvist, 1950), <em>L. temulentum</em> and <em>L. persicum</em> are known as weeds in cereals (e.g. Dore, 1950). All three are mimicry weeds, the morphology of the seeds andlor the habit of the plant is similar to the crop in which it grows. Until a few decades back, these three species had a significant impact as weeds, but due to enhanced seed cleaning techniques the distribution area of these species has largely decreased (Hubbard, 1954).Other examples of the influence of man on the genus <em>Lolium,</em> are the species <em>L. perenne</em> and <em>L. multiflorum.</em> Their distribution area has largely increased due to sowing by man. Scholz (1975) stated that man has had an enormous influence on the development of both species. According to Scholz (1975), this influence started no more than a few thousand years back, with the cutting and burning of forest for replacement by grassland for cattle, and the discovery of hay making. This has made it almost impossible to determine in which parts of the world both species are indigenous. Not only the distribution area of both species is influenced by man but also the phenotype. Selection changes the phenotype in favour of character states desired by man, such as increased yield. Tyler (1979) observed that after a period in which the standard of management is relaxed, natural phenotypes reoccur. This is confirmed with the results from Chapter 5: Dutch perennial ryegrass populations, collected after a period of more relaxed management, have a distinct phenotype compared to cultivars. Tyler (1979) also indicated that the differences between wild and cultivated forms are extremely blurred for <em>L. perenne.</em> This statement is confirmed by the results from Chapter 6: for allozymic variation, cultivars show absolutely no reduction in variation compared to natural populations. Ellenberg (1963) calls the type of plant as <em>L. perenne</em> semi-domesticated, as the crop is not harvested each year but only kept at an acceptable production level using reseeding. During each phase of their lifecycle, populations are exposed to selection forces. Leading to the situation that in grasslands cultivars are often mixed with plants that have been exposed to enviromnental selection for a number of years. This makes the distinction between natural and cultivated grassland extremely vague.Chapter 5 illustrates, as management is the factor that optimizes the amount of genetic variation found within a location, the enormous influence man has had and still has on the amount of variation in phenotypes of <em>L. perenne.</em> Reduction of the influence of man would probably lead to the existence few differing perennial ryegrass phenotypes, and could in some areas even mean extinction of perennial ryegrass. In the Netherlands, foreland and salt marshes are the only original habitats for grassland (Bink et al., 1984). Although <em>L. perenne is</em> a species with much competitive ability, it would suffer from a large reduction of distribution area in case management of grasslands was totally abandoned. Because mainly under man-made conditions, e.g. fertilizing, treading, intensive grazing, drainage, <em>L. perenne</em> expresses this competitive nature.For <em>L. rigidum</em> the influence of man is less strong. <em>L. rigidum is</em> used in some parts of the world (e.g. Australia) as a cultivated fodder crop, but in Europe this is not current. In Europe the fate of <em>L. rigidum</em> depends on the perspectives of <em>L. rigidum</em> as a fodder crop in dry areas or as a crossing parent in breeding programmes. Next to its presence in cultivation <em>L. rigidum is</em> well capable to maintain itself under less cultivated circumstances, this in contrast to, especially, <em>L. perenne.</em> Hartley (1956) states that <em>L. rigidum</em> originates from the Mediterranean region and that <em>L. perenne</em> and <em>L. multiflorum</em> originate from the Eurasian region. The ancestral species of the genus <em>Lolium is</em> supposed to have originated in the Mediterranean region (Malik, 1967). This would indicate that <em>L. rigidum</em> could be the wild form for both cultivated species. The relation between wild and cultivated is often confirmed by the reduction of genetic variation within the cultivated forms. Brown (1978) mentions two examples, based on allozyme variation, for which this assumption is valid. <em>Lycopersicon pimpinellifolium</em> has 61 % unique allelic forms compared to those in <em>L. esculentum.</em> Both species share 37% of the allelic variants and 2% is unique for <em>L. esculentum. Oryza perennis</em> has 47% unique peroxidase alleles, and 22% unique esterase alleles, compared to 0. <em>sativa.</em> Both species share 53 % and 78 % of the alleles respectively. The results from Chapter 3 do not indicate a reduction in allelic variation within <em>L. perenne</em> and <em>L. multiflorum,</em> compared to <em>L. rigidum.</em> This indicates that, if <em>L. perenne</em> and <em>L. multiflorum</em> indeed did arise from <em>L. rigidum,</em> this speciation is of recent origin. Phylogenetic relations between species cannot be determined on basis of these data.<em>L. loliaceum is</em> not known as a weed nor as a crop plant, it mainly grows under poor and maritime conditions. The influence of man on populations of this species is not large, therefore it is not likely that this species becomes extinct nor that its distribution area will suddenly increase. Phenotypic developments are expected to be gradual and slow.For <em>L. canariense</em> the same holds true as for <em>L. loliaceum,</em> it is not a crop nor a weed and grows under poor conditions, making it a stable and localized species.The screening of the <em>Lolium</em> species for allozymic variation, added little to the species determination within the genus <em>Lolium.</em> The pattern of allozyme diversity could hardly be linked with taxonomic classification (Chapter 3); mainly because all allelic variants were common in each population screened. As pointed out in the discussion of Chapter 3, this maybe caused by the small number of enzyme systems screened. A question that can be asked is whether increase of the number of allozymes could lead to better results for genotypic screening. In Chapters 3 and 6 the calculated diversity statistics for the cross-breeding <em>Lolium</em> species were above the average for other wind- pollinated cross-breeding species (Hamrick & Godt, 1990). These statistics indicated that a larger proportion than average of the loci screened were polymorphic, and also that the average heterozygosity of the loci was far above the mean. Extension of the number of loci screened would therefore most likely result in finding monomorphic loci or less variable polymorphic loci and would not enhance the results. In literature, analyses of <em>L. perenne</em> populations for several other allozymes are reported. These allozymes are Glutamate- oxaloacetate-transaminase (GOT, Hayward & McAdam, 1977; Arcioni et al., 1988; Charmet et al., 1993), Isocitrate dehydrogenase (IDH: Lallemand et al., 1991; Charmet et al., 1993), Peroxidase (PRX: Charmet et al., 1993) and Superoxide dismutase (SOD: Charmet et al., 1993). All authors report results that confirm the expectation that higher number of allozymes screened do not improve the elucidation of speciation. Again, the within- population variation is too large compared to the betweenpopulation variation.Another option would be to make use of molecular markers, e.g. restriction fragment length polymorphism (RFLP). Few reports are known for <em>Lolium</em> species, using these techniques. Darbyshire & Warwick (1992) report on the results for one <em>L. perenne</em> population, which was compared with several other grass populations classified in 26 <em>Festuca</em> species and the genera <em>Vulpia, Poa</em> and <em>Puccinella.</em> Eleven restriction endonucleases and twelve restriction fragments from chloroplast DNA of <em>Petunia hybrida</em> Vilm. were used in this analysis. In total 341 <em></em> bands were observed of which 108 (31.7%) <em></em> were polymorphic. Of these 108 bands, 34 <em></em> were detected in the <em>L. perenne</em> population. Only one plant was analyzed from each population. Chloroplast DNA variation in other <em>Lolium</em> species (Lehväslaiho et al., 1987 <em>;</em> Soreng et al., 1990) is <em></em> only reported for one <em>L. multiflorum</em> population, using five restrictionenzymes and direct end labelling. Again only one plant has been analyzed and compared with a large set of populations and genera mainly from the family <em>Poaceae. L. multiflorum</em> differs in 11 bands on a total of 144 shared bands with <em>Festuca pratensis.</em> Only one report is known (Wu et al., 1992) <em></em> on the between-population variation within <em>L. perenne</em> for RFLP's. Five cultivars of perennial ryegrass were screened, using 2 <em></em> restriction enzymes and 37 <em></em> probes from <em>Festuca pratensis.</em> Twenty-four (65 %) <em></em> of these probes hybridized, resulting in on average 69% <em></em> polymorphism between the five cultivars . On average 3.2 different banding patterns were observed for each restriction enzyme-probe combination. Again only one plant was analyzed for each population.No reports on the between-species and the within-population variation are known for any of the <em>Lolium</em> species.The results from Chapter 3 and Chapter 6 <em></em> indicate that substantial variation is found within populations of the cross-breeding <em>Lolium</em> species, which makes results based on only one plant per population unreliable (Wu et al, 1992). <em></em> It remains necessary to analyse a minimum number of plants for the crossbreeding <em>Lolium</em> species, unless an acceptable bulk sample can be taken. This would be desirable as otherwise the cost and time needed to analyse a population using molecular markers could be limiting. The danger of using a bulk sample would be that no differences between populations and even between species can be observed (as would be the case for a bulk sample when screening for allozyme variation). Screening of five enzyme systems resulted in a maximum of 10 bands observed <em>(</em> 13 <em>,</em> if the heterozygous bands were also counted), in case a plant was heterozygous (maximum variation) for each enzyme. This is a much better result as reported by Darbyshire & Warwick (1992), 34 <em></em> bands out of 132 <em></em> restriction enzyme-probe combinations. It is equal to the theoretical maximum number of bands reported by Wu et al. (1992), 96 <em></em> bands in case of heterozygosity at all 48 restriction enzyme-probe combinations. The preliminary conclusion would therefore be that RFLP analysis will not greatly enhance the distinction of crossbreeding <em>Lolium</em> species and populations.For the inbreeding <em>Lolium</em> species the analysis of few plants is sufficient. The observed variation would probably increase compared to the observed allozyme diversity (Chapter <em>3:</em> fixation for four of the five enzymes), as the number of possible markers would greatly increase using RFLP's. The use of molecular markers for the screening of inbreeding <em>Lolium</em> populations would therefore be a valuable extension of the knowledge on these species.<u>Genetic resources: <em>in situ</em> conservation</u>In the general introduction three research questions were mentioned, concerning the genetic resources of <em>L. perenne.</em> What are the answers to these questions after analyzing the results from four years of research? Firstly, the Dutch populations do form, morphologically, a distinguishable group of forms within the genetic variation for perennial ryegrass. This conclusion is based on the observation of morphological characters only, as the heritability of these characters is better determined than for agronomically important traits as winter hardiness, spring growth etc. The date of ear emergence is one of the most important characters, both agronomically and for the recognition of breeders rights. Results indicate that genetic variation for this character is substantial in the Netherlands: in natural surroundings, populations varying from very early till very late heading could be collected. It is expected that if the Dutch populations show this amount of genetic variation for morphological characters, results can be analogous for agronomical characters. The morphological variation found within the Dutch populations is not comparable with the variation found in the cultivars used in these trials. Although date of ear emergence indicates that some Dutch populations are heading as late as the cultivars, morphologically they are distinct. The Dutch populations are a.o. more prostrate growing and shorter at ear emergence, this could indicate that this phenotype is natural for <em>L. perenne</em> in the Netherlands. The fact that Dutch populations are morphologically clearly distinct from the cultivars and that there was also substantial variation observed between Dutch populations, indicates that <em>in situ</em> conservation is a realistic option for <em>L. perenne.</em> Weibull (1989) gives several advantages and disadvantages for the <em>in situ</em> conservation of forages. The advantages are: continued co-evolution of populations and the possibility to study the ecology of the species. It is also possible to make successive collections, and it avoids space and time consuming activities for storage en regeneration for genebanks. A combination of the genetic resources conservation objective with other objectives like nature conservation could be an option. This possibility is clearly illustrated with the present results, as all Dutch populations were collected in areas managed by nature conservation organisations.Disadvantages are that it is difficult to determine how many sites, and which sites should be preserved to optimize genetic variation. Natural populations are vulnerable to external factors, such as human influences and extreme weather conditions. Also the costs of the maintenance of conservation sites maybe high, and access of breeders can be a problem in case of a combination with nature conservation objectives.For the allozyme variation no differences between the Dutch populations and the cultivars were found. Allelic variants were very common in all populations, the cultivars showed much larger differences in allelic frequencies than the Dutch populations. <em>In situ</em> conservation would be very successful in retaining genetic diversity at the allozyme level. The data were not useful for selection of accessions for genebanks. Phillips et al. (1993) reported for <em>Avena sterilis</em> L. (inbreeder), the wild progenitor of <em>A. sativa</em> L., the possibility to separate populations in six different groups based on 23 loci. Selection of genebank accessions can be facilitated using these six groups, combined with morphological data.Francisco-Ortega et al. (1992) observed for <em>Chamaecytisus proliferus</em> (L. fil.) Link a totally different pattern. Morphologically this species can be separated into seven subspecies, which are morphologically distinct and ecologically each occupy a distinct niche. Allozyme diversity shows no differentiation between these seven subspecies.Just like in the genus Lolium, almost all allelic variants are common and widespread, and the within-population variation is very large. Also in this case allozyme data were considered of no use for the selection of genebank accessions.Generally, the usefulness of screening for allozyme variation varies substantially. Compatibility behaviour and age of the genus/species are the major factors, explaining the value of this kind of data.