2018 Volume 87 Issue 1 Pages 115-123
Lilium japonicum Thunb. which has pink or white-colored funnel-like flowers, is distributed in Kyushu, Shikoku, and the western part of Honshu, the main island of Japan. L. auratum Lindl. which has large white flowers with yellow central stripes and red spots, is distributed in the eastern part of Honshu. Natural hybridization of these two species has only been found on the Izu Peninsula of Honshu. However, details of the variation and hybridity of the interspecific hybrid population of these species on this peninsula remain unknown. In the present study, we conducted a morphological examination using 43, 21, and 90 individuals of L. japonicum, L. auratum, and the putative hybrid, respectively, from six, four, and ten populations of the Izu Peninsula, respectively. In addition, we performed simple sequence repeat (SSR) analysis using 47, 41, and 106 individuals of L. japonicum, L. auratum, and the putative hybrid, respectively, from six, four, and ten populations, respectively. Putative hybrid populations that resembled L. japonicum in morphology and SSR profile were found in the southern to eastern part of the peninsula, whereas those that resembled L. auratum and those exhibiting an intermediate morphology and SSR profile were found in the southern part of the peninsula. Large morphological variations exist in putative hybrid in the southern population, and interspecific hybridization has occurred in the southern and eastern populations. These results suggest that the center of natural hybridization is located in the southern part of the Izu Peninsula.
Six species including Lilium auratum, L. japonicum, L. rubellum, L. alexandrae, L. nobilissimum, and L. speciosum fall within the section Archelirion on the basis of morphology and molecular data (Comber, 1949; Nishikawa et al., 1999). All species in this section, except L. speciosum, are endemic to Japan. This section has been used as a source of important parental species for the Oriental hybrid cultivar group of lilies (Shimizu, 1971, 1987). The internal transcribed spacer (ITS) region of 18S-25S nuclear ribosomal DNA suggests small genetic distances among Archelirion species (Nishikawa et al., 1999, 2001). L. alexandrae and L. nobilissimum are monophyletic; however, the relationships among other species of Archelirion remain unclear. Three spacer regions in the chloroplast DNA showed that L. auratum is closely related to L. rubellum and that L. japonicum is closely related to L. nobilissimum, L. speciosum, L. alexandrae, and L. auratum var. platyphyllum (Nishikawa et al., 2002).
L. japonicum Thunb. is distributed in Shikoku, Kyushu, and the western part of Honshu, the main island of Japan (Sakaguchi, 1976), and the Izu Peninsula forms the eastern border of the distribution along the Pacific coast of Honshu for this species (Ogiwara, 1966; Shimizu, 1987). This species is characterized by umbellate inflorescences and white-to-pink funnel-like flowers. In addition, it has no central stripes, papillae, or spots on the tepals. The flowering period of this species occurs from June to July (Shimizu, 1987). It has a large morphological diversity (Nishimura and Atsumi, 2000), and four varieties are recognized by their morphological and physiological characteristics (Shimizu, 1987): 1) L. japonicum var. albomarginatum; 2) L. japonicum var. platyfolium; 3) L. japonicum var. angustifolium; and 4) L. japonicum var. abenanum. L. auratum Lindl. is mainly distributed in the eastern part of Honshu. This species is characterized by racemose inflorescences, large white flowers with yellow central stripes, and red to brown colored spots on recurved tepals. The flowering period usually extends from July to August (Komata, 2007; Shimizu, 1987).
The Izu Peninsula is located in the central Pacific coast of Honshu. The climate of this peninsula is described as mild oceanic owing to the influence of the warm Japan Current. In addition, the Izu Peninsula area receives heavier rainfall than the nearby central coastal region (Hirohito, 1980). L. japonicum (Shimizu, 1971, 1987) is distributed in the southern and eastern parts of this peninsula (Ogiwara, 1966) (Fig. 1).
Distributions of Lilium japonicum and L. auratum, and the location of the Izu Peninsula. The area marked by bold lines on the peninsula shows the endemic area of L. japonicum. Numbers of investigated sites are shown in Table 1.
Natural hybridization has occurred in many plant species (Anderson and Stebbins, 1954) and has extensively contributed to angiosperm diversity and evolution (Soltis and Soltis, 2009). L. japonicum and L. auratum are diploid (2n = 24), and artificial interspecific hybridization using these species has been conducted to breed the Oriental hybrid cultivar group of lilies (Asano, 1986). The distributions of L. japonicum and L. auratum overlap in the central area of Honshu; consequently, natural hybridization between these species has been suggested only on the Izu Peninsula (Ogiwara, 1966; Watanabe, 1987). Ogiwara (1966) reported that the karyotype of some L. japonicum individuals occurring in this peninsula was different from that of individuals of other regions, and it resembled that of L. auratum. Some L. japonicum individuals on this peninsula had red spots on their tepals, similar to L. auratum. Watanabe (1987) found morphological variations in L. japonicum, L. auratum, and putative hybrids on this peninsula. He named these hybrids ‘Izu-yuri’ and classified the putative hybrid individuals into three types: 1) L. japonicum type; 2) L. auratum type; and 3) an intermediate type. Watanabe (1987) also suggested that interspecific hybridization had occurred in the distribution area of L. japonicum in this peninsula.
Microsatellites or simple sequence repeat (SSR) markers have many advantages over other markers, such as high levels of polymorphism and transferability. SSR markers for lilies have been developed in some studies. Horning et al. (2003) developed six SSR markers for L. philadelphicum and Kawase et al. (2010) developed three SSR markers for L. japonicum. In our previous study, SSR markers revealed the natural hybridization of L. auratum and L. japonicum on this peninsula (Yamamoto et al., 2014). Expressed sequence tag (EST)-SSR markers applicable for evaluation of genetic diversity and identity in the Lilium genus have recently been developed. Lee et al. (2011) developed 19 EST-SSR markers and Yuan et al. (2013) developed 118 EST-SSR markers using an EST database. We used these EST-SSR markers in our previous study to estimate genetic diversity of L. auratum wild populations (Yamamoto et al., 2017).
In the present study, we characterized wild populations of L. japonicum, L. auratum, and their putative hybrids located on the Izu Peninsula by morphological and SSR analysis. We also analyzed the morphological diversity and genetic variation of putative hybrids to reveal the hybrid zone in the peninsula.
We conducted morphological analysis for the southern and eastern parts of the Izu Peninsula (Fig. 1). Plant materials included in the morphological and SSR analysis are summarized in Table 1. Forty-three L. japonicum individuals from six populations (Minami-Izu A, 2; Minami-Izu B, 6; Minami-Izu C, 17; Minami-Izu D, 15; Minami-Izu E, 2; Shimoda A, 1); 21 L. auratum individuals from four populations (Shimoda H, 5; Shimoda I, 5; Matsuzaki, 8; Higashi-Izu D, 3); and 90 putative hybrids of L. auratum and L. japonicum from ten populations (Kawazu, 8; Higashi-Izu A, 22; Higashi-Izu B, 11; Higashi-Izu C, 15; Shimoda B, 3; Shimoda C, 7; Shimoda D, 7; Shimoda E, 4; Shimoda F, 9; Shimoda G, 4) were investigated. In total, 154 individuals were used for the morphological analysis.
Populations and numbers of individuals used for morphological and SSR analysis.
The following 18 characteristics were measured during the flowering period: 1) leaf length; 2) leaf width; 3) number of flowers; 4) outer tepal length; 5) outer tepal width; 6) number of papillae in the outer tepal; 7) color of spots on the outer tepal; 8) number of spots on the outer tepal; 9) inner tepal length; 10) inner tepal width; 11) number of papillae in the inner tepal; 12) number of spots on the inner tepal; 13) color of spots on the inner tepal; 14) color of central part of the tepal; 15) color of nectar furrow in the tepal; 16) flower shape of the perianth; 17) corolla length, and 18) stigma color. Discrimination of spot, stigma and tepal colors was done according to the Royal Horticultural Society (R.H.S) color chart. If the spot color was the same in the inner and outer tepal, we used this characteristics as spot color. Flower diameter and number of flowers were affected by temperature in lilies (Lucidos et al., 2013). To avoid misleading correlations that can be affected by climate or annual changes, we selected the following 13 characteristics for further statistical analysis: 1) leaf length; 2) leaf width; 3) outer tepal length; 4) outer tepal width; 5) number of papillae in the outer tepal; 6) number of spots on the outer tepal; 7) inner tepal length; 8) inner tepal width; 9) number of papillae in the inner tepal; 10) number of spots on the inner tepal; 11) color of central part of the tepal (pink, white); 12) spot color (absent, yellow, yellow + red, and red), and 13) stigma color (green, gray, and purple). We compared these 13 characteristics among populations using principal component analysis (PCA) in SPSS version 20.0 (IBM, Armonk, NY, USA).
Plant materials for SSR analysisIn total, 194 individuals were used for the SSR analysis (Table 1). Forty-seven samples of L. japonicum were collected from six populations (Minami-Izu A, 3; Minami-Izu B, 6; Minami-Izu C, 17; Minami-Izu D, 18; Minami-Izu E, 2; Shimoda A, 1); 41 samples of L. auratum were collected from four populations (Shimoda H, 16; Shimoda I, 10; Matsuzaki, 10; Higashi-Izu D, 5); and 106 samples of the putative hybrid between two species were collected from ten populations (Kawazu, 10; Higashi-Izu A, 22; Higashi-Izu B, 11; Higashi-Izu C, 15; Shimoda B, 3; Shimoda C, 9; Shimoda D, 11; Shimoda E, 11; Shimoda F, 9; Shimoda G, 5).
SSR analysisDNA was isolated from 20 mg of fresh leaves, stamen, or stigma using the modified cetyltrimethylammonium bromide (CTAB) method (Doyle, 1990). Seven EST-SSR markers (L20, L59, eL16, ivflmre252, ivflmre294, ivflmre330, and ivflmre850) were selected from the studies of Lee et al. (2011) and Yuan et al. (2013) (Table 2). Polymerase chain reaction (PCR) was performed in reaction mixtures containing 0.4 μL of deoxyribonucleoside triphosphate (dNTP) mix, 0.5 μL of 10 × reaction buffer, 0.3 μM each of forward and reverse primer pairs, 0.025 μL of Ex Taq polymerase (Takara, Tokyo, Japan), 20 ng of DNA, and sterile distilled water in a final volume of 5 μL. PCR amplification was performed using a T100 thermal cycler (BioRad, Hercules, CA, USA). For the L20, L59, and eL16 markers, the reaction cycles consisted of an initial denaturation at 94°C for 2 min, followed by 35 cycles of denaturation at 94°C for 45 s, annealing at 55°C for 45 s, and extension at 72°C for 1 min, followed by a final extension step at 72°C for 5 min (Lee et al., 2011). For the ivflmre252, ivflmre294, ivflmre330, and ivflmre850 markers, a touchdown PCR amplification was performed with the following reaction cycles: denaturation at 94°C for 5 min, followed by 10 cycles of denaturation at 94°C for 30 s, annealing at 63°C for 30 s (reducing by 0.5°C in each cycle), and extension at 72°C for 45 s. Subsequently, PCR was performed with 25 cycles of denaturation at 94°C for 30 s, annealing at 58°C for 30 s, and extension at 72°C for 45 s, followed by a final extension at 72°C for 5 min (Yuan et al., 2013). Forward primers were fluorescently labeled with either NED or HEX (Applied Bio Systems, Carlsbad, CA, USA). PCR products were electrophoresed with an internal size standard (Gene scan-350; Applied Bio Systems) using a 3100 Genetic Analyzer (Applied Bio Systems). Genotypes were analyzed using Peak Scanner 1.0 (Applied Bio Systems).
Primer sequences of SSR markers (Lee et al., 2011; Yuan et al., 2013) used in this study.
Bayesian clusters were determined by Structure v.2.3.4 (Falush et al., 2007; Pritchard et al., 2000). The numbers of distinct clusters (K) varied from one to ten. Ten iterations were performed for each K, with a burn-in of 100000 and a Markov chain Monte Carlo (MCMC) of 100000 iterations. The optimal K was estimated by the calculation of ΔK (Evanno et al., 2005). In addition, a combination of assignment tests with canonical discriminant analysis was conducted (De Rick et al., 2013). Using this method, an assignment table was produced, and the assignment values were used as inputs for canonical discriminant analysis.
Flowering period and morphological variations of 13 characteristics in six populations of L. japonicum, four populations of L. auratum, and ten populations of the putative hybrid on the Izu Peninsula are shown in Tables 1 and 3. Flowers of L. japonicum, L. auratum, and the putative hybrids are shown in Figure 2.
Mean values of 13 morphological characteristics of 20 populations.
Flowers of Lilium japonicum, L. auratum, and putative hybrids. A, L. japonicum; B, L. auratum; C, Hybrid (L. japonicum-like); D, Hybrid (intermediate); E, Hybrid (L. auratum-like).
Populations of L. japonicum (populations 1–6) had no papillae on their tepals, and their flowering period was earlier than that of other populations. Populations of L. auratum (17, 18, 19, and 20) had red spots on their tepals, and their morphological characteristics were typical of L. auratum. Populations of the putative hybrid in the Kawazu and Higashi-Izu (populations 7–10) had few papillae on their tepals and their flowering period was later than that of L. japonicum. Morphological characteristics of these populations were more similar to those of L. japonicum than those of L. auratum. Putative hybrid populations of Shimoda B and Shimoda C (populations 11 and 12, respectively) also showed L. japonicum-like phenotypes; however, individuals of these populations had no papillae on their tepals, and their flowering period was later than that of L. japonicum populations. Individuals of Shimoda D (population 13) showed hybrid morphological characteristics; the number of papillae, number of spots on their tepals, and their leaf widths were intermediate between the two species. The putative hybrid populations of Shimoda E, Shimoda F, and Shimoda G (populations 14, 15, and 16 respectively) showed L. auratum-like phenotypes; however, they had yellow spots or pink tepals, which were not found in L. auratum populations.
The eigenvalues of principal component 1 (PC1) and PC2 were 8.00 and 1.85, respectively (Table 4). The percentage contributions of PC1 and PC2 were 61.55% and 14.24%, respectively. The first and second contributions cumulatively accounted for 75.79% of the total. PC1 explained the majority of flower characteristics and leaf width, whereas PC2 explained the leaf length. A two-dimensional scatter diagram of the first and second components for the 13 morphological characteristics showed that individuals of L. japonicum (population 1–6) and L. auratum (population 17–20) were separated (Fig. 3). Individuals of putative hybrids with L. japonicum-like populations (populations 7–12) or L. auratum-like populations (populations 14–16) were plotted closely by the range of either L. japonicum or L. auratum. Putative hybrid individuals with an intermediate phenotype (population 13) were plotted between the two species.
Factor loadings of 13 morphological characteristics in the first and second principal components (PCs) of the PCA.
Two-dimensional scatter diagram of the first and second components using 13 morphological characteristics in Lilium japonicum, L. auratum, and putative hybrids. Population codes for symbols are shown in Table 1.
The highest ΔK determined by the Bayesian cluster analysis was 2. When K = 2 (Fig. 4), individuals of L. japonicum from Minami Izu and Shimoda A (populations 1–6) and putative hybrid individuals of Kawazu (population 7) were mainly assigned to cluster I. Most putative hybrid individuals of Higashi-Izu A, B, and C (populations 8–10) and Shimoda B, C, and D (populations 11–13) were assigned to cluster I; however, some individuals were assigned to both clusters I and II. Putative hybrid individuals from Shimoda E, F, and G (populations 14, 15, and 16, respectively) and individuals of L. auratum from Shimoda H (populations 17) were mainly assigned to cluster II, although some individuals showed mixed clusters. Individuals of L. auratum from the Shimoda I, Matsuzaki, and Higashi-Izu D (populations 18, 19, and 20, respectively) were assigned to cluster II.
Bayesian cluster analysis of 20 populations of Lilium japonicum, L. auratum, and putative hybrids by SSR analysis, estimated from the data with K = 2. Population codes for symbols are shown in Table 1.
Canonical discriminant analysis separated L. japonicum (populations 1–6) and L. auratum (populations 17–20) on axis 1 (Fig. 5). Putative hybrid individuals from Kawazu (population 7), Higashi-Izu A, B, and C (populations 8–10), and Shimoda B, C (populations 11 and 12) were plotted closely by the range of L. japonicum individuals (populations 1–6). Putative hybrid individuals from Shimoda E, F, and G (populations 14–16) were plotted closely by the range of L. auratum individuals (populations 17–20). Putative hybrid individuals from Shimoda D (population 13) were plotted between the two species.
Discriminant analysis derived plot of the parent species and putative hybrids in 20 populations. Population codes for symbols are shown in Table 1.
Natural hybridization between L. japonicum and L. auratum has been suggested by previous studies (Ogiwara, 1966; Watanabe, 1987); however, details of the hybrid zone have not yet been analyzed. Our morphological observations and SSR analysis revealed that the putative hybrid zone contained L. japonicum-like, L. auratum-like, and intermediate type individuals. The populations of Kawazu (population 7), Higashi-Izu A, B, and C (populations 8, 9, and 10), and Shimoda B and C (populations 11 and 12) showed L. japonicum-like morphological characteristics. The distribution of L. japonicum is endemic to the southern to eastern parts of the Izu Peninsula (Ogiwara, 1966). In the Izu Peninsula, the flowering season of L. japonicum is later than that in other regions (Inagaki et al., 2000). The late flowering of L. japonicum in this peninsula increases its chances of hybridization with L. auratum. The populations of Shimoda E, F, and G (populations 14, 15, and 16) showed L. auratum-like characteristics. Ogiwara (1966) suggested that gene flow has occurred from L. auratum to L. japonicum. In this study, gene flow from L. japonicum to L. auratum was suggested, because the characteristics and SSR profile of Shimoda E, F, and G (populations 14, 15, and 16) were very similar to L. auratum (populations 17, 18, 19, and 20). It was suggested that reverse introgression has occurred in the Shimoda area. Principal component and canonical discriminant analysis showed that Shimoda D (population 13) has a larger diversity. Horie et al. (2012) suggested that the center of the hybrid zone showed larger morphological diversity than the peripheral areas in a study of Epimedium diphyllum and E. sempervirens var. rugosum. Interspecific hybridization of Arisaema angustatum and A. suwoense in the Izu Peninsula was reported (Kakishima, 2012). These two species grows sympatrically and natural hybridization has occurred in northern parts of the peninsula. Northern populations consist of a hybrid and A. angustatum. However, the introgression did not occur in southern populations. Kakishima (2012) suggested the transition of Arisaema species vegetation and formation of Izu Peninsula. A. suwoense spread from the southern area, and that hybridization occurred in northern populations first. In our study, hybridization of L. japonicum and L. auratum occurred in the southern and eastern parts of the peninsula. L. japonicum spread from Minami-Izu and hybridization may have occurred in the southern part of the Shimoda area.
Factors in hybridizationInterspecific hybridization has been used to breed lilies (Asano, 1987; Shimizu, 1987). Asano (1987) confirmed successful cross-pollination between L. auratum and L. japonicum. However, there have been few reports of a hybrid zone or natural introgression of wild populations in the Lilium genus. The natural distribution areas of L. auratum and L. speciosum do not overlap in natural habitats. Hybridization between artificially planted L. auratum and wild L. speciosum was reported in the Shikoku area (Shimizu, 1987). This hybridization occurred via pollen flow from planted L. auratum into wild individuals of L. speciosum. The natural distribution areas of L. auratum and L. rubellum overlapped. However, the flowering period of L. rubellum occurs from May to June in the wild population (Shimizu, 1987), while the flowering period of L. auratum occurs from July to August in this area (Komata, 2007; Shimizu, 1987). Hybridization of L. auratum and L. rubellum is not found in wild populations owing to the discrepancy in their flowering periods. The distribution areas of L. japonicum and L. auratum overlapped on the Izu Peninsula (Fig. 1). The flowering period of L. auratum extends from July to August, whereas that of L. japonicum is usually earlier than L. auratum (Shimizu, 1987). This means that interspecific gene flow between these two species is usually restricted in wild populations. In the present study, the flowering period of L. japonicum populations (populations 1–6) was during early June; therefore, interspecific gene flow was restricted in these populations (Table 1). Interestingly, we found that the flowering period of putative hybrids of L. japonicum-like individuals (populations 7, 8, 9, 10, 11, and 12) was later than that of L. japonicum populations (Table 1). The flowering period of L. auratum-like individuals (populations 14, 15, and 16) was earlier than that of L. auratum populations (populations 17, 18, 19, and 20). Flowering periods of L. auratum-like individuals and L. japonicum-like individuals overlapped in the southern part of the Shimoda area, suggesting the possibility of hybridization among these populations. Inagaki et al. (2000) reported that the flowering period of L. japonicum in the Izu Peninsula was correlated with longitude, and that flowering period of eastern populations were later than western populations. In our study, flowering period of L. japonicum-like individuals overlapped with L. auratum-like individuals, so this may be a factor in interspecific introgressions.
The main pollinators of L. japonicum have been identified as hawk moth (Acosmeryx naga and Sphinx constricta) (Inagaki, 2003). On the other hand, the pollinators of L. auratum are butterfly (Papilio bianor) during daytime and hawk moths (Meganoton analis) during nighttime (Morinaga et al., 2009). Floral volatiles of L. auratum have been reported (Morinaga et al., 2009), and their concentrations are higher at nighttime than in the daytime; therefore, it has been suggested that L. auratum may attract nocturnal visitors such as hawk moths. Pollinators of these species were not identified in the hybrid zone of the Izu Peninsula. Hawk moth species may be common pollinators among these lilies, resulting in hybridization.
Our results indicate that interspecific hybridization of L. auratum and L. japonicum occurs in the southern and eastern parts of the Izu Peninsula and that there is a high morphological diversity in the southern region of the Izu Peninsula, indicating that this area may be the center of the hybrid zone.
We are grateful to Dr. M. Yamagishi of the Research Faculty of Agriculture at the Hokkaido University for useful suggestions.
