Preliminary DNA fingerprinting of the turf grass Cynodon dactylon ( Poaceae : Chloridoideae )

Identification o f different cultivars o f turf grasses is often very difficult. In a preliminary attempt to identify different cultivars o f Cynodon dactylon (L.) Pers.. random amplified polymorphic DNA (RAPD) analyses o f some well-known cul­ tivars used in South Africa, i.e. Bayview. Cape Royal, Florida. Hamsmith. Silverton Blue. Skaapplaas and Titdwart. as well as 10 potential new cultivars, were done. These results were used to determine the genetic distances among cultivars. Only five primers w ere needed to obtain a specific fragment pattern for each cultivar. The degree o f amplification w as used as an additional criterion by including all visible fragments, excluding very faint fragments and only including the brightest frag­ ments. The neighbour-joining trees o f C. dactylon showed best resolution from the data set w ith all visible fragments includ­ ed. although fragment intensity did not affect the tree topology. The cultivars Silverton Blue and Bayview exhibited the greatest genetic variation and two potential new cultivars were identified. RAPD analyses can. therefore, be used to distin­ guish between different C. dactylon cultivars and to determine the genetic variation between them by calculating genetic distances.


INTRODUCTION
The genus Cynodon Rich, comprises (Gibbs Russell et al. 1990).These species are morphologically very simi lar.Cynodon bradleyi, C. dactylon and C. transvaalensis are cultivated as turf grasses.Often potential new culti vars are introduced, but are these really new cultivars or are some just variable morphological forms o f existing cultivars?
Amplification conditions for RAPD analysis are simi lar to those used in a normal polymerase chain reaction, except that only one primer is used instead o f two primers with specific sequences (Williams et al. 1990).As a result, amplification in RAPD analysis occurs everyw here in a genome, where it contains two comple mentary sequences to the primer that are within the length-limits o f the polymerase chain reaction (PCR).which is ± 3 kb.The PCR patterns obtained from RAPDs are dependent on both the template and the specific PCR primer.Yu et al. (1993) observ ed the fragment size range to be from 0.5 to 2.5 kb and the fragment numbers from 1-10.
Polymorphisms detected by the RAPD technique are inherited as dominant markers in a Mendelian fashion and can be generated in any species without prior DNA sequence information (Williams et al. 1990: Welsh et al. 1991).Marsan et al. (1993) showed that DNA fragments, from inbred maize lines, were always present in one or both o f the respective parental lines, thus suggesting that RAPD fragments were stably transmitted from genera tion to generation.
A general characteristic o f the RAPD profile is the difference in fragment intensities.These differences in fragment intensities were therefore, also used as criteria in determining genetic variation within and between known cultivars and unknown specimens.
The aim o f this preliminary study is to use DNA pro files generated by the RAPD method to identify various known Cynodon cultivars from vegetative material, and to identify potential new cultivars.by comparing them with some well-known cultivars currently used in the industry.A further purpose is to use the RAPD data to calculate the genetic distances between the different cul-tivars, thereby investigating the variation within and
The reproducibility o f the technique was tested by duplicating each reaction (44 reactions for 22 specimens per primer).This was done by performing amplifications on identical DNA samples in two different reactions.These findings confirmed that the fragment pattern for a particular combination o f primer and DNA was repro ducible tor replicates, both in and between experiments.
Between 5-10 jil o f the amplification product was mixed with gel loading buffer and separated on a 1% (m/v) agarose gel in TBE containing ethidium bromide (0.4 mg/ml).The gel was run in 0.3X TBE ( IX TBE = 0.089 M Tris-HCl, 0.089 M boric acid.0.002 M EDTA) or 0.5X TAE (IX TAE = 0.04 M Tris-HCl.1.142 ml acetic acid.0.001 M EDTA) buffer at 120 V for ± 60 minutes.DNA lambda molecular weight markers VI or X were included in each gel.The fragments were viewed under UV light and documented with a 35 mm photo graph.

Fragment and phylogenetic analysis
The fragments were manually scored for each primer as present ( 1) or absent (0), for all the cultivars studied.
Furthermore, the fragments were divided into three cate gories according to intensity o f the fragments: bright, medium and faint fragments.The data were classified in three different sets, namely (a) including all visible frag ments, (b) excluding faint fragments, and (c) using only bright fragments (Table 2).
The different fragment intensities observed with the amplification products were also scored, by comparing the fragments within a specific specimen, for each primer.This was done due to different specimens am pli fying at different intensities.
The data were analysed with PAUP* (phylogenetic analysis using parsimony *and other methods) 4.0b8a (Swofford 1998).Cluster analysis was performed bv using the neighbour-joining method (NJ) as implement ed in this softw are and neighbour-joining trees were con structed.using total character difference as distance m ea sure.Cape Royal w as used as the outgroup in this study, being the morphologically distinct cultivar.Bootstrap values were calculated from 500 replicates (Felsenstein 1985) with resampling o f all 96 characters.Jackknife values were calculated from 500 replicates, with 50% deletion and the emulate Jac resampling option in effect (Lanvon 1985).

RESULTS AND DISCUSSION
The 22 specimens o f C dactylon came from seven known cultivars.duplicate specimens (collected from different localities) o f five o f these cultivars and ten potential new cultivars (Table 1).Table 2 contains infor mation on the total number o f fragments scored, the per centage o f fragments that showed no replication, the per centage o f faint, medium and bright fragments, the mini mum and maximum number o f fragments per specimen, and the range o f fragment sizes.
For primer OPA11, Cape Royal 2 specimen and for primer OPA16, Cape Royal 1 and Silverton Blue 1, the duplicated reactions failed, as a result o f total PCR fail ure (and thus not failure o f repeatability).This informa tion was not used in the calculations o f the percentage of fragments that showed no repetition (Table 2).
The percentage o f fragments, which showed no repe tition in the duplicates o f a reaction, varied from 1.33% in OPB03 to 2.82% in OPA16, with an average o f 2.03% (Table 2).This indicates that fragment reproducibility was high with all the primers used.OPBÓ3 being the most reproducible.Most o f the fragments that showed no repetition were o f faint intensity.
A series o f tests, done on different DNA extractions from the same plant and different amplification o f the same DNA sample, indicated that RAPD results are reli able.Well-amplified regions corresponded in all repeats from the same sample.The only differences observed were in faint fragments found in certain repeats.
A general characteristic o f the RAPD profile is the difference in fragment intensities.Many speculations for the reason o f this phenomenon have been given.One explanation is that the difference may be linked to the degree o f homology between primer and template DNA (Thormann et al. 1994).Caetano-Anollés et al. (1991) speculated that it might be the result o f amplification of multiple copies in the genome.
All five primers exhibited differences in the duplicate specim ens o f the cultivars Cape Royal, Tifdwarf, Florida.Bayview' and Silverton Blue, which ranged from faint to bright fragment differences.These results indi cate varying degrees o f variability w ithin these cultivars, especially Silverton Blue and Bayview.One fragment was consistent throughout all the specimens for two primers, namely a ± 700 bp fragment with primer OPA-16 and a ± 570 bp fragment with primer OPB-06.All the primers exhibited a few other prominent fragments in most o f the specimens.Very few unique cultivar-specific fragments were found, which could be linked to the small sample size.
For the neighbour-joining analysis, the three different data sets for Cynodon (according to fragment intensity) were used separately.Though their intensities differ, the three data sets gave neighbour-joining trees with the same topology.The resolution decreased with fewer parameters (number o f fragments), therefore, the neigh bour-joining tree using all visible fragments was the best resolved and will be discussed further (Figure 1): SAG.01 groups with the Silverton Blue clade.with rela tively high bootstrap and jackknife support.It is proba bly not a new cultivar.but a morphological variant o f this cultivar.SAG. 13 and Harrismith seem to follow the same pattern.There is, however, no substantial support for this grouping.O f the other potential new cultivars.SAG.02-SAG.05form a monophyletic clade and SAG.06 + S A G .09-SA G .il form another monophyletic cluster.When comparing the fingerprinting patterns for the dif ferent specimens, the close affinities between specimens SAG.02, SAG.03, SAG.04 and SAG.05 (Figure 1; Table 2) were also evident.The groupings SA G .02-SA G .04 and SAG.09 + SAG.11 probably represent two potential new cultivars, with the variation within the clades being so small as to indicate that the specimens in each cluster are probably the same cultivar.These close affinities are corroborated by the bootstrap and jackknife support val ues within these groups, which are 100% and 80-85% respectively.The other members o f these clades, SAG.05 and SAG. 10 + SAG.06, are probably closely related vari ant forms.The distances within some o f the existing cul tivars are very large, which indicate large levels o f genet ic variation within these cultivars.This is especially true for the two cultivars Bayview and Silverton Blue.This is corroborated by the variability in fragment patterning observed in the different specimens for these cultivars.These results, in which these two cultivars are clearly non-monophyletic, might indicate that either the taxono my o f these cultivars are confused or the samples are in fact not purebred cultivars any more.The last possibility is very feasible in this group of grasses that constitute a heterogeneous group o f varieties with considerable geno typic as well as phenotypic variation and in which outcrossing is frequent.The other duplicated cultivars form distinct groupings with high support bootstrap values for Cape Royal and Tifdwarf (97% and 92% respectively).Florida also exhibits some variability but not to the same extent as Silverton Blue and Bayview.
The number o f specimens studied per cultivar was only two, due to the preliminary nature o f the study.By increasing this number, the variability within cultivars can be investigated in more detail.
Very small genetic differences can be detected with RAPDs.In some cases these differences may include only a single DNA change.A single difference in the fragmenting patterns o f different specimens does, there fore, not indicate separate cultivar status.It was.howev er.possible to distinguish between the different Cynodon cultivars with the RAPD fingerprinting patterns.More primers included and more samples per cultivar, would help to further resolve relationships, especially where the status o f a cultivar is uncertain.
Although the reproducibility o f this RAPD technique can be influenced by factors that may vary, such as tem plate quantity and primer structure (Kernodle et al. 1993;Multani & Lyon 1995), the use of a standardized RAPD protocol and sufficient replication can ensure repro ducible RAPD patterns (M ultani & Lyon 1995).Furthermore, all reactions were always amplified simul taneously, and found to be repeatable across different amplification times.
These markers have the potential to be employed as genetic fingerprints for future identification.

CONCLUSIONS
This study indicated that different Cynodon cultivars differ genetically, and these variations can be determined by RAPDs.
The only two specimens with a similar fragmenting pattern, irrespective o f the primer used, were SAG.03 and SAG.04.However, these specimens show' similar patterns to SAG.02 and SAG.05 with most primers.This indicates that these four specimens are genetically very similar and could well be the same cultivar.This was reflected by the neighbour-joining analysis where SA G .02-SA G .05 and SAG.06 + S A G .09-SA G .il form definite m onophyletic groups with the clusters SAG.02-SA G .04 and SAG.09 + SAG.l 1. which appear to be new cultivars.This is supported by bootstrap and jackknife values and very little variance within these clusters.SAG. 13 appears to be related to the Harrismith cultivar and SAG.01 to the Silverton Blue cultivar.
Furthermore, the variability within existing cultivars was very high in some instances, questioning their status as true cultivars.Due to the variable nature o f the species it is very difficult to recognize the different cultivars of these turf grasses vegetatively, especially when they are frequently cut on lawns, bowling greens or golfing greens.This complicates the unequivocal identification o f these cultivars.

TABLE 1 .-Cultivar names and voucher numbers o f Cynodon dacty- between the studied specimens. ion specimens MATERIALS AND METHODS Plant material
Table 1) are housed in the Geo Potts H erbarium , U niversity o f the Free State, B loem fontein (BLFU).The cultivars used were Bayview, Cape Royal, Florida, Harrismith.Silverton Blue, Skaapplaas and Tifdwarf, as well as 10 potential new cultivars (SA G .01-06 & 09-13).

TABLE 2 .-Comparison between five primers used with respect to no. o f fragments observed, repeatability, fragment intensity and range of fragments
.FIGURE 1