Introduction
Perilla (2n = 40) is a self-fertilizing annual plant belonging to the Lamiaceae family and is a crop primarily used in East Asia for its seeds and leaves. Taxonomically, this species is divided into two varieties, Perilla frutescens var. frutescens and P. frutescens var. crispa, based on their morphological characteristics and uses (Makino, Reference Makino1961; Lee and Ohnishi, Reference Lee and Ohnishi2001). As well as being cultivated in South Korea, the two varieties (or cultivated types) of Perilla are also cultivated in other areas of East and Southeast Asia, especially in Japan and China (Li, Reference Li1969; Lee and Ohnishi, Reference Lee and Ohnishi2001, Reference Lee and Ohnishi2003; Nitta et al., Reference Nitta, Lee and Ohnishi2003). P. frutescens var. frutescens is called ‘Ren’ in China, ‘Dlggae’ in South Korea and ‘Egoma’ in Japan, while P. frutescens var. crispa is called ‘Zisu’ in China, ‘Cha-jo-ki’ or ‘Jaso’ in South Korea and ‘Shiso’ in Japan (Lee and Ohnishi, Reference Lee and Ohnishi2001, Reference Lee and Ohnishi2003; Nitta et al., Reference Nitta, Lee and Ohnishi2003). In South Korea, P. frutescens var. frutescens, an oil crop and vegetable crop, is widely used as a wrap vegetable using the leaves and as a seed oil, which contains many unsaturated fatty acids (linoleic acid, linolenic acid, oleic acid, etc.) and provides health benefits (Lee and Ohnishi, Reference Lee and Ohnishi2001, Reference Lee and Ohnishi2003; Park et al., Reference Park, Sa, Hyun, Lee and Lee2021). Moreover, the leaves of P. frutescens var. frutescens are abundant in perilla ketone, which gives them their unique fragrance (Sa et al., Reference Sa, Jang, Lee, Park, Cho, Sung and Lee2023).
Recently, as the consumption of meat and raw fish has been increasing along with changes in eating habits, the consumption of P. frutescens var. frutescens as a fresh leafy vegetable (locally known as ‘kkaennip’) akin to lettuce has been on the rise. Furthermore, the seeds of P. frutescens var. frutescens are rich in functional ingredients such as omega-3 fatty acids, and their seed oils are gaining attention in South Korea as a health food (Park et al., Reference Park, Sa, Hyun, Lee and Lee2021; Sa et al., Reference Sa, Jang, Lee, Park, Cho, Sung and Lee2023). For this reason, P. frutescens var. frutescens is extensively cultivated and utilized in South Korea (Lee et al., Reference Lee, Nitta, Kim, Park, Yoon, Shin and Ohnishi2002; Cho et al., Reference Cho, Park, Heo, KJ and Lee2025).
Conversely, in P. frutescens var. crispa, the leaves have a high anthocyanin content and a high perilla aldehyde content, giving it a unique aroma, so it is used as a leafy vegetable, spice, pickle and medicinal plant in Japan and China (Lee and Ohnishi, Reference Lee and Ohnishi2001, Reference Lee and Ohnishi2003; Nitta et al., Reference Nitta, Lee and Ohnishi2003; Sa et al., Reference Sa, Jang, Lee, Park, Cho, Sung and Lee2023). However, people in South Korea dislike the aromatic scent of the P. frutescens var. crispa plant and rarely cultivated it. Currently, weedy plants of P. frutescens var. crispa grow on some roadsides and fields around farms (Lee and Ohnishi, Reference Lee and Ohnishi2001; Lee et al., Reference Lee, Nitta, Kim, Park, Yoon, Shin and Ohnishi2002; Sa et al., Reference Sa, Choi, Ueno, Park, Park, Ma and Lee2013).
Overall, the two cultivated types of Perilla have a rich history of cultivation and use in East Asia, where Perilla continues to be an important part of culinary traditions and herbal medicine practices. In South Korea, archaeological evidence for Perilla crop is not yet known, but in nearby Japan the two varieties of Perilla seeds have been found in archaeological sites (Kasahara, Reference Kasahara1981). Both types of Perilla seeds have been found in carbonized starch cakes dating back about 5,000 years and discovered in Nagano, Gifu and Fukushima prefectures in Japan (Matsutani, Reference Matsutani, Katoh, Kobayashi and Fujimoto1983). Nitta et al. (Reference Nitta, Lee and Ohnishi2003) reported that these seeds are thought to have been used to flavour foods, similar to ancient nuts and sesame seeds. Additionally, in East Asia, Perilla seed oil was also used in industrial processes such as in the manufacture of lacquer and paper umbrellas, while Perilla seed oil, rather than animal oils, was used for lamps in the past (Nitta et al., Reference Nitta, Lee and Ohnishi2003). Therefore, in East Asian countries, including South Korea, Perilla is thought to have been cultivated as a very important food and industrial crop since ancient times. In South Korea, Perilla has been a folk plant since ancient times and has been cultivated as an oil or medicinal crop in farms across the country. However, today, P. frutescens var. crispa is rarely cultivated because of the decline in its use for medicinal purposes, and only P. frutescens var. frutescens is cultivated on farms. Also the two weedy types of Perilla crop live around farms or fields in a natural state. Therefore, our research team has visited farms and habitats across the country every year for the past 20 years to collect genetic resources of landrace and weedy type accessions of Perilla crop. These collected resources were donated to the Genetic Resources Center of the Rural Development Administration in South Korea, which currently has more than 2,000 accessions.
For effective management of gathered genetic resources, an understanding of morphological traits is critical. Over a long period of time, humans have chosen and domesticated crops, resulting in a variety of physical traits that vary based on the cultivated location. Phenotypic characterization is becoming a more vital stage in plant breeding programmes as it allows breeders to assess variations and choose high-performance accessions effectively. Phenotypic characteristics, along with genetic diversity analysis, are essential for establishing core populations (Shadmehr et al., Reference Shadmehr, Ramshini, Zeinalabedini, Parvizi Almani, Ghaffari, Izadi Darbandi and Fooladvand2017); therefore, morphological analysis is important for the effective utilization and development of Perilla.
Meanwhile, South Korean environmental conditions can be divided into those of the Central region, Southern region and Jeju Island. Among these, the Central region is a border region between temperate and cold climates with the Yeongseo region of Gangwon-do and the northern region of Gyeonggi-do in particular mainly having a cold climate. The Southern region mainly has a temperate climate, but the Southern coast of the Southern region and Jeju Island have a completely temperate climate, with winters averaging between 3°C and 5°C. Due to these differences in climatic conditions, morphological or physiological characteristics may vary depending on the collection site and cultivation area. Despite climatic differences across collection sites in Korea, there has been limited physiological and morphological research on Perilla crop. Regionally adapted Perilla crops exhibit distinct physiological and morphological traits, making them valuable resources for cultivar development.
To conserve crop genetic resources and develop new varieties, it is essential to first collect and study them, and then evaluate and understand their genetic diversity. Therefore, this study was conducted to provide a better understanding of the morphological variation of cultivated and weedy types of P. frutescens var. frutescens and weedy type of P. frutescens var. crispa collected and preserved by the RDA-Genebank of the Republic of Korea. The morphological diversity of 500 Perilla accessions collected from various regions of South Korea was analysed using 7 qualitative traits and 10 quantitative traits. The results of this study are expected to offer valuable insights into morphological variations associated with different collection locations, thereby assisting Perilla breeding programmes and potentially facilitating the establishment of core collections for the long term conservation of Perilla genetic resources in the future.
Materials and methods
Plant material
For this study, in 2022 we received over 1,227 accessions of Perilla genetic resources preserved at the RDA-Genebank of the Republic of Korea (http://genebank.rda.go.kr) and selected 500 accessions based on their collection locations. Among these 500 Perilla accessions, 200 accessions were cultivated var. frutescens (CF), 200 accessions were weedy var. frutescens (WF) and 100 accessions were weedy var. crispa (WC), and each collection site was selected from a total of 96 sites in the Central region, Southern region and Jeju Island (Fig. 1, Supplementary Table 1). Of the 500 accessions of three types of Perilla, 232 accessions were selected from the Central region of Korea, 239 accessions were selected from the Southern region and 29 accessions were selected from Jeju Island. Among the 232 accessions in the Central region, 58 accessions were collected in Chungcheongbuk-do, 64 accessions in Chungcheongnam-do, 65 accessions in Gangwon-do and 45 accessions in Gyeonggi-do. Among the 239 accessions from the Southern region, 97 accessions were from Gyeongsangbuk-do, 41 accessions from Gyeongsangnam-do, 66 accessions from Jeollabuk-do and 35 accessions from Jeollanam-do. The average number of accessions used from each area in the Central and Southern regions, excluding Jeju Island, was 58.9 accessions. By region, in the Central region, there were 94 accessions of CF, 80 accessions of WF and 50 accessions of WC; and in the Southern region, there were 101, 92 and 46 accessions of CF, WF and WC, respectively. From Jeju Island, there were 5 accessions of CF, 20 accessions of WF and 4 accessions of WC.
Map of South Korea showing collection sites of 500 Perilla accessions of three types (refer to Supplementary Table 1 for accession numbers and remainder of the summary). Central region – Chungcheongbuk-do, Chungcheongnam-do, Gangwon-do, Gyeonggi-do. Southern region – Gyeongsangbuk-do, Gyeongsangnam-do, Jeollabuk-do, Jeollanam-do. CF: cultivated var. frutescens; WF: weedy var. frutescens; WC: weedy var. crispa.
Evaluation of morpho-agronomic traits
To investigate the morphological characteristics of native genetic resources of Perilla species collected in South Korea, we cultivated Perilla accessions using standard conventional methods at a farm of the College of Agriculture and Life Sciences at Kangwon National University, located in Chuncheon, Gangwon-do, South Korea, over 2 years (2022, 2023). Seeds from 500 Perilla accessions of three types were sown in mid-May, grown in a greenhouse for about a month, and then transplanted to the field in mid-June. In detail, 30 seeds from each of the 500 Perilla accessions were sown and grown in a greenhouse, and then 10 plants per accession were transplanted to the field in mid-June. One week before transplanting, a compound fertilizer (21% nitrogen, 17% phosphorus, 17% potassium) was applied to the soil. On the designated day, transplanting was carried out following sufficient irrigation. The spacing between accessions was maintained at 1 m, while individual plants within each accession were spaced 30 cm apart.
The morphological characteristics survey examined 7 qualitative traits and 10 quantitative traits, following the standard trait evaluation criteria of the RDA-Genebank of the Republic of Korea (Table 1, Supplementary Fig. 1). Six of the qualitative traits investigated (colour of leaf surface [QL1], colour of leaf reverse side [QL2], colour of stem [QL3], colour of flower [QL4], degree of pubescence [QL5], colour of seed [QL7]) were visually distinguished, and seed hardness (QL6) was measured by hand, as reported by Lee and Ohnishi (Reference Lee and Ohnishi2001). If a seed broke, it was classified as soft type; if it did not break, it was classified as hard type. The emergence period, flowering period and maturity period were recorded during the growth period, and eight quantitative traits (days of heading [QN1], days of flowering [QN2], days of maturity [QN3], plant height [QN4], number of internodes [QN5], number of branches [QN6], length of inflorescence [QN7] and number of florets [QN8]) were assessed during the flowering period. Seed characteristics (weight per 100 seeds [QN9] and seed size [QN10]) were measured after a drying period of approximately 1 month post-harvest, with seed size measured in millimetres using a Mitutoyo dial caliper.
Traits used in a morphological survey of 500 Perilla accessions collected from South Korea
The criteria for investigating morphological characteristics were based on a previously published paper by Lee and Ohnishi (Reference Lee and Ohnishi2001).
Statistical analysis
Statistical analysis was performed on the 17 morpho-agronomic traits analysed in the 500 Perilla accessions of the three types. The results for the 7 qualitative traits and 10 quantitative traits were analysed using Microsoft Excel. Heatmap analysis was performed using MetaboAnalyst version 6.0 to compare the 10 quantitative traits among and within the CF, WF and WC accessions. Additionally, the correlation between the 17 morpho-agronomic traits was analysed using MetaboAnalyst version 6.0. Principal component analysis (PCA) was performed using NTSYS-pc V2.1 (Rohlf, Reference Rohlf2000) with the 17 morpho-agronomic traits and data of two types of location (region and city). To determine the statistical significance of the 10 quantitative traits, analysis of variance was performed using the SPSS program.
Results
Variation of morpho-agronomic traits between three types of perilla crop
The 500 Perilla accessions were classified according to the results of the morphological characteristics survey of seven qualitative traits (Fig. 2, Supplementary Table 1). In detail, for QL1 (colour of leaf surface), 418 accessions were green, 32 accessions were reddish-green, 27 accessions were greenish-red, 21 accessions were purple and 2 accessions were dark purple. For QL2 (colour of leaf reverse side), 378 accessions were green, 24 accessions were reddish-green, 14 accessions were greenish-red and 84 accessions were purple. For QL3 (colour of stem), 404 accessions were green, 34 accessions were reddish-green and 62 accessions were purple. For QL4 (colour of flower), 397 accessions were white, 55 accessions were pastel pink, 44 accessions were pink and 4 accessions were purple. For QL5 (degree of pubescence), 451 accessions had slight pubescence, 42 accessions had pubescence and 7 accessions had heavy pubescence. For QL6 (seed hardness), 204 accessions had soft seeds and 296 accessions had hard seeds. For QL7 (colour of seed), 201 accessions were dark brown, 119 accessions were brown, 122 accessions were light brown, 31 accessions were grey, 23 accessions were white and 4 accessions were mixed.
Bar graph showing the frequency distribution of seven qualitative traits in 500 Perilla accessions. QL1: colour of leaf surface; QL2: colour of leaf reverse side; QL3: colour of stem; QL4: colour of flower; QL5: degree of pubescence; QL6: seed hardness; QL7: colour of seeds.
A comparison of the seven qualitative characteristics among the CF, WF and WC accessions showed that most of the 200 CF accessions exhibited green colour for colour of leaf surface (QL1), colour of leaf reverse side (QL2) and colour of stem (QL3), white flowers (QL4), slight pubescence (QL5) and soft seeds (QL6). However, some CF accessions displayed reddish-green colour in QL2, pink flowers in QL4, normal or heavy pubescence in QL5 and hard seeds in QL6. For QL7, CF accessions were classified into dark brown, brown, light brown, grey, white and mixed colours. Among the 200 WF accessions, most accessions exhibited green colour for QL1, QL2 and QL3 and white flowers (QL4). For QL5 and QL6, most WF accessions had slight pubescence and hard seeds. However, some WF accessions displayed a reddish-green colour in QL2, pink flowers in QL4, normal or heavy pubescence in QL5 and soft seeds in QL6. For QL7, most WF accessions were dark brown, while the remaining were classified as brown (31 accessions), light brown (37 accessions), grey (21 accessions), white (3 accessions) and mixed (3 accessions). Among the 100 WC accessions, there was greater diversity in colour for QL1 and QL2 than with the CF and WF accessions. Additionally, most WC accessions had purple stems (QL3), pastel pink or pink flowers (QL4), slight pubescence (QL5) and hard seeds (QL6). For QL7, the 100 WC accessions were classified into dark brown (60 accessions), brown (27 accessions) and light brown (13 accessions) (Supplementary Table 1).
The results of the survey of the 10 quantitative traits for the three types of Perilla are shown in Table 2 and Supplementary Fig. 2. The averages for days of heading, flowering and maturity were all highest for the CF accessions, with the WC accessions having the lowest averages for days of heading and flowering and the WF, WC accessions the lowest average for days of maturity. Similarly, the averages for plant height, number of internodes and number of branches were all highest for the CF accessions, followed by the WF accessions and then the WC accessions. In contrast, the averages for length of inflorescence and number of florets were in the reverse order with the highest for the WC accessions and the lowest for the CF accessions. For seeds, the averages for both weight per 100 seeds and seed size were highest for the CF accessions and lowest for the WC accessions. Although most of the 10 quantitative traits analyzed showed statistically significant differences among the three types of Perilla, no statistically significant differences were found between WF and WC accessions for the QN3 trait (Table 2).
Morphological characteristics of 10 quantitative characteristics in 500 accessions of three types of Perilla crop
*Different letters denote a significant difference at the p < 0.05 level (Scheffé’s multiple comparison test). CF: cultivated var. frutescens; WF: weedy var. frutescens; WC: weedy var. frutescens.
The heatmap analysis was conducted using the 500 accessions of the three types of Perilla, divided into three regions (Central, Southern and Jeju Island), based on the 10 quantitative traits (Supplementary Fig. 3). According to the results, the accessions from Jeju Island exhibited higher values for QN1, QN2 and QN3 compared with the accessions from the other two regions. Additionally, the accessions from Jeju Island showed higher values for QN5 and QN6 and lower values for QN7 and QN8 compared with the accessions from the Central and Southern regions. In the Central region accessions, QN10 showed relatively high values compared with the other two regions, while QN5, QN6 and QN9 showed low values. Finally, the accessions from the Southern region showed the highest plant height (QN4), higher values for QN9 and QN7 and lower values for QN10 compared with the accessions from the other two regions.
Correlation analysis of 17 morpho-agronomic traits
In this study, correlation analysis was conducted for 17 morpho-agronomic traits (Supplementary Table 2). The results showed high positive correlations (0.7 or higher) among four traits (days of heading [QN1], days of flowering [QN2], days of maturity [QN3] and plant height [QN4]) related to the three types of Perilla. Additionally, traits related to inflorescences, length of inflorescence (QN7) and number of florets (QN8), also exhibited a high positive correlation (0.801). Moreover, high positive correlations were observed between seed size (QN10) and weight per 100 seeds (QN9) (0.760), QN4 and number of internodes (QN5) (0.760), QN4 and number of branches (QN6) (0.762) and QN5 and QN6 (0.981). However, negative correlations were found between QN4 and colour of leaf reverse side (QL2) (−0.690), colour of stem (QL3) (−0.628) and colour of flower (QL4) (−0.608) and between QN9 and seed hardness (QL6) (−0.654). Additionally, QN7 and QN8, as well as QN5 and QN6, were found to be highly correlated in the dendrogram of Supplementary Fig. 3.
PCA of 500 Perilla accessions
In this study, PCA was conducted using 17 morpho-agronomic traits and data of two types of location (region and city) for the 500 Perilla accessions of three types (Fig. 3, Supplementary Table 3). The first and second components represented 38.8% and 13.4% of the total variance, respectively, accounting for 52.2% of the total variance. In the first component, plant height (QN4, 0.830), number of internodes (QN5, 0.763), number of branches (QN6, 0.763) and weight per 100 seeds (QN9, 0.743) contributed significantly positively, while traits related to plant colour, colour of leaf reverse side (QL2, −0.819), colour of stem (QL3, −0.783), colour of flower (QL4, −0.773) and colour of leaf surface (QL1, −0.697), contributed significantly negatively. In the second component, region (0.676) and days of maturity (QN3, 0.621) contributed positively. Among the 19 traits (with two types of location), degree of pubescence (QL5) and colour of seed (QL7) did not show a significant influence on either the first or second axes of variation for the Perilla accessions (Supplementary Table 3).
Projection of 500 Perilla accessions of three types in the first and second principal components (● = 200 accessions of cultivated var. frutescens; ■ = 200 accessions of weedy var. frutescens; ▲ = 100 accessions of weedy var. crispa; ◆ = 17 agricultural traits and data of two types of location).
The PCA results show that the 200 CF accessions were predominantly distributed to the right of the origin, while the 200 WF accessions were mostly concentrated in the centre, with a few exceptions being outliers. The 100 WC accessions were widely distributed to the left side. These results indicate a clear separation between CF and WC accessions, with some exceptions, and a certain degree of differentiation between CF and WF accessions (Fig. 3). Furthermore, the relationships among the 17 morpho-agronomic traits observed in the PCA results were consistent with those identified in the correlation analysis (Supplementary Fig. 4).
Discussion
Knowledge of the morphological diversity observed among wild and cultivated crop species, which are widely distributed and show regional differences, is essential for understanding their evolutionary process (Gould and Johnston, Reference Gould and Johnston1972; Wyatt and Antonovics, Reference Wyatt and Antonovics1981). Morphological variation among germplasm collections of a crop species results from microevolutionary adaptations to different environments. Additionally, crop domestication by humans represents an evolutionary transformation in which cultivated plants develop morphological and physiological differences from their wild relatives (Schwanitz, Reference Schwanitz1966; Harlan, Reference Harlan and Chapman1992).
In this study, we evaluated 7 quantitative characters and 10 qualitative characters in 500 accessions of three types (200 of CF, 200 of WF, 100 of WC) of Perilla (Supplementary Table 1). According to the results, the accessions of the three types of Perilla showed various differences depending on the quantitative and qualitative traits investigated in this study. In particular, the CF accessions showed significant differences in plant- and seed-related traits compared with the other two types (WF, WC) of Perilla accessions (Table 2, Supplementary Fig. 2). For example, for seed-related traits, most CF accessions exhibited large and mainly soft seeds, whereas most WF and WC accessions showed small, hard seeds. Additionally, weight per 100 seeds of CF accessions was heavier than that of WF and WC accessions. Along with seed related traits (seed hardness [QL6], seed colour [QL7], seed weight [QN9], seed size [QN10]), other characteristics such as plant height (QN4) and seed dormancy are important characteristics for distinguishing cultivated crops from wild species and represent traits that have been strongly selected during crop domestication (Hancock, Reference Hancock1992; Lee and Ohnishi, Reference Lee and Ohnishi2001; Ha et al., Reference Ha, Sa and Lee2021). Additionally, as reported by Lee and Ohnishi (Reference Lee and Ohnishi2001), this study also demonstrated that the accessions of the three types (CF, WF, WC) of Perilla could be distinguished by morphological characteristics, indicating that there is morphological diversity among the three types of Perilla. Furthermore, CF accessions showed relatively higher values for most quantitative traits compared with the WF and WC accessions, and these traits showed strong correlations with each other (Table 2, Supplementary Table 2, Supplementary Fig. 4). However, the WF and WC accessions scored higher values in the length of inflorescence (QN7) and the number of florets (QN8) compared with CF accessions, suggesting that wild or weedy Perilla accessions tend to have more and smaller seeds and a higher shattering tendency compared with the CF accessions, which were selected and domesticated for larger seeds (high oil content) and a low shattering tendency. These results suggest that CF accessions are thought to have evolved more in seed characteristics than WF and WC accessions through the domestication process, as previously by Lee and Ohnishi (Reference Lee and Ohnishi2001).
Meanwhile, heatmap analysis of the 500 Perilla accessions of the three types using 10 quantitative traits showed differences among most accessions from the three regions of South Korea: the Central region, Southern region and Jeju Island (Supplementary Fig. 3). According to these results, the accessions of Jeju Island showed higher values in the days of heading (QN1), days of flowering (QN2), days of maturity (QN3), number of internodes (QN5) and number of branches (QN6) traits; the Central region accessions showed higher values in the QN10 trait; and the Southern region accessions showed higher values in the plant height (QN4), length of inflorescence (QN7) and weight per 100 seeds (QN9) traits. Therefore, these traits are considered useful traits for distinguishing the accessions collected from the three regions of South Korea through heatmap analysis (Supplementary Fig. 3). Additionally, the PCA results showed that yield-related traits (number of internodes [QN5], number of branches [QN6], weight per 100 seeds [QN9], seed size [QN10]), as well as plant height (QN4), were negatively correlated with plant colour-related traits colour of leaf surface (QL1), colour of leaf reverse side (QL2), colour of stem (QL3) and colour of flower (QL4). These traits showed clear differences among the accessions of the three types of Perilla. As reported by Lee and Ohnishi (Reference Lee and Ohnishi2001), CF and WC accessions could be easily distinguished based on morphological characteristics, while WF accessions showed intermediate characteristics between CF and WC accessions. This result was also observed in the PCA results shown in Fig. 3. Although CF and WC accessions were clearly differentiated morphologically in terms of leaf and seed characteristics in this study, WF accessions were not clearly distinguished from either CF or WC accessions. Previous studies by Nitta and Ohnishi (Reference Nitta and Ohnishi1999) and Lee and Ohnishi (Reference Lee and Ohnishi2001, Reference Lee and Ohnishi2003) suggested that WF accessions may have originated either as an escape type from CF accessions or through hybridization between CF and WC accessions. For Perilla crop germplasm collected in South Korea, we classified the samples based on whether they were cultivated on farms or growing wild around farms or fields. However, among the wild accessions, some, although growing naturally around the fields, had similar characteristics to CF in terms of seed size, etc. These accessions are therefore considered escape types derived from CF. As noted in previous reports by Nitta et al. (Reference Nitta, Lee and Ohnishi2003) and Lee and Ohnishi (Reference Lee and Ohnishi2001), it is believed that seed or gene exchange has occurred frequently between cultivated and weedy types of Perilla crop throughout the long period of its cultivation in South Korea.
No previous study has evaluated accessions of cultivated and weedy types of Perilla crop collected from different regions of South Korea. In this study, however, 500 Perilla accessions were grouped into three regions (Central, Southern and Jeju Island) and assessed. Significant differences in morphological and physiological characteristics were observed (Supplementary Fig. 3). The PCA results of these Perilla accessions categorized by region are shown in Supplementary Fig. 5. In addition to being well classified into CF, WF and WC accessions, the Perilla accessions were also effectively categorized based on collection areas by the regions of Central, Southern and Jeju Island. Five traits (days of heading [QN1], days of flowering [QN2], days of maturity [QN3], number of internodes [QN5], length of inflorescence [QN7], number of florets [QN8]) showed significant differences among the three regions, which were also confirmed by the PCA results. In detail, for QN2 (days of flowering), the average for the Central region was 97.5; for the Southern region it was 96.3; and for Jeju Island, it was 108.2. For QN3 (days of maturity), the average for the Central region was 129.3; for the Southern region, it was 127.8; and for Jeju Island, it was 136.5 (Supplementary Table 4). These differences are because of the milder temperatures in the Southern region and Jeju Island compared with the Central region, which leads to crops being adapted and grown as late-maturing varieties in the former regions while being adapted and grown as early-maturing and mid-maturing varieties in the Central region. The results indicate that geographical regions have an impact on morphological and physiological characteristics regardless of the three types of Perilla accessions collected from South Korea. Perilla is a typical monoculture plant. According to previous report by Ma and Lee (Reference Ma and Lee2017), late-maturity type of Perilla accessions is distributed in southern China, while early- and mid-maturity type of Perilla accessions is distributed mainly in northern China. Therefore, the ecological characteristics of Perilla accessions, which are classified into early and mid-maturing, and late maturing types according to geographical distribution in South Korea, are likely to be characteristics that have been adapted to the local environment over a long period of time during the cultivation process.
As a result, the Perilla genetic resources analysed in this study provided useful information for understanding the morphological variation according to the geographical distribution of three types (CF, WF, WC) of Perilla accessions in South Korea. In addition, these 500 Perilla accessions could be valuable for selecting useful genetic resources for the development of new Perilla cultivars, and they might provide crucial data for the establishment of core collections necessary for the management and preservation of Perilla germplasm in the future.
Conclusion
Our study conducted a diversity analysis of phenotypic traits for 500 Perilla accessions of three types collected from various regions in South Korea. The results revealed diversity within each type of Perilla and statistically differences between CF, WF and WC in 7 qualitative traits and 10 quantitative traits. PCA results of the 500 Perilla accessions clearly distinguished CF from WF and WC. Additionally, morphological and physiological variations were observed across the Central, Southern and Jeju Island regions of South Korea, regardless of type, with significant differences found in five traits (days of heading [QN1], days of flowering [QN2], days of maturity [QN3], length of inflorescence [QN7]and number of florets [QN8]). Therefore, this study will improve our understanding of the morphological and geographical distinctions among the three types of Perilla accessions in South Korea, and it will be helpful for the construction of core collections and breeding programmes for the efficient conservation and management of Perilla genetic resources.
Supplementary material
The supplementary material for this article can be found at https://doi.org/10.1017/S1479262125100075
Acknowledgements
The authors thank the National Agrobiodiversity Center, National Institute of Agricultural Sciences, RDA, Jeonju 54874, South Korea, for supporting the genetic resource management institute at Kangwon National University.
Author contributions
H.P. and J.K.L. wrote the manuscript and designed the experiments. H.P. and J.C. performed the experiment and analysed the data, and T.H.H. helped to draft the manuscript. All authors commented on previous versions of the manuscript and approved the final manuscript.
Funding statement
This study was supported by a National Research Foundation of Korea (NRF) grant funded by the Korean government (MSIT) (No. 2022R1F1A1063300), Republic of Korea.
Competing interests
The authors declare that they have no conflicting interests.
