Introduction

Relationships between species of Lolium L. and Festuca L. have long been remained as a controversy in taxonomy of the subtribe Loliineae. Lolium which was first classified in the tribe Triticeae Dumortier based on the morphology of inflorescence, was later transferred to the tribe Poeae (= Festuceae) by Nevski (1934). Many non-morphological evidences from different sources of data supported this transfer (see Darbyshire, 1993). Inter-generic hybridization between Festuca subgen. Schedonorus (P. Beauvois) Petermann and outbreed species of Lolium (specially, L. perenne L.) along with other evidences from cytology, anatomy, molecular markers and genomic and plastid DNA, supported the union of F. subgen. Schedonorus and Lolium. Darbyshire (1993) suggested the realignment of F. subgen. Schedonorus with the genus Lolium, and introduced new combinations: Lolium subgen. Schedonorus Darb., L. arundinaceum Darb., L. giganteum Darb., L. mazzettianum Darb., and L. pratense Darb. Festuca subgen. Festuca sect. Ovinae Fries (syn.: sect. Festuca), encompasses two controversial aggregates namely F. ovina L. and F. rubra L., according to the classification proposed by Hackel (1882), and with hundreds of species, subspecies, varieties, subvarieties, and formes, published so far. Stace et al. (1992) reduced all named morpho-anatomical variations into those two aggregates. He stated that only two characters (Sheaths of young tiller-leaves: fused/free, and tillers: extravaginal/ intravaginal) could definitely be used to distinguish between Festuca ovina and F. rubra aggregates; supporting the notion proposed by Hackel (1882) who divided the section into two main groups; Intravaginales and Extravaginales vel Mixtae. Phylogenetic studies based on cpDNA-RFLP (Darbyshire and Warwick, 1992) and ribosomal ITS (Charmet et al., 1997; Gaut et al., 2000; Torrecilla and Catalan, 2002) have demonstrated the paraphyly of Festuca and that, Festuca may include Lolium and Vulpia C.C.Gmel., therefore, choosing between the transfer of F. subgen. Schedonorus as a new genus (Soreng and Terrell, 1998), or realignment of it with genus Lolium (Darbyshire, 1993) remained to be studied. Torrecilla and Catalan (2002) worked on two main lineages in festucoids, namely “fine-leaved” and “broad-leaved fescues”, and demonstrated that Lolium species were close relatives of broad-leaved fescues, and that the polyphyletic Vulpia was a close relative of fine-leaved fescues lineage. They confirmed that a monophyletic Festuca might encompass species from Vulpia, Leucopoa Griseb., Schedonorus and Lolium, while realignment of Schedonorus with Lolium would remain the rest of the clade as a large polyphyletic assemblage. Phylogeny of the festucoids based on nucleotide sequences of ITS and trnL-F regions (Catalan et al. 2004) showed that Lolium, Mycropyropsis and Festuca subgen. Schedonorus were close relatives, and they fell into a single monophyletic clade. Although several appreciated studies have been already performed in this complex group (Catalan et al., 1997; Catalan, 2002; Torrecilla et al., 2003; Catalan et al., 2004; Torrecilla and Torrecilla et al., 2004; Catalan, 2006; Muller and Catalan, 2006), the festucoids, Festuca s. str. and Lolium s.l. are still open and interesting subjects to be studied for more details.

Bor (1970) described six species of genus Lolium s. str. for Iran: L. perenne L., L. multiflorum Lam. (syn: L. italicum A.Br.), L. rigidum Gaud. (syn: L. strictum Presl.), L. persicum Boiss. and Hohen. ex Boiss., L. temulentum L. and L. loliaceum (Bory and Chaud.) Hand.-Mazz which was later considered as a synonym for L. rigidum subsp. lepturoides Sennen and Mauricio (legitimate name). They coincided with members of Festuca subgen. Shedonorus in high mountain elevations and mesic habitats along Alborz and Zagros chains. Flavonoids have long been proved as important characters in plant systematics and biosystematics researches, and they were still continue to take part specially in biosystematics researches (Sharifi-Tehrani and Ghassemi Dehkordi, 2011; Ghassemi Dehkordi et al., 2012; Sharifi-Tehrani et al., 2012).

This study was aimed to evaluate the relationships in Lolium sensu Darbyshire in Iran, using flavonoids spot profiles and quantitative morphological characters. Lolium specimens were studied here along with specimens from Festuca subgen. Schedonorus and from more distantly sister group, Festuca subgen. Festuca. This was the first report on the numerical analysis of quantitative morphological characters of Lolium s.l. in Iran. This study was also the first one to report application of digitally measured flavonoids spot profiles in the chemotaxonomy of the group in Iran. Relevance of this study was due to the importance of members of Lolium s.l. as economic hay and forage plants in Iran and the neighboring countries in the west of Mediterranean region, and also the relative absence of F. sclerophylla Boiss. ex Bisch. and L. persicum in previous studies.

Materials and Methods

Plant material

Plant materials were collected from wild populations throughout their distribution ranges in Iran (Table 1). Specimens were paper-dried, and determined using identification keys available in Flora Iranica and Flora of Turkey (Bor, 1970; Davis et al., 1988). Seventy eight samples were chosen for flavonoid extractions or morphological studies (Tables 1, 2). Total flavonoids were extracted from 0.6 to 1.0 gram of dried leaves of 59 selected specimens belonging to ten species from three closely related genera.

Table 1. Plant material collected from wild populations in Iran. Acc: accession numbers of each specimen; TLC lane: corresponding flavonoid profile in Figure 1; S: corresponding plot in Figure 2

Table 2. List of morphological characters

Extraction of flavonoids and TLC

Plant materials (leaves) were ground to a fine powder using mortar and pestle. Extraction was performed using 80% MeOH for 36 h according to Markham (1982) with modifications. Solvent of the filtrate was evaporated using rotary evaporator in 50-60° C under relative vacuum. Dried extracts were dissolved in distilled water and filtered to discard fatty substances. Aqueous extract was dried again and dissolved in 5 ml MeOH. One dimensional thin layer chromatography was performed using 20 cm × 10 cm Aluminum sheets covered with silica gel 60F254 (Merck). Solvent system consisted of water: 20, acetic acid: 20, iso-propanol: 10, butanol: 50. Separated flavonoid spots on TLCs were visualized under UV 254nm and 366nm and digitally photographed using a Canon EOS 500D digital camera. Skewness of images was corrected in Adobe Photoshop software ver. 13.0 CS6 x64 extended. Then, images were calibrated in ImageJ software ver. 1.47s (Rasband, 2008) and for each specimen, the flavonoid spots profile were plotted. Migration distance for each spot was measured in ImageJ software, and data transferred to Microsoft Excel 2013 to calculate the Rf values (Table 3). The area under each flavonoid spot in each profile was measured in ImageJ software and data were transferred to Microsoft Excel 2013 to calculate the percentage of each spot in its corresponding profile, where the bar graphs for each profile were drawn.

Table 3. Rf values and percentage of each identified flavonoid spot in its corresponding profile. Letters in first column are same as letters in Figure 2. Numbers in first row are numbers of spots on the chromatogram from small Rfs to large Rfs (right to left on chromatogram)

Morphology

Sixty eight quantitative morphological characters from both vegetative (17) and reproductive (51) characters were used for morphological study (Table 2). Each character was measured 3 to 6 times (independent measures on same specimen) to calculate averages and standard deviations (Table 4). Measurements were performed on several digital images taken with 15 megapixels Cannon EOS 500D camera capable to connect to stereomicroscope, and using millimeter-papers to calibrate the images. Calibration of images was performed using ImageJ software, and measurements were transferred to Excel datasheets to calculate the basic statistics (min, max, average, and SD). Inapplicable characters such as ‘upper glume length’ for Lolium spp. specimens were considered as missing values. Data were converted to NTS format of NTSYS-pc software and analyzed using Simint, Njoin, SAHN, Eigen and Mod3D modules. Cosine distance (dissimilarity) coefficient () was used for calculating dissimilarity matrix, and UPGMA was used as the sorting method in cluster analysis. Same coefficient was used for PCO analysis.

Results and Discussion

Flavonoids spot profiles

Solvent system was optimized to achieve best separation of flavonoid spots in one dimensional TLCs (Figure 1). Two dimensional test TLCs moved spots on diameter of TLC and confirmed efficient separation of spots using one dimensional TLCs. To document the process of optimization of solvent system, a database system (unpublished) was designed to help storage and retrieval of our TLCs data containing components of each tested solvent, list of samples on each chromatogram, TLC images, and to making specialized reports.

Figure 1. TLC chromatograms of selected specimens. Number beneath each lane refers to numbers in ‘TLC lane’ and corresponding specimen in Table 1

Flavonoids were separated in range Rf = 0.045 to Rf = 0.93. Number of identified flavonoid spots (separate Rf values) in each extract ranged from 4 spots in L. persicum and V. myuros (K, M in Figure 2), to 11 spots in L. perenne (H in Figure 2). Percentage of spots in their corresponding extracts ranged from 1.44% (spot 1 in L. multiflorum; F) to 59.64% (spot 5 in L. rigidum; I).

Plots of F. arundinacea, F. pratensis, F. sclerophylla and L. multiflorum (A; B, D; C; F in Figure 2) were topologically similar, showing large spots in Rf range 0.5-0.7. Plots of L. perenne samples (G, H in Figure 2) were close to this group, but with less strong spots (area under curve, or percentage of picks). Four plots belonging to V. myuros, L. persicum and F. gigantea were also similar, although the members of this group were distantly connected together. Plot of F. alaica was distinct, containing a large spot in Rf 0.45. Flavonoid spot profile of L. persicum was most similar to that of F. gigantea (syn. L. giganteum sensu Darbyshire). Percentage of each spot and its corresponding Rf value for all specimens are presented in Table 3.

Figure 2. Plots showing location (Rf), intensities and percentage of each flavonoid spot in selected specimens. Arrows point to pick of each spot, numbers above each arrow: first number corresponds to Rf values, and second numbers are percentage of each spot in its profile. Bar charts are drawn according to percentage values. A. Festuca arundinacea; B. F. pratensis; C. F. sclerophylla; D. F. pratensis; E. L. rigidum; F. L. multiflorum; G. L. perenne; H. L. perenne; I. L. rigidum; J. L. rigidum; K. L. persicum; L. Vulpia myuros; M. V. myuros; N. F. gigantea; O. F. alaica.

Festuca arundinacea L. (syn.: Lolium arundinaceum (L.) Darb.): 10 distinct spots were identified (profile A) for this species which Rfs ranged from 0.13 to 0.8. The most prominent spot had Rf = 0.63 which contained 26.6% of the total flavonoids in the corresponding extract. Festuca pratensis Hudson (syn.: Lolium pretense (Hudson) Darb.): 10 distinct spots were identified (profiles B, D) for this species which Rfs ranged from 0.06 to 0.84. The most prominent spot had Rf = 0.66 (profile D) which contained 21.7% of the total flavonoids in the corresponding extract. Festuca sclerophylla Boiss. et Hohen. (syn. Leucopoa sclerophylla (Boiss. et Hohen.) Krecz. et Bobr.): 10 distinct spots were identified (profile C) for this species which Rfs ranged from 0.11 to 0.93, and the most prominent spot had Rf = 0.59 which contained 33.3% of the total flavonoids in the corresponding extract. Lolium multiflorum Lam. (syn. L. italicum Braun): 9 distinct spots were identified (profile F) for this species which Rfs ranged from 0.12 to 0.78. The most prominent spot was Rf = 0.6 which contained 33.2% of the total flavonoids in the corresponding extract. Lolium perene L. (syn. L. marschallii Steven): Up to 11 distinct spots were identified (profiles G, H) for this species which Rfs ranged from 0.1 to 0.8. The most prominent spot was Rf = 0.6 (profile G) which contained 38.4% of the total flavonoids in the corresponding extract. Lolium rigidum Gaudin: Up to 9 distinct spots were identified (profiles E, I, J) for this species which Rfs ranged from 0.1 to 0.83. The most prominent spot was Rf = 0.6 (profile I) which contained 59.6% of the total flavonoids in the corresponding extract. Lolium persicum Boiss. and Hohen. ex Boiss.: 4 distinct spots were identified (profile K) for this species which Rfs ranged from 0.14 to 0.46. The most prominent spot was Rf = 0.19 which contained 27.7% of the total flavonoids in the corresponding extract. Vulpia myuros (L.) C.C.Gmel.: 4 distinct spots were identified (profiles L, M) for this species which Rfs ranged from 0.05 to 0.68. The most prominent spot was Rf = 0.05 (profile M) which contained 27.8% of the total flavonoids in the corresponding extract.

Festuca gigantea (L.) Vill. (syn. Lolium giganteum (Linnaeus) Darb.): 4 distinct spots were identified (profile N) for this species which Rfs ranged from 0.11 to 0.59. The most prominent spot was Rf = 0.4 which contained 49.6% of the total flavonoids in the corresponding extract. Festuca alaica Drobow: 5 distinct spots were identified (profile O) for this species which Rfs ranged from 0.06 to 0.61. The most prominent spot was Rf = 0.45 which contained 44.6% of the total flavonoids in the corresponding extract.

Morphology

Phenetic relationships between species belonging to Lolium s. str. plus those Festuca spp. routinely hybridize them (i. e. Lolium s.l., excluding F. gigantea) were studied using quantitative morphological characters. Measurements were performed using digital images taken from different parts of specimens, while a millimeter paper was in background of each photo. After calibration of images in ImageJ software, measurements were performed and a scale bar (white on black background) was superimposed on each photo and backgrounds were replaced with black color. Fertile parts of florets in six Lolium s.l. species are shown in Figure 3.

Figure 3. Fertile parts of florets in six Lolium s.l. species. Scale Bars length = 0.2 mm. A. Festuca pratensis; B. L. multiflorum; C. L. rigidum; D. L. perenne; E. F. arundinacea; F. L. persicum

Results (Table 4) showed that quantitative morphological characters contained variations that could be used both for description of taxa and multivariate analysis to elucidate the phenetic relationships between them. Table 4 summarizes the data, as the first least- and most-variable characters for each taxon are reported, with annotations for their min, max, SD, and P-values (P is the standardized value, and is defined here as: range divided by the mean).

Table 4. Measurement of morphological characters. The unit of all measures is mm

Multivariate analysis

Multivariate analysis (cluster and PCO analyses) of six Lolium s.l. based on quantitative morphological characters was performed to elucidate the phenetic relationships between them. Resultant dendrogram using Cosine coefficient and UPGMA method (Figure 4A) showed that L. persicum specimens were grouped together and distantly separated from the rest of specimens. Specimens belonging to F. pratensis were also grouped together (except for F. pratensis acc.# 833; a specimen collected from Mt. Dena in central Zagros region). Specimens belonging to F. arundinacea were divided into two well defined groups. Geographical location of populations from which specimens of F. arundinacea were collected, did not support the sub-grouping of these specimens. However, the partitioning of F. arundinacea specimens based on quantitative morphological data was interesting. Another interesting result was the misplacement of one specimen belonging to L. persicum (acc.#799, collected from Ardabil province, Khalkhal) which was grouped with F. pratensis specimens. Cophenetic analysis (Figure 4B) showed that there was moderate levels (r = 0.58) of correlation between resultant dendrogram and the underlying distance matrix. However, the results were strongly confirmed with resultant plot from PCO analysis (Figure 4C) which showed the separation of L. persicum, partitioning of F. arundinacea and misplacement of L. persicum acc.#799. Results of PCO analysis confirmed that the first three axes had collected and expressed 82 percent of the variation held in raw data matrix (Table 5).

Figure 4. Multivariate analyses. A. Cluster analysis based on Cosine dissimilarity coefficient, and UPGMA as sorting method; B. Co-phenetic plot; C. Principal Coordinates analysis based on Cosine dissimilarity coefficient for quantitative data

Table 5. Axes loadings in PCO analysis. More than 82 percent of variation was expressed by three first axes

Partitioning of Lolium persicum was reported in a previous work by Sharifi-Tehrani et al. (2008), so that the population Ardabil, Khalkhal (the population from which acc.#799 was collected), was distantly separated from other populations. Although population Khalkhal was clearly identified as L. persicum based on diagnostic characters and available keys, however, quantitative morphological studies and microsatellites provided evidence supporting for its separation from other L. persicum populations. To elucidate the taxonomic status of this population, more extensive studies are required.

SDS-PAGE analysis of seed protein profiles of taxa belonging to genera Festuca and Lolium proved to be useful for classification of festucoids (Aiken et al., 1998). Their method consisted scoring of each protein band identified based on Rf values, and analyzing the qualitative data using Jaccard’s coefficient. Their resultant dendrogram showed that members of Festuca subgen. Schedonorus (F. arundinacea and F. pratensis), had constructed a major group with Lolium spp., within which F. pratensis specimens were sub-grouped together, and F. arundinacea was closer to L. rigidum than to other outbreeding Lolium spp.

Results obtained from analysis of quantitative morphological characters in our study were concordant to Aiken’s results. F. arundinacea although divided into two separate groups, showed close relationships with L. rigidum. Analysis of quantitative morphological data provided more resolution in F. arundinacea, and the separation of L. persicum Acc.#799 (Ardabil, Khalkhal) from other populations in this study was also in concordance with the molecular analysis by Sharifi-Tehrani et al. (2008). The application of flavonoids spot profiles for classification of Festuca and Lolium species in this study was comparable to SDS-PAGE profiles of seed proteins Aiken et al. (1998). Flavonoids spot Profiles belonging to the members of F. subgen. Shedonorus were similar to profiles of their relatives in genus Lolium. Presence/absence of spot profiles were not used here as qualitative data to elucidate the relationships, or to classify taxa, as the homology of spots are to be certified. Observed variations in spot profiles of the studied species, specially, in L. rigidum, L. perenne, and F. pratensis claimed for their applicability for investigating the chemical variation between the populations within the species level. TLC chromatograms of flavonoid extracts in this study provided sharp-enough bands which could be scored and analyzed. Close relationships between F. pratensis and L. multiflorum which was demonstrated by Pasakinskiene et al. (1998) through analysis of GISH bands, was also confirmed by both morphological data (Figure 4C; PCO plot) and flavonoids spot profiles (Figures 2F, 2B and 2D).

Application of quantitative morphological characters for phenetic classification of Lolium spp. was reported in a recent work by Oshib-Nataj et al. (2011), where the resultant dendrogram clustered the 33 specimens into the 5 species. Relationships between some Iranian members of festucoids (including Lolium) using AFLPs (Majidi et al., 2006; Majidi and Mirlohi, 2010) showed the close relationships between F. arundinacea and F. pratensis specimens, to which group, the specimens belonging to L. perenne and L. rigidum were connected. Results obtained from analysis of AFLPs were concordant with the previous foundings about relationships in this genus. The study also demonstrated the application of AFLPs for genetic relationships studies in cool season grasses. The phenetic analysis of Iranian species of Lolium based on 27 morphological characters, measured on 68 specimens (6 species) dispersed L. perenne and L. multiflorum (closely related taxa) among other Lolium species which claimed for the inapplicability of morphological characters for phenetic analysis of Lolium, i. e. (Mirjalili et al., 2008). Relationships between Lolium species in the resultant clustering scheme could hardly be accepted (see also Mirjalili and Bennett, 2006) regarding the previous results from many other literatures (see Darbyshire, 1993 and refs. there in for review).

Results of the current study based on quantitative morphological data and flavonoids spot profiles, were in concordance with the previous foundings about species relationships in this group, and produced interpretable groupings in both cluster- and PCO analyses which also further confirmed our previous founding about L. persicum population Khalkhal (Ardabil province of Iran). Flavonoid spot profile of F. gigantea was different from those of F. arundinaceae and Festuca pratensis. These results did not support the results of seed protein electrophoresis analysis as they produced similar seed protein profiles (Bulinska-Radomska and Lester, 1985).

Conclusions

For complex plant groups (such as Lolium s.l.) misleading characters should be identified and avoided. Many non-reproductive characters and certain reproductive characters (see Bulinska-Radomska and Lester, 1988) may lead to uninterpretable results. Careful selection of morphological characters, along with adoption of proper methods for analysis, will have great impact on resulting phenograms. Measurements of morphological characters reported in Table 4 may be of interest for those researchers working on gene pools of these taxa using molecular markers or for plant breeders or physiologists who want to know how these taxa may vary in their different morphological characters. The grouping of Iranian F. arundinacea populations into two distinct subgroups was an interesting result in our study which supported for existence of two forms in mixed populations. The applicability of quantitative morphological data to reveal phenetic relationships in this taxonomically complex group, and the potential usage of flavonoids to further description of taxa with biochemical data, and to study the variation held in their populations, were reported in this study and are intended for a more extensive study in tribe Poeae.

Acknowledgments

This study was part of the MSc thesis fulfilled by S. Raeisi-Chehrazi, conducted by H. Saeidi and M. Sharifi-Tehrani. Authors would like to thank deputy of research and office of graduate studies at the Universities of Isfahan (UI) and Shahrekord (SKU), for their supports. Authors also appreciate two anonymous referees for their invaluable comments.