MORPHOLOGICAL CHARACTERIZATION OF Cyclamen sp. GROWN NATURALLY IN TURKEY: PART II

The morphology of 279 accessions of Cyclamen sp. growing naturally in Turkey, namely C. alpinum (syn. C. trochopteranthum), C. graecum, C. hederifolium (syn. C. neapolitanum) and C. mirabile, was characterized. Plants with intact tubers were collected from locations in Antalya, Isparta, Aydın, Muğla, İzmir and Denizli, determined by GPS, where they grow naturally in spring and autumn. The morphology of the four Cyclamen species was characterized using one year old regenerated plants based on 27 morphological traits (13 flower, 11 leaf, 2 plant, 1 tuber). There were distinct differences among these accessions related to petal colour, pedicel length, leaf length and width, leaf shape, and tuber diameter. Even though principle component analysis confirmed the grouping of characters into species-specific clusters, a wider range of morphological data as well as molecular data are needed for more reliable conclusions to be drawn about the classification of these Cyclamen species.

ic affinity within and among species, define the relationship between populations and detect whether there is genetic drift or migration between them [Dolezalova et al. 2003, Alhajjar et al. 2011]. The combination of morphological, chemical, and molecular marker analyses will result in a better classification of the genus Cyclamen. The objective of this study was to characterize the morphological traits of four Turkish Cyclamen species (C. mirabile, C. graecum, C. hederifolium and C. alpinum) growing in natural conditions and to analyze their relationships using cladistic analyses.

MATERIALS AND METHODS
Plant material. Cyclamen plants were collected with intact tubers from locations where they grow naturally, in spring and autumn, when they were in flower: C. graecum  [Davis 1978, Davis et al. 1988, Guner et al. 2000]. Cyclamen populations were surveyed in each province and 50 plants were collected from each location. Plants were sprayed with water to retain high relative moisture in a plastic bag packed into a cardboard box used for transporting the material to an unheated greenhouse in Adana where they were propagated. For propagation, plants were sprayed with an insecticide (Viaduct, water-soluble granules; 5% Emamectin benzoate (Shenzhen Chuangye Industry Co. Ltd., China) then planted in 3-L plastic pots filled with peatmoss, sand, and perlite (1:1:1, v/v/v). Pots were placed on a raised bench in a greenhouse. The plants were shaded as needed using a net providing 60% shade, irrigated weekly with 300 ml of tap water per pot and fertilized every two months with 2 g/pot of NPK 20-20-20 + ME fertilizer (Nutri-Leaf EC Fertilizer, Miller Chemical & Fertilizer Corp., Hannover, PA, USA) [Curuk et al. 2015]. Plants were transferred to a cooler place during summer and were not watered. When leaves started to emerge in the next season, pots were transferred them to a greenhouse where plants were once again watered and fertilized. Cyclamen species were characterized morphologically one year later in regenerated plants (not all plants regenerated and the number of regenerated plants for each species were 56 accessions out of 100 in C. mirabile, 154 accessions out of 205 in C. hederifolium, 10 accessions out of 100 in C. graecum and 59 accessions out of 209 in C. alpinum).
Morphological characterization. There is no description list in UPOV and IPGRI for the Cyclamen genus. Thus, the properties of the four Cyclamen species studied were determined with a description list (tab. 2) modified from Debussche and Thompson [2002] containing important leaf-, flower-and tuber-related features of Cyclamen. A total of 27 phenotypic characters (13 flower, 11 leaf, 2 plant, 1 tuber) including 13 quantitative (7 flower, 5 leaf, 1 tuber) presented in Table 3 and 14 qualitative traits (6 flower, 6 leave, 2 plant) were evaluated in most of the accessions. A few genotypes were excluded because values were missing, because of the use of an insufficient num-208 P. Curuk, Z. Sogut, T. Izgu, B. Sevindik, E.M. Tagipur, J.A. Teixeira da Silva, S. Serce...

_____________________________________________________________________________________________________________________________________________
Acta Sci. Pol. ber of morphological characters or mistiming of measurements for parameters resulting in a failure to characterise them with confidence. Flower number/plant, pedicel length (cm), petal length (mm), petal width (mm), basal corolla ring diameter (mm), pistil length (mm), stamen length (mm), leaf number/plant, lamina length (cm), lamina width (cm), lamina length/lamina width ratio, petiole length (cm) and tuber diameter (mm) were examined. Length was determined by a ruler or a digital compass (Mitutoyo CD-15D, Kawasaki, Japan). Statistical analysis. All phenotypic characters and quantitative morphological traits were measured in triplicate for each accession. Mean values were recorded in a Microsoft Excel (2010) spreadsheet and raw data were then coded to allow analysis using Unistat 4.0 for Windows. Statistical analyses were conducted using SAS [SAS Institute Inc., NC, USA, 1990]. Data for the 13 quantitative characters was analyzed to determine means, standart deviations, minimum and maximum values of each species using SPSS.14.
In terms of quantitative traits, specifically petal number, different values were observed in C. hederifolium ( fig. 5) accessions. This may be because the temperature of the cultivation site was higher (average temp. 19.1°C) than their natural location (average temp. 17.9°C in Izmir, and 17.6°C in Aydin) [MGM 2015].   . 9). Tubers were larger in C. graecum and C. hederifolium. Cyclamen species with more than 30 chromosomes, such as C. hederifolium (2n = 34) and C. graecum (2n = 84), can develop very large tubers [Clennett 2002]. Thus, variation among species in certain traits may be associated with differences in chromosome number associated with polyploidization [Debussche et al. 2004]. It can also be associated with the age of the accessions since Cyclamen can not produce sister tubers but enlarge with ageing [Le Nard and De Hertogh 1993]. Plant growth habit was weak in 28 (10%), medium in 159 (57%), and strong in 92 (33%) accessions. Flowering preced-

Hortorum Cultus 15(5) 2016
ed leaf development in 274 (98.2%), or followed leaf development in 5 (1.8%) accessions. The fruit is a capsule, often drawn down to the soil level by the twisting of the floral stalk [Davis 1978, Le Nard and De Hertogh 1993, Debussche and Thompson 2002. Pedicel coiling was present in all accessions (279, 100%) of all four species (fig. 8). These findings were adjusted with Clennett's cladistic data matrix [Clennett 2002]. These species share several characteristic features that diagnose them as a monophyletic group, such as a well-developed tuberous subterranean bulb formed by swelling of the hypocotyl, conspicuously reflexed corolla lobes, and coiled fruiting pedicels. In all Cyclamen species except for four, the pedicel coils down to the soil surface when the fruit ripens, usually coiling from the apex downwards, but in C. rohlfsianum the pedicel coils from the base towards the apex, and in C. graecum it coils two ways from the middle ( fig. 8B). In two other species, C. persicum and C. somalense, the pedicel does not coil, but becomes curved at anthesis ]. All Cyclamen species have the same broad growth aspect, but differ from each other in characters such as size and structure of the tubers, denticulation of the leaf margin, width of corolla mouth, presence or absence of auricles at the corolla mouth, chromosome number, and the time of year in which the flowers develop , Mammadov et al. 2016]. This range of phenomena was also confirmed for the four species examined in this study, and in another four species examined in our previous study [Curuk et al. 2015]. The standard deviations, as well as minimum and maximum values were compared (tab. 3). Morphological diversity does not correspond to geographic region but rather to species. Among the species sampled, C. hederifolium contained the highest mean values for flower number/plant (7.2), petal length (20.6 mm), pedicel length (11.4 cm), lamina length (5.3 cm) and tuber diameter (5.4 cm). C. alpinum and C. graecum shared the highest mean values for petiole length (10.45 cm). The highest mean values for leaf number/plant were 15.6 in C. mirabile and followed by 13.3 in C. hederifolium. Tuber A C B D diameter ( fig. 9) was greatest in C. graecum and C. hederifolium (45.13 and 53.89 mm, respectively). These results are broadly compatible with the description of the Flora of Turkey [Guner et al. 2000]. Also, in Sahin and Burun's [2010]     Based on a cluster analysis (data not shown) performed using morphological data, accessions were clustered into the same groups based on species but not on geographic origin. For example, C. coum and C. persicum from part I [Curuk et al. 2015] and C. hederifolium accessions collected from different locations flowered in the same season, and had the same colour of the abaxial leaf surface and the same petal colour (tab. 2). Clustering according to plant species has also been reported in Oncocyclus irises [Saad and Mahy 2009] and chamomile (Matricaria chamomilla) [Solouki et al. 2008].
Principle component analysis. The results of PCA are presented in Table 4. The first PC explained 37% (eigenvalue of 4.44), while the second and third PCs contributed 13 and 12%, respectively to total variance (eigenvalues of 1.58 and 1.41, respectively). PCA explained 62% of total morphological variation. Lamina length (0.41), lamina width (0.38), basal corolla ring diameter (0.39), pistil length (0.38) and petal length (0.35) were important variables defining PC1. Leaf and flower number/plant (eigenvalues of 0.64 and 0.62, respectively) were the two traits that were highly correlated with PC2. Petiole length (eigenvalue of 0.67) was the most important trait for PC3.

Hortorum Cultus 15(5) 2016
From all characters, petiole length was found to be the most discriminative parameter differentiating accessions. Similar trends were observed in C. persicum, C. cilicium, C. pseudibericum and C. coum growing naturally in Turkey [Curuk et al. 2015]. Mih et al. [2008] evaluated the morphology of four selections of Vernonia hymenolepis A. Rich. based on PCs, reporting a high level (80%) of morphological variation. Fig. 10. Plot of four Cyclamen (C. alpinum, C. graecum, C. mirabile, C. hederifolium) accessions collected from Turkey on the first three principle components (PCs) obtained from the analysis of 12 quantitative agro-morphological traits using PCA According to PCA using qualitative and quantitative data obtained from the observations of 27 morphological characters in these four Turkish Cyclamen species, accessions could not be separated into different groups ( fig. 11). While PC1 explained 30% of total morphological variation, PC2 explained 8% and PC3 8% of total variation, with a total of 46% variation accounted for. Aalaei et al. [2007] found that effective traits related to the colour of flowers, petioles, young leaves on the adaxial site, and flower shelf life could be categorized into six groups that covered a total of 86.89% of total variation. In this study, PCA results showed that Cyclamen species that originated from a similar geographic area or the same species that originated from different geographical areas were grouped separately ( fig. 10). Zhao et al. [2007] found that the lack of conformity between genetic and geographic variation exists because of an exchange of genetic material, the introduction of new accessions, genetic drift and natural selection, or human interference. Plants are nearly always restricted to specific geographic and climatic zones in areas with great topographic and climatic variation. However, the appearance of different accessions is related to basic evolutionary elements (selection, adaptation, migration, self-pollination and genetic drift), which are related to environmental and anthropogenic activities over time [Martins et al. 2006]. In the present study, during our survey, we came across C. hederifolium which grows naturally in the western part of Turkey, in one location of Urun village, Osmaniye located in the southern part of Turkey. This might be because of human interference, introduction of new accessions or exchange of genetic material. Fig. 11. Plot of cyclamen accessions collected from Turkey on the first three principle components (PC) obtained from analysis of 27 agro-morphological traits using PC analysis

CONCLUSIONS
The results obtained in this study allowed for the accurate identification and discrimination of four Cyclamen species (C. hederifolium, C. alpinum (syn. C. trochopterantum), C. graecum, C. mirabile) growing wild in Turkey, which boasts over 270 accessions. This set of results complements our investigation of C. persicum, C. cilicium, C. pseudibericum and C. coum, also growing naturally in Turkey [Curuk et al. 2015]. Tuber diameter, leaf width, petal length, petal colour, leaf shape, and the presence of auriculate characters were the most acutely different morphological characteristics. Such a detailed morphological characterization is useful for identifying superior accessions that are hardy to cold, have attractive leaves, or flower fragrance, which could be taken into consideration in future hybridization programs of Cyclamen. In such a programme, a wide range of variability, heritability, genetic advances and positive correlation coefficients among traits can be excellent tools to improve or select accessions of interest [Akbar et al. 2003]. There were distinct morphological differences among all four spe-_____________________________________________________________________________________________________________________________________________ Hortorum Cultus 15 (5) 2016 cies which allowed them to be clearly distinguished based on morphological characteristics alone, as well as on PCA. There is a need to compare Cyclamen species from other Mediterranean countries where they grow naturally, and to compare them with species from Turkey. Additionally, other DNA loci as well as isozyme data need to be analyzed to clarify the taxonomic name and endemic status of populations distributed naturally in Turkey. The use of wild genetic resources to expand the genetic base of Cyclamen, and their preservation in vitro, is still a relatively unexplored topic in Cyclamen biology [Kocak et al. 2014].