Introduction

Perilla frutescens Britt. (Labiatae) is a self-fertilizing crop that is widely cultivated in East Asia. The Perilla plant occurs in two distinct varieties based on its morphology and specific uses: P. frutescens var. frutescens, an oil crop (Deul-ggae in Korean and Egoma in Japanese), and P. frutescens var. crispa, a Chinese medicine or spicy vegetable crop (Cha-jo-ki in Korean and Shiso in Japanese) (Nitta and Ohnishi 1999; Lee and Ohnishi 2001; Lee et al. 2002). Today, the two cultivated types of P. frutescens are extensively cultivated and used in Korea and Japan, although Perilla probably originated from China (Li 1969; Makino 1961; Nitta 2001; Lee and Ohnishi 2003). In Korea, var. frutescens is extensively cultivated as both an oil crop and a vegetable crop. Recently, with increases in meat consumption and the development of various methods of cooking fresh leaves and seeds, var. frutescens has become one of the most important crops in Korea. Conversely, var. crispa is not currently being cultivated due to the decreasing use of its seeds in Chinese medicine, although it is occasionally found in a relict form in Korea (Lee and Ohnishi 2001; Lee et al. 2003, 2007). However, in Japan, var. crispa is extensively cultivated and used as a spicy vegetable crop, whereas the cultivation of var. frutescens has declined to a small scale in mountain areas as a relict form (Nitta 2001). Despite the geographic proximity between Korea and Japan, methods for the cultivation and utilization of Perilla differ profoundly between the two countries (Lee and Ohnishi 2001; Lee et al. 2003; Nitta et al. 2003).

To ensure the success of breeding programs and maximize the use of genetic resources in Korea and Japan, the genetic diversity and relationships among cultivated types of Perilla and their weedy types must be studied. In addition, knowledge of the genetic diversity and population structure of Perilla germplasm collections is an important foundation for crop improvement in Korea and Japan. Recently, DNA-based molecular markers have provided useful information regarding genetic diversity and genetic relationships between cultivated species and their wild relatives. Among the various types of DNA-based markers, SSR markers are the preferred choice for genetic studies because they are highly reproducible, polymorphic, generally codominant and abundant in plant genomes (Powell et al. 1996; Park et al. 2009). In the previous study, Perilla SSRs were respectively developed by Kwon et al. (2005) and Park et al. (2008) and used to analyze Perilla accessions from Korea and East Asia (Lee et al. 2007; Lee and Kim 2007; Park et al. 2008).

On the other hand, weedy plants have recently been identified for two cultivated types of P. frutescens by Nitta and Ohnishi (1999), Lee and Ohnishi (2001, 2003), Lee et al. (2002), and Nitta et al. (2003). Morphological, RAPD, AFLP and SSR analyses demonstrated that these weedy plants can be also grouped into two types: one belonging to the group of var. frutescens and another belonging to the group of var. crispa (Nitta and Ohnishi 1999; Lee and Ohnishi 2001, 2003; Lee et al. 2002, 2007). Recently, Sa et al. (2012) evaluated the morphological characteristics of 54 accessions of cultivated and the weedy types of Perilla species, and reported that the cultivated type of var. frutescens might be regarded as a more domesticated type than the cultivated type of var. crispa. However, by this time, no concrete evidence of the origin of the weedy plants was presented in these analyses.

Therefore, in this study, we used all 18 Perilla SSR primers and the five morphological characteristics in order to determine the genetic diversity and genetic relationships between and within the cultivated and weedy types of Perilla species from Korea and Japan.

Materials and methods

Plant materials and DNA extraction

The materials use herein consisted of 56 accessions (23 of cultivated var. frutescens, 6 of cultivated var. crispa, 17 of the weedy type of var. frutescens and 10 of the weedy type of var. crispa) collected in Korea and Japan (Table 1; Fig. 1). The subset of each collections was deposited in the National Agrobio Diversity Center, Rural Development and Administration, Suwon, Republic of Korea, for permanent seed preservation. In this study, we tentatively classified the Perilla samples as the cultivated type or the weedy type based on a previous study conducted by Lee and Ohnishi (2001). Total DNA was extracted from the leaf tissues of a representative individual plant for each accession following the Plant DNAzol Reagent protocols (GibcoBRL Inc., Grand Island, NY, USA).

Table 1 Accessions of cultivated and weedy types of Perilla crop and their morphological characteristics
Full size table
Fig. 1
figure 1

Collection sites of 56 Perilla accessions in Korea and Japan. See Table 1 for the code numbers. empty circle Cultivated type of var. frutescens, filled circle Weedy type of var. frutescens, empty square cultivated type of var. crispa, filled square weedy type of var. crispa

Full size image

Morphological characteristics analysis

To assess the morphological variations in the cultivated types of Perilla crop and their weedy types, seven individuals of each accession were grown in a field at the Faculty of Agriculture, Kangwon National University, Chun-Cheon, Kangwon Prefecture. Approximately 25 seeds of each accession were sown in a nursery bed in early May and maintained in a glass house for a month. Seven seedlings of each accession were then transplanted into the field in early June. In our study, we examined five morphological characteristics, which were reportedly useful for discrimination between cultivated Perilla and weedy-type Perilla (Lee and Ohnishi 2001). The plants were evaluated at the appropriate growth stages for each accession as described in detail in Table 2.

Table 2 Characters used in the morphological analyses in cultivated and weedy types of Perilla crop
Full size table

SSR analysis and silver-staining

SSR amplifications were conducted in a total volume of 20 μl consisting of 20 ng genomic DNA, 1× PCR buffer, 0.5μM of forward and reverse primers, 0.2 mM dNTPs, and 1 U of Taq polymerase (Biotools, Spain). The PCR profile consisted of initial denaturation at 95 °C for 3 min, followed by 36 cycles of 95 °C for 30 s, 55 °C for 30 s, and 72 °C for 1 min 30 s, with a final extension step of 5 min at 72 °C. After PCR, 5 μl of the final product were mixed with 10 μl of electrophoresis loading-buffer (98 % formamide, 0.02 % BPH, 0.02 % Xylene C, and 5 mM of NaOH). After denaturing and immediate cooling, 2 μl of the sample were loaded onto 6 % denaturing (7.5 M urea) acrylamide–bisacrylamide gel (19:1) in 1× TBE buffer and then electrophoresed at 1,800 volts and 60 watts for 130 min. The separated fragments were then visualized using a silver-staining kit (Promega, USA).

Data analysis

Fragments amplified by SSR primers were scored as present (1) or absent (0), and the genetic diversity was then calculated for each group of accessions according to the formula developed by Nei (1973):

$$ {\text{Genetic diversity}} = 1 - \sum P_{i}^{2} , $$

where P i is the frequency of the ith SSR allele present in the group of accessions. Anderson et al. (1993) measured genetic diversity in terms of polymorphic information content (PIC). The genetic similarities (GS) were calculated for each pair of accessions using the Dice similarity index (Dice 1945). The similarity matrix was used to construct an UPGMA (un-weighted pair group methods using arithmetic averages algorithm) dendrogram with the help of SAHN-Clustering by NTSYS-pc.V.2.1 (Rohlf 2000). In addition, a principal coordinate analysis (PCO) based on the genetic similarity matrices was also carried out to estimate relationships among Perilla accessions using principal component analysis programs of the NTSYS-pc soft ware package (Rohlf 2000).

Results

Genetic variation in the cultivated and weedy-type P. frutescens using morphological and SSR markers

The measurements of five morphological characteristics of 56 accessions of cultivated types of P. frutescens and their weedy types collected from Korea and Japan are summarized in Table 1. Our results indicate that the two cultivated types of P. frutescens from Korea and Japan can be distinguished by several morphological features. Specifically, most accessions of the cultivated var. frutescens from Korea and Japan have a large seed size (>2 mm), soft seeds, four different seed colors (white, gray, brown and dark brown), and green leaves. Comparatively, most accessions of the weedy var. frutescens from Korea and Japan have small seed size (<2 mm), hard seeds that are dark brown in color, and green leaves. However, several accessions of cultivated var. frutescens and its weedy type showed exceptionally different seed characteristics. Namely, three accessions (K015923, K015813, K015869) of cultivated var. frutescens from Korea and two accessions (K015967, K015968) of weedy var. frutescens from Japan have hard seeds and large seed size (>2 mm) (Table 1). Conversely, all accessions of the cultivated var. crispa and its weedy type from Korea and Japan have small seed size (<2 mm), only hard seeds that are brown or dark brown in color, and either purple or green leaves (Table 1).

In SSR analysis, 18 SSR loci were used to evaluate the genetic diversity and genetic relationships on the 56 accessions of the two cultivated types of P. frutescens and their weedy types. A total of 165 alleles were detected segregating in the 56 Perilla accessions, with an average of 9.2 alleles per locus. The number of alleles per locus varied widely from two at KWPE-56 and KWPE-39, to 21 at GBPFM-204. The genetic diversity of each locus ranged from 0.497 at KWPE-56 and KWPE-39 to 0.959 at GBPFM-204, with an average of 0.692 (Table 3). Upon analysis of the four groups of accessions (cultivated and weedy types of var. frutescens and of var. crispa) with the 18 SSRs, the average numbers of alleles were 5.4, 5.7, 2.9 and 3.6 for the cultivated and weedy types of var. frutescens, and cultivated and weedy types of var. crispa, respectively (Table 4). In addition, the average genetic diversity values were 0.549, 0.685, 0.451 and 0.557 for the cultivated and weedy types of var. frutescens and the cultivated and weedy types of var. crispa, respectively (Table 4). Although the number of samples was not large, the results obtained herein implied that the weedy types of var. frutescens and var. crispa harbored comparatively higher numbers of alleles and relatively higher genetic diversity than the cultivated types (Table 4).

Table 3 Characteristics of the 18 SSR loci including repeat motif, allele size range, allele numbers and gene diversity among 56 accessions of cultivated and weedy types of Perilla crop in Korea and Japan
Full size table
Table 4 Estimates of genetic diversity and allele numbers of 18 SSR loci among cultivated and weedy types of Perilla crop in Korea and Japan
Full size table

On the other hand, in our study, among the 165 alleles which were detected from 18 SSR loci, 12 allele bands showed specific allele band patterns in accessions of cultivated and weedy types of P. frutescens. As shown in Fig. 2, SSR primer KWPE-58 revealed 6 alleles among the 56 Perilla accessions and that the cultivated and weedy types of P. frutescens, except for several accessions, had specific alleles. The allele designated by the arrow (on the left) was found in most accessions of cultivated and weedy types of var. frutescens, except for several accessions of cultivated and weedy types of var. frutescens, whereas another allele (designated by arrow on the right) was present in most accessions of cultivated and weedy types of var. crispa and one accession of weedy var. frutescens, except for three accessions of cultivated and weedy types of var. crispa (see Fig. 2a). Thus, these SSR markers may be considered useful molecular markers for distinguishing between the two cultivated Perilla types. Whereas, among the 18 SSR loci, three SSR loci—KWPE-53, GBM-203 and GBM-204—produced high numbers of allele bands (18, 20, 21), and also evidenced duplication band patterns in all accessions of Perilla (see Fig. 2b). Thus, the 18 SSR loci in our study were found to be highly informative molecular markers for characterizing and evaluating the genetic diversity of the two cultivated types of P. frutescens and their weedy types.

Fig. 2
figure 2

An example of an SSR profile of 56 Perilla accessions collected from Korea and Japan, using the SSR primers KWPE-58 (a) and GBPFM 203 (b). The accession numbers of Perilla germplasm are shown in Table 1. Among the two arrows of Fig. 1a, the arrow of the left side indicated specific molecular marker in cultivated and weedy types of var. frutescens, while the arrow of right side showed specific molecular marker in cultivated and weedy types of var. crispa

Full size image

Genetic relationships between the cultivated and weedy types of P. frutescens in Korea and Japan

The phylogenetic tree constructed using UPGMA revealed that the 56 accessions cluster into two major groups with 23 % genetic similarity (Fig. 3). Group I contained 43 accessions that consisted of 23 of the cultivated type of var. frutescens, 17 of the weedy type of var. frutescens, one of the cultivated type of var. crispa, and two of the weedy type of var. crispa. Group II contained 13 accessions, five of the cultivated type of var. crispa and eight of the weedy type of var. crispa. In addition, the group I accessions were further subdivided into three sub-clusters with 32 % genetic similarity. The first sub-cluster contained 14 accessions of the cultivated type of var. frutescens, 13 accessions of the weedy type of var. frutescens and two accessions of the weedy type of var. crispa from Korea and Japan. The second sub-cluster contained nine accessions of the cultivated type of var. frutescens and two accessions of the weedy type of var. frutescens from Korea. The third sub-cluster contained two accessions of the weedy type of var. frutescens from Korea and one accession of the cultivated type of var. crispa from Japan. The group II accessions were further subdivided into two sub-clusters with a 49.5 % genetic similarity. The first sub-cluster contained five accessions of the cultivated type of var. crispa and six accessions of the weedy type of var. crispa from Korea and Japan. The second sub-cluster contained only two accessions (K015873, K015874) of the weedy type of var. crispa from Korea. Taken together, the clustering patterns observed in the present study permitted a clear distinction between two cultivated types of var. frutescens and var. crispa and their related weedy types, with only a few exceptions.

Fig. 3
figure 3

UPGMA dendrogram based on SSR markers. The accessions of the Perilla germplasm are shown in Table 1. empty circle cultivated type of var. frutescens, filled circle weedy type of var. frutescens, empty square cultivated type of var. crispa, filled square weedy type of var. crispa

Full size image

Principal coordinates (PCO) analysis

The results of PCO analysis indicated that the first and second principal components accounted for 14.29 and 8.96 % of the total variation (Fig. 4), respectively. The contributions of the remaining principal components were less than 6.18 % each. The scatterplot of the first two principal components is presented in Fig. 4. In PCO analysis, most accessions of cultivated types of var. frutescens and var. crispa were clearly separated, that is, all accessions of cultivated var. frutescens occupied the negative side, whereas most accessions of cultivated and weedy types of var. crispa were situated on the positive side. Additionally, most accessions of cultivated and weedy types of var. frutescens, including one accession (K015964) of cultivated var. crispa and two accessions (K015810, K015917) of weedy var. crispa, were closely positioned to each other. As a result, the scattering pattern in Fig. 4 agrees closely with the clustering observed on the dendrogram (Fig. 3).

Fig. 4
figure 4

Scatter plot of principal coordinate analysis of accessions of cultivated types of Perilla crop and their weedy types in the first and second principal components, using SSR data. filled diamond cultivated type of var. frutescens, filled square weedy type of var. frutescens, filled circle cultivated type of var. crispa, filled triangle weedy type of var. crispa

Full size image

Discussion

Morphological and SSR variations among accessions of cultivated and weedy types of P. frutescens

In East Asia, two cultivated types of P. frutescens are extensively cultivated and used in Korea and Japan (see “Introduction”). Although Korea and Japan are geographically close, the cultivation and utilization of the two cultivated types of Perilla differ greatly between these two countries. Therefore, understanding the genetic diversity among Perilla germplasm accessions collected from Korea and Japan is essential for the long-term success of breeding programs and maximization of the use of the germplasm resources. According to the results of the present study, Korean accessions of cultivated var. frutescens showed white, gray, brown and dark brown seed colors and had both soft and hard seeds, whereas Japanese accessions had only soft seeds that were white and brown. As a result, accessions of cultivated var. frutescens from Korea showed more morphological variations than those of Japan, most notably with regard to seed hardness and seed color (Table 1). In this study, with a few exceptions, seed size and seed hardness were found to be useful morphological characteristics for discrimination between cultivated and weedy types of var. frutescens, as well as between cultivated var. frutescens and var. crispa; this is consistent with the results reported by Lee and Ohnishi (2001). Conversely, in the case of cultivated and weedy types of var. crispa, most Korean and Japanese accessions showed similar morphological variations in seed and leaf colors (Table 1). As previously reported by Lee and Ohnishi (2001), in the present study, most accessions of cultivated and weedy types of var. crispa could not be distinguished based on these morphological characteristics.

On the other hand, SSRs are useful markers of genetic diversity, since the hyper-variable nature of SSRs results in the production of high levels of allelic variations, even among very closely related varieties (Park et al. 2009). In this study, we demonstrated the successful application of SSR analysis in a study of the genetic diversity among accessions of cultivated types of Perilla crop and their weedy types from Korea and Japan. According to our results, a total of 165 alleles with 18 SSRs were detected segregating in the 56 Perilla accessions from Korea and Japan, which yielded an average of 9.2 alleles per locus (Table 3). This value appears to be high when compared to the effective number of alleles per SSR locus in other major crops, such as the 7.4 detected in wheat (Prasad et al. 2000), 6.8 in rice (Ni et al. 2002) and 3.6 in barley (Hamza et al. 2004). Thus, the high number of alleles in our study reflect the utility of Perilla SSR markers in determining unique genetic profiles of individual genotypes of Perilla crop that should prove useful in future genetic and breeding studies.

Conversely, the accessions of weedy types of var. frutescens and var. crispa exhibited comparatively higher polymorphisms and genetic diversity values than those of their cultivated types in Korea and Japan (Table 4), which is consistent with the results of a previous study by Lee et al. (2003) conducted using AFLP markers with a similar set of accessions of cultivated types of P. frutescens and their weedy types. These results may indicate that many alleles were lost through natural or human selection during the course of evolution from wild species to cultivated species within Perilla. In particular, domestication has reduced the levels of polymorphisms and genetic diversity in cultivated species via direct or indirect human selection and genetic drift during the course of adaptation from wild to cultivated variants. In addition, the cultivated and weedy types of the Perilla crop evidenced much higher polymorphisms and genetic diversity in Korea than in Japan (data not shown). Currently, var. frutescens is extensively cultivated, and many weedy samples of var. frutescens and var. crispa are found in Korea (Lee and Ohnishi 2001; Lee et al. 2002; Nitta 2001; Nitta et al. 2003). These results suggest that Korea may be a secondary center of cultivated and weedy types of Perilla crop in East Asia, as proposed in previous studies conducted by Lee et al. (2002) and Lee and Ohnishi (2003). The SSR markers of the Perilla crop developed herein provide valuable information to help expand our understanding of the genetic diversity in Perilla species from Korea and Japan.

Genetic relationships of cultivated and weedy types of P. frutescens

The phylogenetic tree constructed using UPGMA revealed 56 accessions clustered into two major groups. The results indicated that most accessions of cultivated types of var. frutescens and of var. crispa were sharply separated by SSRs, which was expected based on the distinct morphological characteristics of the two cultivated types of P. frutescens (Lee and Ohnishi 2001). Additionally, most accessions of weedy types of var. frutescens and of var. crispa could be grouped into two types, one belonging to the group of var. frutescens and another belonging to the group of var. crispa, respectively. However, as is shown in Figs. 3 and 4, several accessions of cultivated and weedy types of Perilla crop were ambiguous in terms of their classification on the basis of the aforementioned SSR markers. In previous studies of Perilla, weedy plants were identified for two cultivated types of var. frutescens and of var. crispa by Nitta and Ohnishi (1999), Lee et al. (2002) and Lee and Ohnishi (2001, 2003). Nitta and Ohnishi (1999) suggested that the weedy type in Japan originated from hybrids of cultivated types of var. frutescens and var. crispa, or that it may be a form that has escaped from cultivation. Additionally, Lee et al. (2002) and Lee and Ohnishi (2003) suggested that the weedy types of P. frutescens in East Asia as well as in Korea are the key taxa to understanding the origin of the cultivated type of var. frutescens and of var. crispa. In the group I and group II clusters obtained in the present study, most accessions of the weedy type of var. frutescens and var. crispa were closely related to their corresponding cultivated types. These findings indicate that the two cultivated types of P. frutescens may have differentiated from their respective weedy types. Although the wild species of the two cultivated types of P. frutescens have not yet been identified, the results of the present study suggest that the weedy types of P. frutescens might be a true wild form of the cultivated type of var. frutescens and var. crispa. In addition, in the group I and group II clusters identified in the present study, as well as the results of PCO analysis shown in Fig. 3, several accessions of the weedy type of var. frutescens and var. crispa could not be clearly discriminated from their cultivated types of P. frutescens. For example, some accessions of the weedy type of var. frutescens and var. crispa of group I, such as K015965, K015915, K015810 and K015917, were placed in the cluster of the cultivated type of var. frutescens. In addition, three accessions (K015810, K015917, K015964) of cultivated and weedy types of var. crispa were placed in the cluster of cultivated and weedy types of var. frutescens (see Figs. 3, 4). These results indicated that some accessions of the weedy type of var. frutescens and var. crispa might be derived from either seeds escaped from cultivation or hybrids between cultivated and weedy types of P. frutescens, as previously proposed by Nitta and Ohnishi (1999), Lee et al. (2002), Lee and Ohnishi (2003) and Lee and Kim (2007). Therefore, further analyses will be necessary to clarify the taxonomic position of these exceptional weedy samples, since we have tentatively classified the weedy Perilla samples based on cultivation conditions and morphological characteristics (see Introduction). However, as shown in Fig. 3, most accessions of cultivated var. frutescens in group I could be subdivided further into two sub-clusters. The first sub-cluster contained 14 accessions of cultivated var. frutescens from Korea and Japan, and the second contained nine accessions of cultivated var. frutescens from Korea. The results of the present study indicate that the geographical locations of most accessions of cultivated var. frutescens in Korea and Japan are not related to their positions in the phylogenetic tree. These findings demonstrate that the cultivated var. frutescens might indeed have diffused from Korea to Japan, and that cultivated var. frutescens may be considered the most highly domesticated of the plants, since seed color, seed hardness and seed size were more variable in the accessions of cultivated var. frutescens than those of the weedy var. frutescens and both the cultivated and weedy types of var. crispa. Furthermore, the growth patterns of cultivated types of P. frutescens from Korea and Japan differ profoundly. Most notably, var. frutescens is cultivated by sowing seeds in nursery beds or directly in the field during spring in Korea, while seedlings of var. crispa that sprouted in the field from naturally shattered seeds during autumn of the previous year are transplanted into fields for cultivation in Japan. The naturally occurring volunteer seedlings of var. crispa can also be cultivated without transplantation (Nitta et al. 2003).

One question, however, remains to be answered. Is var. crispa a domesticated form or not? According to the seed characteristics of Perilla crop, the weedy type of var. frutescens and cultivated and weedy types of var. crispa have small and hard seeds (less than 2 mm) with seed dormancy. Conversely, most accessions of the cultivated var. frutescens have large, soft seeds (larger than 2 mm) which are free from seed dormancy. Seed dormancy is considered to be a characteristic strongly selected against during domestication, and hence it should not appear in cultivated crops (Hancock 1992). In addition, in the SSR analysis, most accessions of the cultivated and weedy types of var. crispa could not be clearly discriminated from each other. With the results of morphological characteristics and the SSR data, like as the previous report by Sa et al. (2012), we proposed that the cultivated type of var. frutescens may be sufficiently differentiated from the weedy type of var. frutescens, whereas var. crispa may not be yet sufficiently differentiable from the weedy type of var. crispa.

This study demonstrates the efficacy and utility of SSR analysis in the study of genetic diversity of two cultivated types of P. frutescens and their weedy types. The genetic diversity and genetic relationships among genetic resources of cultivated types of P. frutescens and their weedy types by SSR analysis is also expected to prove helpful to the Perilla crop breeding programs of Korea and Japan. The results of SSR marker analysis also help to clarify the genetic relationships between var. frutescens and var. crispa and have improved our understanding of the differentiation of the two cultivated types of P. frutescens and their weedy types in Korea and Japan.