VARIABILITY OF MORPHOLOGICAL AND ANATOMICAL TRAITS IN NATURAL POPULATIONS OF Festuca rubra AND F . nigrescens

Morphological and anatomical variability in 24 subpopulations of two native species of Festuca occurring in Poland was studied. The principal component analysis and cluster analysis of morphometric data and DNA classified specimens or sub-populations of Festuca into two groups, corresponding to the two species. F. rubra and F. nigrescens are similar to each other in most morphological characters: the number of the florets, the length of the spikelet, the length of the lemma awn, the width and thickness of the leaf, the length of the cauline leaf, the length of hairs in the leaf, the number and diameter of vascular bundles in the leaf, the hairiness of branches of the panicle, the hairiness of the lemma. The key morphological features distinguishing both species include the number of ribs in the leaf, the width of the cauline leaf (second from the top), caespitose, presence of rhizomes and quantity of the sclerenchymatous tissue in the leaf blade. Apparent differences were detected between the single sub-populations. However, between the intraspecific sub-populations, diversity in terms of the following traits was observed: the length of the stems, number of florets in the spikelet, the length of the spikelet and lemma awn, the width of the cauline leaves, the number of the ribs and quantity of the sclerenchymatous tissue in the leaf blade.


INTRODUCTION
The genus Festuca contains several hundred species.These were placed by H a c k e l (1882) in six sections.One of the sections, (sect.Ovinae Hackel., now sect.Festuca) is by far the largest (for example, 129 of the 170 species of Festuca recognized in Europe by M a r k g r a f -D a n n e n b e r d (1980) belong to it) and includes the great majority of fine-leaved species of the genus in North Temperate regions.H a c k e l (1882) recognized only eleven species in sect.Festuca, but two of these (F.ovina L. and F. rubra L.) were divided into a complex hierarchy of subspecies, varieties and lower rank taxa.Many of them are now recognized at the species level by modern workers.Of the 129 species recognized by M a r k g r a f -D a n n e n b e r d (1980), 91 are referred to as F. ovina sensu Hackel and 21 to F. rubra sensu Hackel.Fourteen of the latter 21 fit into our concept of the F. rubra aggregate (nigrescens, rubra, richardsoni, pyrenaica, trichophylla, pseudotrichophylla, cyrnea, oelandica, junicifolia, diffusa, rivularis, rothmaleri, nevadensis, cretacea), corresponding to H a c k e l ' s (1882) F. rubra subsp.eu-rubra, pyrenaica, dumetorum and nevadensis.Taxa of the F. rubra aggregate possess extravaginal non-flowering shoots (often as long rhizomes), and leaf-sheaths which are closed almost to the mouth and lack deeply infolded thin margins.They have been recorded as varying from diploid to decaploid (2n = 14, 28, 42, 56 and 70).
F. rubra L. and F. nigrescens Lam. are two widely distributed and ecologically important grasses that exemplify very variable species in the aggregate.In Flora Europaea, M a r k g r a f -D a n n e n b e r d (1980) distinguished seven lower rank taxa from F. rubra L.: subsp.arenaria, rubra, litoralis, asperifolia, pruinosa, juncea, thessalica and two from F. nigrescens Lam.: subsp.nigrescens and microphylla.
Up to date, there have been no detailed reports of F. rubra and F. nigrescens diversity in Poland.The primary aim of this study, which is a preliminary step for further investigations, is to determine the actual ranges of morphological diversity of F. rubra and F. nigrescens.The structure of the phenotypic variation in morphological characters is therefore analyzed to investigate the relationship between the species.In this study, a molecular system has been used with the objective of clarifying the taxonomic status of F. rubra and F. nigrescens.

MATERIAL AND METHODS
Plant material included samples of Festuca rubra L. from 14 sub-populations (N-252 individuals) and samples of Festuca nigrescens Lam. from 14 sub-populations (N-252 individuals) collected in Poland from different habitats (Fig. 1).50% of the F. rubra sub-populations originated from grassland habitats while the others were from well-lighted pine forests.Eight F. nigrescens sub-populations were from dry meadow communities, five from sand dunes, whereas sub-population N-97, characterised by a considerably bigger number of ribs in the leaf, a broader cauline leaf and the longest stems, originated from half-shadow mixed forest.Selected specimens from the particular populations were sampled as vegetative rhizomes from their natural habitats and, to avoid mutual contact, planted on distant plots in Maria Curie-Skłodowska University Botanical Garden in Lublin (Fig. 2).Each accession consisted of 18 plants of one sub-population.The study was carried out in the years 2001-2003 and in 2007-2010.The studies involved twenty three traits: 16 quantitative and 7 qualitative (Table 1 and 2) and consisted mainly in biometric measurements of traits that distinguish both species.Some observations such as the stem length and number, the length of panicle, presence of rhizomes and caespitose were carried out directly on living plant specimens on the experimental plots.The other characters which demanded precise observation of flower and leaf elements were counted or measured with a ruler or an Opta-Tech stereoscopic zoom microscope.The anatomical analysis was performed with the use of a Nikon light microscope.
For the phenetic analysis, univariate and multivariate methods were used.In the univariate methods, descriptive statistics of quantitative characters was calculated for each taxon, based on the entire data set.Box-and-whisker plots were used to display these data.Differences among the means of characters were tested using a one-way ANOVA.
Multivariate methods of analysis were carried out using version 7.1 of the STATISTICA programme (StatSoft Inc. 2007).The principle component analysis (PCA), and cluster analysis (CA) were performed (S o k a l and R o h l f , 1981).PCA was used to detect groups of specimens and to estimate the contribution of each variable to the analysis.PCA was based on correlation matrices of the whole set of quantitative characters (Table 1) and three qualitative characters (caespitose, presence of rhizomes, distribution of the sclerenchymatous tissue in the leaf blade), (Table 2).Cluster analysis was used to assess the groupings among Festuca sub--populations based on 16 quantitative characters and three qualitative traits (Table 1 and 2).Standardised data were used to compute the distance matrix based on average taxonomic distance, and this was subjected to the unweighted pair-group method, arithmetic average (UPGMA) clustering algorithm.
Semi-specific PCR were used for evaluation of genetic variability among F. rubra (R=14) and F. nigrescens (N=14) individuals.The sequences of primers were based on the consensus sequences of the intron--exon junction, 7 and 9 bases in length, common for plants and necessary for effective splicing (B r o w n , 1986).The additional bases were added at random to extend the length of the primers (R a f a l s k i et al. 1997).Verification of classification of the specimens into the particular similarity groups was based on the analysis of 680 DNA fragments.The results obtained facilitated computing the indices of genetic distance between the specimens, performing the cluster analysis and estimation of compatibility of the molecular analysis results with the origin of the ecotypes examined.The UPGMA cluster analysis was performed using version 7.1 of the STATISTICA programme (StatSoft Inc. 2007).
Sampled sub-populations of Festuca rubra (R) and F. nigrescens (N) were collected by A. Dąbrowska.

Quantitative characters
The results of the ANOVA analysis showed a significant variation (p<0.05) between Festuca species for the morphological characters (Table 1).However, the most useful characteristic for discriminating between the studied species of Festuca is the number of ribs in the leaf and the width of cauline leaf (second from the top).On the upper surface of the leaf blade of F. rubra there were not more than 6 triangular-shaped ribs, while in F. nigrescens there were from 6.6 to 9 ribs, slightly rounded at the top (Fig 2).The cauline leafs of F. rubra were flat with the average width of 2-3 mm; in F. nigrescens they were often compound and rarely exceeded 2 mm.The remaining quantitative characters overlapped between the species of Festuca.
There was no apparent difference between the two ecotypes of F. rubra from grassland and woodland.Examination of variation in the traits for these two ecotypes revealed that plants from grassland habitats produced more florets in the spikelet and had a longer second spikelet and a longer lemma awn (measured at the first floret in the second spikelet) in comparison with the ecotypes from woodland habitats (Fig. 3).
The UPGMA dendrogram for the Festuca accessions (Fig. 4) shows two major clusters.The first major cluster contains four sub-clusters of F. rubra accessions.Three of these contain accessions of F. rubra from woodland habitats (R-W) and the other and most numerous contains accessions of F. rubra from grassland (R-G).Both ecotypes of F. rubra appear to be very similar and are placed very close to each other in a single cluster.The second cluster contains accessions of F. nigrescens at some distance.It interesting to note that the accession of F. nigrescens (N-97) from the half-shadow mixed forest reserve is distinct.Exa-mination of the trait ranges for this accession revealed that N-97 is unusual, compared with the other F. nigrescens sub-populations, in having a bigger number of ribs, wider cauline leaves (second from the top) and longer stems (Table 1).The outlying values of these traits are likely to have resulted in separation of this population into a unique cluster.
There are some morphological differences between plants of different habitats within F. rubra those differences, however, are too inconsistent and imprecise to allow recognition of two intraspecific taxa within F. rubra.One would expect that, under such circumstances, a significant degree of morphological diversification would have occurred, making it possible to visually distinguish plants of different habitats.

Qualitative characters
F. rubra specimens formed lax tufts and rhizomes.The sclerenchymatous cells in the leaf blades were distributed in 5-8 small groups (3-,4-layered).All F. nigrescens individuals were densely caespitose and did not form rhizomes.Sclerenchymatous cells formed 4-, 6-or even 8-layered groups located under the lower epidermis opposite the vascular bundles and at the edge of the leaf blade.In cross-section, F. nigrescens leaf blades were wedge-shaped, with concave sides, whereas in F. rubra they were triangular (Table 2 and Fig. 2).
F. rubra individuals from grasslands had shorter rhizomes, which in the woodland specimens were -considerably longer.The sclerenchymatous tissue in the grassland specimens was distributed in remarkably bigger groups, compared with the woodland individuals.
Fourteen of the 18 individuals of F. nigrescens (N-97) that accessed from half-shadow mixed forest had hairy branches of the panicle; almost all individuals had hairy lemma and leaf-sheaths.
The panicle branches of most F. rubra individuals were usually hairy (73%), whereas only 52% of F. nigrescens plants had the same attribute.Most of the Festuca plants examined had hairy lemma and leaf-sheaths (Table 2).As regards hairiness, this character is not species-specific, since both species have the same attribute.

Quantitative and qualitative characters
The output from the PCA (Fig. 5) showed the pattern of morphological variation of Polish Festuca accessions based on 19 morphological characters: 16 quantitative and 3 qualitative characters (Table 1  and 2).The first coordinate axis presents the main directions of the species variation in the analysed group.The left side of the ordinate plane contains F. nigrescens specimens, and the right side -individuals of F. rubra.The gradient represented by the second axis is connected with habitat preferences which result in internal variability within species.The upper and central parts of the diagram show F. rubra ecotypes from grassland habitats (R-G), whereas the lower part presents ecotypes from the woodland habitats (R-W).The vector length indicates relative significance in formation of species variation.The longest vector of the intensity of changes represents the traits: caespitose, the number of ribs in the leaf, presence of rhizomes, quantity of the sclerenchymatous tissue and the number of stems.The first four traits are most correlated with the second coordinate axis, whereas the latter character is most correlated with the first axis.The other vectors are considerably shorter, which testifies to their lower importance in species variation.The directions of most vectors point to the upper diagram plane.

Molecular analysis of genetic variation
The dendrogram of Euclidesian distances for the Festuca accession (Fig. 6) shows two major clusters, within which remarkable distances between Festuca species can be observed (Table 3).The first cluster containing accessions of F. nigrescens is divided into two sub-clusters, in which the specimens display mutual relatedness, possibly due to their place of origin.Eight specimens from the first sub-cluster were from northern sub-populations, and the other six of the second sub-cluster from southern ones (Fig. 1).Ecotype N-97 from the half-shadow mixed forest is a distinctly separate group characterised by a bigger number of ribs, a broader cauline leaf (the second from the top) and longer stems (Table 1).The highest coefficients of genetic distance were calculated for this ecotype and a majority of F. nigrescens specimens from the northern sub-populations (Table 3).The results of this analysis also confirm distinctness of this ecotype probably resulting from the place of its origin.A similarly distant place is held by N-101 from the northern sub--population, in comparison with the specimens from the southern sub-population.
The second cluster contains accessions of F. rubra from grassland (R-G) and accessions of F. rubra from woodland (R-W) habitats.Considerable Euclidesian distances indicating high variation are observed between the profiles (R-G and R-W).The genetic distance coefficients between the extremely located specimens from these profiles exceed 0.400 (Table 3).This confirms the validity of division into similarity groups (R-G and R-W) shown by the morphological analysis (Table 1, Fig. 3 and 4) and the habitat preferences revealed by the PCA analysis (Fig. 5).

N-
Fig. 3. Variation of morphological characters for two ecotypes of F. rubra from grassland (R-G) and woodland (R-W) habitats.

DISCUSSION
The results obtained in the present study are partly consistent with investigation data reported by other authors (M a r k g r a f -D a n n e n b e r d , 1980; F a l k o w s k i , 1982; S z a f e r et al. 1986; C on e r t , 1996), who regarded the presence of rhizomes w F. rubra and absence thereof in F. nigrescens as the most important trait differentiating the two species.However, the results of this study do not correspond to those obtained by H u b b a r d (1973), who found the hairless lemma and leaf-sheaths in F. rubra to be the significant difference between the two species.Nevertheless, according to my observations, the pattern of variation within both Polish Festuca species is very similar as regards the lemma and leaf-sheath hairiness.89% and 80% of individuals of F. rubra and F. nigrescens, respectively, have a hairy lemma.Lemma hairiness, especially on the upper part, has a dominant character in both native species of Festuca in Poland.However, hairiness is a dominant characteristic of the leaf-sheaths; 79% and 88% of individuals of F. rubra and F. nigrescens, respectively, have hairy uppermost leaf-sheaths, 74% and 81% have hairy penultimate leaf-sheaths.Both studied species of Festuca consist of many forms as regards the hairiness of the lemma and leaf-sheaths, from almost glabrous types, through a multitude of intermediate forms, to abundantly hairy types.Almost glabrous types are rather rare within both species; most frequent are largely hairy individuals or plants of intermediate appearance.The above-mentioned statement has not been confirmed in the research conducted by S z a f e r (1919), M a r k g r a f -D a nn e n b e r d (1980) and F a l k o w s k i (1982), who ascribe little value to hairiness as a trait of individuals from the genus Festuca.The disagreement between the results presented here and those obtained by the above-mentioned authors is explained by the fact that two different species were compared.Exact boundaries of morphological similarity between taxa in the F. rubra aggregate are difficult to establish.Hence, there are difficulties in unambiguous identification of a particular taxon.1984).Occurrence of this phenomenon between the two F. rubra and F. nigrescens populations under study is highly probable, since they exhibit many common inflorescence features, which according to C o n e r t (1996) are assigned to the individual species, i.e. in comparison to F. nigrescens, F. rubra should have a longer panicle with glabrous branches, a bigger number of flowers in the spikelet, and a longer lemma awn.No distinct boundaries of taxonomic distinctiveness related to the aforementioned traits were found in the present study.At present, it is difficult to suggest the taxonomic status of these taxons.F. rubra and F. nigrescens are cleistogamous and largely autogamous species; however, it hybridizes to an inconsiderable extent.The taxons in question may, therefore, be a product of hybridization.Hybrid zones may confound morphological characters alone; therefore, other markers -preferably chromosome count -would be needed in further assessment of these populations.Further investigation and confirmatory chromosome count in a region where both Festuca species sympatrically appear are needed to cast light on this interesting issue.
It is interesting that ecological segregation among accessions of Festuca was detected on the basis of morphological traits.The results of PCA and genetic analysis can only be interpreted to support the recognition of single sub-populations in F. rubra and F. nigrescens.
These observations suggest that characterisation of the floristic composition of the potential collection would be helpful in maximizing the genetic diversity among ecotype populations.A similar conclusion was drawn by S u z u k i et al. (1999,2006)

CONCLUSIONS
The main characters for identifying the two species studied: the number of ribs in the leaf, width of the cauline leaf (second from the top), caespitose, presence of rhizomes, and quantity of the sclerenchymatous tissue in the leaf blade.
Common characters of the species studied: the number of florets in the second spikelet, the length of the stem, the length of the lemma awn, the length of the second spikelet, the width and thickness of the leaf, the number of spikelets in the panicle, the number and diameter of vascular bundles in the leaf, the length of the cauline leaf, the weight of 1000 grains, the length of hairs in the leaf, the length of the panicle, the number of the stems, hairiness of the panicle branches, the hairiness of the lemma and leaf-sheaths.
Differences related to the area of plant origin: the number of florets in the second spikelet, the length of the second spikelet, the length of the lemma awn (measured at the first floret in the second spikelet), number of ribs in the leaf, width of the cauline leaf (se-cond from the top), the length of the stem, the length of the rhizomes, quantity of the sclerenchymatous tissue in the leaf blade.

Fig. 1 .
Fig. 1.Map of Poland with geographic locations of the sub-populations of Festuca rubra (square) and F. nigrescens (circle).

Fig. 4 .
Fig. 4. UPGMA dendrogram of 28 accessions of Festuca species from Poland based on Euclidean distances computed among accession means of quantitative and qualitative morphological traits listed inTable 1 and 2.

Fig. 6 .
Fig. 6.Association among sub-populations Festuca revealed by UPGMA cluster analysis of 680 DNA fragments.R-G accessions of F. rubra from grassland, R-W accessions of F. rubra from woodland habitats.
Numerous authors for example, M i r e k et al. (2002), F r e y (2003) or R u t k o w s k i (2003) have their own concept of taxonomic divisions.Both species of Festuca studied occur in the same plant communities, which renders hybridisation possible (S t a c e , 1975; S t a c e and A i n s c o u g h , ; N o v a et al. (2006); Š m a r d a (2008); Š m a r d a et al. (2008); A r m o n i e n et al. (2010); Y a m a d a (2011).

Table 1
The range for the 16 quantitative characters and the results of ANOVA.Differences between species were considered significant at the level of p<0.05

Table 2
Descriptions, abbreviations and attributes of qualitative characters and their frequencies in Festuca rubra and F. nigrescens.
p = characters used in PCA *estimate at the first floret in second spikeletTable 3.