Cyclamen sp. are part of the Primulaceae family, being widely cultivated throughout the Mediterranean area as an ornamental plant, but also for its pharmaceutical properties. Of the 24 known species, Cyclamen persicum Mill. is distributed from south-central Turkey to Lebanon-Syria, being the most significant in terms of production, because of its adaptability to new ecological conditions. C. persicum is a much admired ornamental plant of high economic importance, especially in the Netherlands, Germany and Italy. Cyclamen is particularly essential in traditional and modern medicine, in addition to its decorative and commercial importance (Sarikurkcu, 2011; Turan and Mammadov, 2018; Cornea-Cipcigan et al., 2019). Cyclamen possesses cultural and religious relevance, symbolizing empathy and devotion in the Mediterranean culture, is cultivated in Islamic churchyards and monasteries, and is a consecrated flower in Japan (Grey-Wilson, 2015). In addition to their ornamental importance, Cyclamen sp. also possesses strong antioxidant properties due to the phenolic composition, and anti-cancer activities reported in HeLa, non-small cell lung cancer H1299 cells, human colorectal cancer cells (HCT 116 and HT-29), human breast adenocarcinoma (MDA-MB-231) and human fibroblasts BJ cells (Mihci-Gaidi et al., 2010a; Mihci-Gaidi et al., 2010b; Yildiz et al., 2013; Cornea-Cipcigan et al., 2022a). Due to a shortage of plant material, their use for the production of pharmaceuticals has become a severe problem, potentially leading to the loss of plant populations and variety, natural habitat deterioration, and/or species extinction. Classification of morphological traits has been done in only a few Cyclamen species endemic to Turkey (Curuk et al., 2015; Curuk et al., 2016).

The materials preserved in gene banks, such as wild, cultivated and selected accessions represent important sources of variability for breeders. Thus, breeders have used a variety of approaches throughout the years to investigate and assess the level of diversity in plant populations (Bhandari et al., 2017). In Cyclamen, genetic variation has been evaluated through morphological characters and molecular markers, such as the internal transcribed spacer regions (ITS) (Anderberg et al., 2000), random amplified polymorphic DNAs (RAPD) (Naderi et al., 2009; Taşkin et al., 2012), and sequence-related amplified polymorphisms (SRAP) (Simsek et al., 2017) (Robarts and Wolfe, 2014; Yagi et al., 2014; Samarina et al., 2021).

As the availability of specific molecular markers for Cyclamen is limited, SRAP markers represented the basis of the molecular investigations performed in this study. Also the PCR-based start codon targeted (SCoT) markers which were only used previously on other species were included in the analysis. Both types of markers constitute simple, reproducible and low cost techniques for characterizing germplasm collections (Collard and Mackill, 2008), with the aim of genetic diversity screening (Igwe et al., 2017), species identification or phylogenetic analysis (Jalilian et al., 2018).

Although molecular markers have been extensively used to evaluate species variety, extremely low or insignificant correlations have been observed among dissimilarity matrices generated using both phenotypic and molecular data (Gupta et al., 2018). As a result, if the non-overlapping data is derived from phenotypic and genotypic divergence matrix, combining them may offer a full picture of a population’s variety (Houmanat et al., 2021). Multivariate analysis approaches (HCA or PCA), are commonly used to precisely classify various plants based on their agro-morphological, molecular, chemical composition, or bioactivities that are regularly compared with correlation coefficients (Granato et al., 2018). To accurately evaluate the genetic variation among individuals, it is important to choose a suitable dissimilarity coefficient and hierarchical clustering method since both have an impact on the outcomes of genetic diversity analysis (Darkwa et al., 2020). Using both agronomic variables and molecular markers may contribute to a more comprehensive genetic diversity analysis within and among species, phylogenetic investigations, and fingerprinting in various plants (Jalilian et al., 2018; Long et al., 2020). Consequently, to evaluate the genetic diversity of Cyclamen genotypes, joint hierarchical cluster analysis was constructed between phenotypic and molecular data. Choosing the most suitable clustering method is significant to determine proper genetic dissimilarity and diversity between and within populations and clustering, since different correlations may give inconclusive or dissimilar results. To best of our knowledge, comparison between clustering methods based on phenotypic and molecular markers (SRAP and SCoT markers) has not been employed in Cyclamen until now. Our studies objectives were (1) the comparison of several correlation matrices and HCA for investigating genetic diversity in Cyclamen, and (2) the evaluation of genetic diversity and divergence in Cyclamen genotypes by the use of morphological, molecular, and combined data.

Prior research has found considerable morphological differences between closely related accessions or populations, due to the climatic conditions of various habitats, and lower heredity of reproductive and vegetative traits in Cyclamen (Yesson and Culham, 2011; Cornea-Cipcigan et al., 2022b). In the present study, PCA was performed to assess the similarities and differences between genotypes using morpho-agronomic characters. PCA generally supported the assemblage of phenotypic traits into species-specific clusters, as seen by C. persicum genotypes that mainly grouped in the 2^nd and 3^rd quadrants. Significant variables composing PC1 were NS, PL, FA, LN, CA, and BCRD as seen by the highest eigenvalues, whereas PC2 accounted for the color characteristics of flowers ( Supplementary Table S3 ). Using the 36 morphological parameters from the first ten PCs, the UPGMA grouped the genotypes based on their canopy architecture and plant vigor in cluster I, lower canopy area and absence of leaf pattern in cluster II. Highest canopy area and similar flower area dimensions were grouped in cluster III, whereas the IV^th cluster comprised paler pink colored genotypes with lower flower number, and strong degree of lobbing. These are in accordance with morphological descriptions of different Cyclamen species, as the most distinctive variables were found to be quantitative with the highest eigenvalues in the first PCs (Curuk et al., 2015). Several researchers claim that morphological characters are less effective for assessing genetic diversity owing to the shift in environmental conditions and plant growth development; HCA groups the samples mainly based on their morphological characters rather than geographical origin (Shinwari et al., 2014; Curuk et al., 2015; Curuk et al., 2016). Nonetheless, morphological and agronomic characterization is significant for describing plant populations and for selecting varieties with favorable traits for crop development that is of great interest for plant breeders. Other classifications, such as chromosome number, duration of green period, early or late blossoming period and flowering habit, disease spot detection and fruit color were successfully assessed in different ornamental and medicinal crops that may demonstrate more complex distribution classification (Chaudhary et al., 2012; Cui et al., 2019; Lu et al., 2021).

A phenotypic trait that often reflects the physiological status of the plant is the color of the flowers and leaves (Noman et al., 2017; Cui et al., 2019). The concentration of secondary metabolites with considerable chromatic elements, that may have decorative, medicinal, and economic relevance, is indicated by the color of the flowers and leaves (Cornea-Cipcigan et al., 2022a). Color variation also gives data on the genetic inter- and intraspecific diversity of populations and/or species, recognizing the progress of evolutionary principles from a phylogenetic standpoint (Tanaka et al., 2008; Klančnik et al., 2016). Color parameters were assessed to evaluate the color of Cyclamen flowers and leaves that may aid in differentiating similar varieties or species. Genotypes that presented higher values in L* grouped in the same cluster (cluster II, Figure 3A ). Furthermore, distinct color characteristics were observed in dark red and grey-white genotypes that grouped in the same cluster (III and IV, Figure 3A ). Regarding the leaves, C. mirabile ‘Tilebarn Nicholas’ (C22) is considered an ‘outlier’ due to its purple-green leaf pattern. Genotypes C. persicum ‘Petticoat pure white’ (10) and C. cilicium (25) presented similar L*, b* and C* in the dark and silver patterns (cluster I, Figure 3B ). C. persicum genotypes presented similarities in a* and hue in dark lamina and b* and C* in silver pattern, whereas the following cluster comprised genotypes with similar a* (darker green lamina), C* and shade in leaves, together with similarities in terms of petiole hue (II and III, Figure 3B ). The genotypes with similar petiole color characteristics grouped in cluster IV. Similar findings were observed in C. purpurascens, where semi-silvery and patterned leaves had higher L* values compared with green and silver lamina. According to the study’s findings, all genotypes showed positive b* values, indicating that the leaves had a faint yellow tint (Osterc et al., 2014; Cornea-Cipcigan et al., 2019). Additionally, higher hue levels were noted in C. purpurascens flowers and lower values in the leaves (Osterc et al., 2018). Consistent with previous reports, the values of b* were significantly lower in dark-colored flowers and higher in leaves (Cui et al., 2019).

Genetic diversity analysis using the Euclidean distance generated with the SRAP (r = 0.95) and SCoT (r= 0.82) markers partitioned the Cyclamen genotypes into four groups. This study firstly evaluated the effectiveness of SCoT markers to determine genetic variation between Cyclamen genotypes. SCoT markers have been chosen based on their successful use for analysis of genetic diversity and cultivar identification in multiple plants. A relatively high percentage of polymorphic bands (82%) was detected, consistent with the proportion of polymorphism reported in other medicinal and ornamental species, such as sage (Etminan et al., 2018), chrysanthemum (Feng et al., 2016) and orchids (Tikendra et al., 2021). In the SRAP analysis, a high percentage of polymorphic bands (75%) and a clear discrimination between genotypes was observed, as shown by the high cophenetic correlation coefficient (r=0.95). SRAP primers have been chosen due to their high reproducibility compared to RAPDs and ease of assay performance than AFLPs. Previously, UPGMA dendrograms were constructed to assess the similarities and differences between Cyclamen species form different regions of Turkey showing a clear discrimination between C. persicum genotypes and other species. Furthermore, C. pseudibericum and C. alpinum genotypes clustered in different groups based on their geographic origin, demonstrating their narrow genetic diversity (Taşkin et al., 2012; Simsek et al., 2017). Consistent with previous studies on SRAP markers (Simsek et al., 2017), C. hederifolium and C. persium clustered in the same group ( Figure 6A ). Furthermore, C. intaminatum and C. cilicium grouped in the same sub-cluster, while C. alpinum was closely followed by C. coum, possibly due to the same chromosome number (2n = 30). These genotypes are genetically related as already reported (Compton et al., 2004). The combined SRAP and SCoT analysis revealed a more accurate genetic estimate between the genotypes. For example, the 2^nd cluster organized the genotypes that mainly correspond to subgenera Gyrophoebe with few exceptions, whereas 3^rd cluster comprised genotypes corresponding to Cyclamen subgenus. This is in accordance to previous studies that revealed a similar phylogeny using cluster analysis (Anderberg et al., 2000; Compton et al., 2004).

Plants that present similarities in terms of agro-morphological characters may have significantly different molecular characteristics, and vice versa. Discrepancies in genotypic and phenotypic data revealed in clusters might be due to the effects of environment-genotype interaction often noticed in quantitative inherited characteristics. Understanding the nature and degree of genetic variety in crops requires non-overlapping and complementary morphological and molecular data that are highlighted by the absence of relationship between the diversity matrices (grey lines). Several researches investigated the discrepancies between phenotypic and genotypic characteristics, such as maize (Hartings et al., 2008), wheat (Soriano et al., 2016), sweet sorghum (Da Silva et al., 2017), kalanchoe (Al-Khayri et al., 2022), sesame (Stavridou et al., 2021). An entanglement rate of 0.65 was obtained among 77 passion fruit genotypes showing dissimilarities between the molecular and agronomic data (Preisigke et al., 2020). The same divergence in sample distribution in dendrograms was identified in 65 hot pepper accessions using genetic and agronomic data (Moreira et al., 2018). The dendrogram’s irregular genotype ordering might be explained by the lack of phenotypic trait specificity of the used molecular markers. They amplify more stable parts of the genome but are unable to relate a genetic marker to a particular phenotypic characteristic.

The present study assessed the variation between clustering methods based on phenotypic and molecular markers (SRAP and SCoT markers) which have not been employed in Cyclamen. Several HCA for evaluating genetic diversity in Cyclamen were compared, followed by combined analysis using morphological and molecular data. With a few outliers for morphological data (r = 0.73), high cophenetic correlation coefficients were noticed for the others. UPGMA method was demonstrated to generate the highest coefficient values for the majority of matrices between morphological, colorimetric, and molecular data, implying that these matrices and distances are well represented as dendrograms. In line with the current study’s findings compared to other methods, the UPGMA clustering approach was shown to provide significant correlation coefficients for genetic diversity investigations in safflower (Houmanat et al., 2021), yam (Granato et al., 2018), sweet potato (Paliwal et al., 2022), and gladiolus (Singh et al., 2018).

To evaluate genetic diversity in plant populations, a method that creates a single matrix from the phenotypic and genotypic dissimilarity matrices was presented (Da Silva et al., 2017). Combined analysis of morphological and molecular data particularly with SRAP markers, showed the highest performance and similarities, whereas SCoT markers and color characteristics demonstrated moderate similarities. For a better classification of genotypes, more complex and significant correlations may be revealed using molecular approaches. However, evaluation of diversity using only molecular markers is insufficient. As a result, the molecular assessment of genetic diversity may adequately describe genetic variability, when combined with phenotypic variance. In Cyclamen, genetic biology advances have permitted the development of correlations between molecular markers and morphological characteristics (Yesson and Culham, 2006; Naderi et al., 2009). In this aspect, the combination of agronomic parameters, color characteristics and molecular markers is efficient for assessing genetic diversity across and among species with greater precision, taxonomic study, and genotyping.

Color indicators of leaves, petioles, flowers and pedicel were measured separately for color phenotypic data. The present study also attempted to establish a more precise characterization of leaf and flower color in order to be used as a framework for Cyclamen breeding. Furthermore, this study firstly evaluated the potential of SCoT markers for analyzing genetic diversity among Cyclamen genotypes. SRAP markers proved to be useful tools for the separation of Cyclamen species and/or genotypes, particularly when combined with phenotypic data, as seen by the high entanglement analysis (0.30) among dendrograms. Conversely, similarities were noticed in the colorimetric data for leaves and SCoT data (entanglement=0.47), compared with the colorimetric analysis of flowers and SCoT markers (entanglement=0.45). Thus, combined HCA of phenotypic and molecular data, particularly with SRAP markers, showed the highest performance and similarities, whereas SCoT markers and color characteristics demonstrated moderate similarities. As a result, we underline that SCoT markers along with color characters are suggested for the characterization of germplasm banks with a large number of genotypes in order to detect duplicate accessions since they are more efficient and economically feasible. Conversely, phenotypic data along with SRAP markers more accurately describe in terms of genotype evaluation for later use in breeding programs.

The original contributions presented in the study are included in the article/ Supplementary Material . Further inquiries can be directed to the corresponding authors.

MC-C, RM and CS: conceptualization. CS and RM: methodology. MC-C: software. MC-C: formal analysis. MC-C and CS: investigation. CS and RM: resources. MC-C, CS and RM: writing—original draft preparation. MC-C, CS, RM and DP: writing—review and editing. DP: supervision. RM and CS: funding acquisition. All authors contributed to the article and approved the submitted version.