Introduction

The genus Lilium includes approximately 100 species and an abundant number of commercially important cultivars (McRae 1998; Van Tuyl et al. 2011). Lilies are widely used in medicine, food industry, and the flower market. China is the center of the distribution of Lilium, with approximately 55 species (nearly 50% of the global total) (Liang and Tamura 2000), and most wild species have excellent commercial properties. For example, Lilium regale Wilson, Lilium sulphureum Baker and Lilium henryi Baker have been extensively used in breeding (McRae 1998; Prosevičius and Strikulyte 2004). Lilium lancifolium Thunberg (3x, 2n = 36) is a natural triploid in the genus (Noda 1986) and is native to eastern temperate Asia, in China, Japan and Korea. It is widely planted as an ornamental in subtropical and temperate areas, and its bulbs and flowers (without stamens) are edible (McRae 1998; Lim 2014). The cultivation of L. lancifolium in China has a long history, as it is one of the edible lilies and a legal medical resource, recorded in The Pharmacopeia of People’s Republic of China and other historical Chinese herbal records (Lim 2014; Liu 2015).

Lily species are taxonomically classified into seven different sections (Lilium, Martagon, Pseudolirium, Archelirion, Sinomartagon, Leucolirion and Oxypetalum) based on various morphological and physiological characteristics (Comber 1949; Khan 2009). According to the description of McRae (1998), hybrid lilies are presently categorized into ten divisions (Asiatic hybrids, Martagon hybrids, Candidum hybrids, American hybrids, Longiflorum hybrids, Chinese trumpet hybrids, Oriental hybrids, Orienpet hybrids, Species and Miscellaneous hybrids) under the horticultural classification. At present, Most of popular cultivars are interspecific hybrids between the sections (e.g., Oriental × Trumpet hybrids, Longiflorum × Oriental hybrids, Longiflorum × Asiatic hybrids, etc.), and some are interspecific hybrids (Longiflorum hybrids, Asiatic hybrids and Oriental hybrids) within the sections (especially Leucolirion, Archelirion and Sinomartagon). Lilium species exhibit great variation in DNA content, ranging from 13.7 to 47.9 pg/1C with an average of 36.0 pg/1C (Bennett and Leitch 2012). Accordingly, the genetic backgrounds of modern hybrids with different ploidies are complicated, and most genetic loci have a high degree of heterozygosity (Zhou 2007; Xie 2012). Furthermore, some hybrids are sterile, especially triploids (Asano 1982). In our previous research, we found that the abnormal meiotic frequency of pollen mother cells during meiosis was closely related with heterozygosity. (For example, the wild species L. regale had the lowest observed abnormal meiotic rate of 7.65% at the tetrad phase, and two Oriental hybrids had lower abnormal meiotic rates than the Oriental × Trumpet hybrid ‘Robina’, which had a high abnormal meiotic rate of 71.57%; Li et al. 2012.) Therefore, pollen viability is related to heterozygosity in lilies.

In Lilium breeding, crossing is the most important tool to combine agronomic traits in new hybrids. Within a section it can be made with relative ease, but intersectional hybridization is very difficult (Barba-Gonzalez 2005). To address this challenge, studies of the factors affecting pollen growth in intra- and interspecific pollinations, pre- and post-fertilization barriers in interspecific crosses and the interaction between the pollen tube and pistil have been conducted (Ascher and Peloquin 1966, 1968; Ascher 1975; Amaki and Yamamoto 1988; Van Tuyl et al. 1982, 1991; Janson 1992; Janson et al. 1993), and many different techniques have been developed to overcome pre- and post-fertilization barriers in lily breeding (Asano and Myodo 1977; Asano 1978; Van Creij et al. 1990; Van Tuyl et al. 2011). Whether intra- or inter-sectional crosses are involved, both pollen viability and embryo development are important for ensuring successful controlled pollination. In addition, when performing crosses of compatible lilies within sections or hybrid divisions, stigma receptivity is an important factor affecting seed set.

Stigma receptivity and pollen viability are critical for the effective initiation of the pollen-pistil interaction (Shivanna 2003). A stigma is receptive when it has the ability to support pollen germination, and the onset of stigmatic receptivity is accompanied by a number of changes that occur upon stigma maturation (Sanzol and Herrero 2001). The duration of receptivity can vary from a few hours to up 10 days (Dafni 1992; Shivanna 2003). According to Dafni (1992), stigma receptivity can be investigated to identify the optimum flower age for artificial pollination procedures and to increase pollination efficiency. In addition, a sticky secretion that indicates receptivity usually accompanies the female stage of the flower (Heslop-Harrison and Shivanna 1977). In Eucalyptus woodwardii, the amount of stigma secretion and the ability to support pollen germination and tube growth were found to increase with time, reaching a peak at 7 days after anthesis (Sedgley and Smith 1989). In Lilium longiflorum, exudate production was studied from approximately 8 days before anthesis to 2 days after anthesis, and Janson et al. (1994) observed that the exudate production on the stigma and in the style started before anthesis, although it was unclear whether any of the exudate fluid originated from the micropyle. At approximately 8 days before anthesis, the stigma remained largely dry. At three to 4 days before anthesis, some papillae at the stigma were collectively covered by a layer of exudate, and at 2 days after anthesis, a thick layer of exudate was visible on the stigma (Janson et al. 1994). Moreover, the stigmatic sap and the stylar sap accelerated the growth of pollen tube in L. longiflorum (Amaki et al. 1989). Both the speed and the guidance of the pollen tube appeared to be determined by the properties of the exudate (Janson et al. 1994; Willemse et al. 1995). The production of stigma secretions appears to indicate receptivity for pollination and does not appear to be triggered by pollination. The pollen viability phase is the time during which pollen is able to germinate on the appropriate (receptive and compatible) stigma (Dafni 1992; Dafni and Firmage 2000). This phase varies among species and ranges from minutes to months after shedding (Shivanna and Johri 1985). Pollen viability studies are recognized as essential for understanding the reproductive performance of species and the successful implementation of breeding programs (Dafni and Firmage 2000).

Studies of pollen viability and stigma receptivity support the successful controlled pollination of Lilium because the application of viable pollen to the receptive stigmas increases seed set. However, systematic information for many Lilium species or hybrid series is lacking. Therefore, the aims of the present research were to determine the timing of the pollen viability and stigma receptivity of 6 Lilium cultivars and 3 species native to China and to elucidate the relationships between stigma receptivity and both embryo sac development and stigma secretion.

Materials and methods

Plant materials

A total of 6 cultivars and 3 species native to China in the Lilium genus were studied (Table 1). Four cultivars were introduced from the Netherlands and included three Oriental hybrids (OO) (‘Sorbonne’, ‘Siberia’, ‘Starfighter’), and one Asiatic hybrid (AA) (‘Tiny pudhye’). The remaining two cultivars, a Longiflorum hybrid (LL) (‘Baiguang No. 3’) was from Japan, and a AA hybrid (‘Jinghe’) were bred by the Institute of Vegetables and Flowers, Chinese Academy of Agricultural Sciences, Beijing, China (Yuan et al. 2016). The three Lilium species native to China were L. henryi, L. sulphureum and L. regale, and were provided by the Institute of Vegetables and Flowers of the Chinese Academy of Agricultural Sciences, Beijing, China. For each genotype, 30–50 plants were planted in experimental fields in Beijing and Kunming, Yunan Province, that belong to the Institute of Vegetables and Flowers of the Chinese Academy of Agricultural Sciences, Beijing, China. Each bud was protected with a sulfuric acid paper bag before anthesis to avoid cross-pollination. Samples were collected from healthy plants from June to September, 2016.

Table 1 Characteristics and locations of plant materials and the sampling times of floral organs
Full size table

Pollen germination

Pollen viability at different flowering times during anthesis was determined based on germination. For each genotype, pollen grains of dehisced anthers were collected from 3 or 6 flowers with the same flowering time between 10:00 and 11:00 am in the field, and their viability was assayed using in vitro liquid medium techniques (Griffin et al. 1982). The same number of flowers was sampled at different flowering times for each genotype. The pollen incubating solution consisted of 5% sucrose, 150 mg/L boric acid, and 20 mg/L CaCl2. Approximately 5 µL of the solution was dropped on a glass slide, and then a small amount of pollen that had been collected using a toothpick was added to it. After incubation at 25 °C for 4 h, pollen germination was recorded by observation under a microscope (Olympus CX31) at ×400 magnification. A pollen grain was considered to have germinated if the pollen tube was twice as long as the diameter of the pollen (Rodriguez and Dafni 2000).

Each replicate was composed of at least 300 pollen grains from 1 or 2 flowers, and every treatment was replicated three times. Pollen germination data were analyzed by ANOVA (SAS 8.1 software and Microsoft Excel 2007), and means were compared by Duncan’s multiple-range test at P = 0.05.

Stigma receptivity

Peroxidase activity was examined on non-pollinated stigma surfaces. Three to six fresh flowers at the same flowering time during anthesis were collected between 8:00 and 9:00 am in the field and maintained under hydration until analysis. Stigmas were then excised from the flowers using scalpels and placed on a glass slide in a drop of benzidine-H2O2 solution (1% benzidine:3% H2O2 hydrogen peroxide:water = 4:11:22, v/v; Dafni 1992). Stigma receptivity was recorded after 10–15 s by observation under a dissecting microscope. Receptivity was determined based on a reaction to the solution, in which the stigma turns blue and bubbles appear on its surface. A deeper blue with more bubbles indicated stronger stigma receptivity.

Embryo sac development

Based on the results from the stigma receptivity experiments, flowers at the first day of flowering and with initial stigma receptivity were collected, and their ovaries were removed to observe the developmental stage of the embryo sac. Five ovaries were selected for each flower development stage, yielding a total of ten ovaries over both development stages for each genotype. The middle part of each ovary was fixed in FAA (50% ethanol:glacial acetic acid:38% formaldehyde = 18:1:1) for at least 2 days at 4 °C. The samples were then dehydrated through a graded ethanol series, infiltrated with xylene and embedded in paraffin wax by conventional methods (Hu 2005). The ovaries were then sectioned on a microtome (10 µm thick sections) and stained with safranine and fast green.

Stigma secretion

The color and quantity of the stigma secretions were observed with the naked eye during anthesis.

Results

Pollen germination

Anthers began to crack on the first day of flowering in all genotypes. For the three OO hybrids and two wild species (L. henryi and L. regale), the pollen germination percentage was highest on the first day of anthesis and then decreased over time (Table 2). The highest pollen germination rate was longest in ‘Baiguang No. 3’ (LL) (from 0 to 2 days after anthesis), followed by L. sulphureum (from 0 to 1 day after anthesis). The AA hybrid ‘Jinghe’ exhibited no pollen germination and ‘Tiny pudhye’ (AA) only exhibited low pollen viability for a short period of time (from 0 to 1 day after anthesis).

Table 2 Pollen germination percentage during the course of anthesis in different lily genotypes
Full size table

Stigma receptivity

According to the results of pollen germination, seven genotypes having better fertility were selected in this experiment (Table 3). All tested genotypes showed stigma receptivity at 1 or 2 days after anthesis, and had 5–8 days of stigma receptivity. (Table 3; Fig. 1a). Stigma receptivity first increased and then decreased over time after anthesis (Table 3; Fig. 1b). The time and duration of the period of highest receptivity were genotype dependent. The period of highest stigma receptivity lasted for only 1 day in L. henryi and for 4 days in L. sulphureum.

Table 3 Stigma receptivity during the course of anthesis in different lily genotypes
Full size table
Fig. 1
figure 1

‘Starfighter’ stigmas from 0 to 7 days after anthesis (DAA). a Stigmas showing no reaction but showing wet secretions (from 0 to 6 days after anthesis; Table 3). b Stigmas after treatment with benzidine-H2O2 solution

Full size image

Relationship between embryo sac development and stigma receptivity

Similarly, during anthesis, embryo sac development was observed in those seven genotypes (Table 4). On the first day of flowering, six genotypes had mature embryo sacs (Fig. 2), whereas development lagged in the other one genotype (‘Sorbonne’). Embryo sacs in ‘Sorbonne’ matured 2 days after anthesis, when the stigmas began to show receptivity. In summary, all the tested genotypes had mature embryo sacs at the beginning of stigma receptivity.

Table 4 Development of embryo sacs at different flowering times
Full size table
Fig. 2
figure 2

Mature ‘Dandie’ embryo sac. PN polar nuclei, Sy synergid, EC egg cell

Full size image

Relationship between stigma secretion and stigma receptivity

No stigma secretion was observed with naked eye in the two AA hybrids. In the remaining genotypes, the secretions were colorless, shiny, wet and sticky (Fig. 1a). The results indicated that when the secretions began to appear on the surfaces of the stigmas, receptivity had begun or would begin 1 day later (Table 5). In most genotypes stigmas were most receptive when the secretions peaked, or reached highest receptivity 1 day later as in the cases of L. henryi and L. sulphureum.

Table 5 Relationship between stigma secretion and stigma receptivity
Full size table

Discussion

During the last few decades, polyploidy cultivars originated from hybridization of cultivars that belong to different taxonomic sections have been replacing the diploid forms due to their superior characteristics (Barba-Gonzalez et al. 2005; Zhou 2007; Zhou et al. 2008; Khan et al. 2009). Moreover, allotriploids are predominantly polyploids. These allotriploids are difficult to use as parents in breeding because the triploids are mostly sterile due to abnormal meiosis (Asano 1982, 1983). According to the hypothesis “Five same genomes of endosperm are essential for its development in Lilium”, male-sterile triploid lilies can be used as the female parent to cross with appropriate diploid and tetraploid males to produce aneuploid progenies, but the breeding efficiency is very low (Zhou et al. 2012, 2014). Therefore, in this research, diploid genotypes representing three important groups of hybrids (Longiflorum, Asiatic and Oriental) were tested, including some cultivars (e.g., the three OO hybrids ‘Sorbonne’, ‘Siberia’ and ‘Starfighter’) that are popular lilies with good characters for cut flowers in China. Another 3 species in the Lilium genus that are native to China were selected: L. henryi, L. regale and L. sulphureum, which are important parents of trumpet hybrids in lily breeding (McRae 1998). All the materials investigated in the study appear to have great potential as donor materials in breeding.

Successful seed set generally depends on the viability of pollen grains, which can be assessed by different methods, including staining with non-vital dyes, in vitro germination tests (Heslop-Harrison et al. 1984; Shivanna and Johri 1985) and in vivo tests, such as analyses of final seed set (Shivanna and Johri 1985; Razora and Zsuffa 1986). Analyzing final seed set is the most reliable method (Shivanna and Johri 1985; Bhattacharya and Mandal 2004) but has the disadvantage of a high time investment, whereas non-vital stains are useful for rapidly determining pollen viability (Kearns and Inouye 1993). The in vitro germination method has been found to be reliable for determining pollen viability in Shorea robusta (Bhattacharya 2011) and was used to determine pollen viability in the present study. In this research, ‘Jinghe’ (AA) did not exhibited pollen germination and ‘Tiny pudhye’ (AA) only exhibited low pollen viability for a short period of time (from 0 to 1 day after anthesis (Table 2). Furthermore, in practice, to cross with diploid AA hybrids, no seeds were got when ‘Jinghe’ was used as the female parent and only very few seeds can be obtained sometimes when ‘Tiny pudhye’ as the female parent (no data shown). So the two AA hybrids were mostly sterile. This maybe relate to AA’s genetic background. These cultivars in the group of Asiatic hybrids are derived from interspecific hybridization among approximately 12 species of the Sinomartagon section, while OO hybrids are bred from interspecific hybridization between 6 species in section Archelirion and LL hybrids originate from section Leucolirion containing 4 species (Woodcock and Stearn 1950; McRae 1998). In addition, in most of those genotypes showing pollen germination, the rates were highest on the first flowering day and then slowly decreased in the days following anthesis, remaining at approximately 50% until the natural shedding of the anthers. In this experiment, the pollen germination rate of ‘Starfighter’ (OO) was very low throughout the flowering period; it is possible that the liquid medium used for pollen germination was not suitable for this genotype. In future research, we will attempt to formulate a liquid medium that is suitable for pollen germination in a wider variety of genotypes.

Stigma receptivity has been studied in only a limited number of ornamental plants (Shivanna 2003; Baskorowati 2009; Kalinganire et al. 2000; Sliva et al. 2013; Souza et al. 2016). It can be determined based on the stigma’s morphological changes, pollen germination on its surface, and staining or testing for enzymatic activity; however, none of these methods provide precise estimates. According to Shivanna (2003), the only definitive method to determine stigma receptivity is through controlled pollination and subsequent analysis of effective pollen germination until fruit and seed set. However, this method is not widely applicable in Lilium because the genetic backgrounds of modern hybrids of different ploidies are complicated, and many crosses are not compatible. Receptive stigmas invariably present several enzymes such as esterases, peroxidases, and acid phosphatases, and the activities of these enzymes on the stigma surface have often been considered to indicate receptivity (Dafni 1992; Dafni and Maués 1998). In this study, peroxidase activity was examined on non-pollinated stigma surfaces with a benzidine-H2O2 solution (Dafni 1992). This solution provides a fast, simple, and easily performed method for determining stigma receptivity. In Shorea robusta, the duration of stigma receptivity was shown to be long, initially increasing and then decreasing during anthesis (Bhattacharya 2011). Our results in Lilium are similar. Stigma receptivity duration was 5–8 days in the 7 tested genotypes. For most genotypes, the stigmas began to be receptive 1 day after anthesis, and all genotypes had receptive stigmas at 2 days after anthesis. The duration of stigma receptivity was long, and at the late stage of flowering, it was not suitable for pollination. This latter finding reflects the facts that the stigma is more prone to falling off after long flowering times and that there might not be sufficient time available for pollen germination on the stigma.

The clarified ovaries of Lilium showed the presence of fritillaria-type embryo sac development (Maheshwari 1950) with eight cells: three antipodals, two synergids, one egg cell, and two polar nuclei (one small (n) and one large (3n)) (Fig. 2). We found the antipodal lifespan to be very short, and the cells easily disintegrated in the mature embryo sac. Furthermore, the haploid and triploid polar nuclei fused into a central cell in the mature embryo sacs only after pollination. Janson and Willemse (1995) also found that the polar nuclei did not fuse until fertilization in L. longiflorum. In the present research, six of 7 genotypes had mature embryo sacs on the first flowering day, and all tested genotypes had mature embryo sacs at the time at which the stigmas first became receptive. The timing of stigma receptivity initiation, the time of highest receptivity, and the duration of stigma receptivity were genotype dependent (Table 3). Therefore, the benzidine-H2O2 method was useful for determining the time of stigma receptivity and can thus facilitate successful controlled pollination. However, the solution is toxic, the procedure is time intensive, and stigma receptivity may be affected by external environmental conditions.

Therefore, we attempted to find a morphological index related to stigma receptivity. In practice, a wet secretion was observed on the stigma surface during anthesis in the OO, OT and LL hybrids and some of the wild species; this secretion was colorless, shiny and sticky. In Melaleuca alternifolia, the secretion is yellow in color (Baskorowati 2009). Therefore, it is possible that the color of the secretion depends on the species. In L. longiflorum, a thick layer of exudate was observed on the top of stigma at 2 days after anthesis (Janson et al. 1994). Similarly, in our study, the secretion began to peak on the stigma surface at the same flowering time in ‘Baiguang No. 3’ (a L. longiflorum hybrid). In Melaleuca alternifolia, the appearance of the secretion indicated that the stigma was receptive (Baskorowati 2009). In our study, its appearance was also a sign of stigma receptivity. When the secretion began to appear on the surface of the stigma, it indicated that the stigma was receptive or would soon be receptive, and when the secretion peaked, the stigma was or would soon be at its most receptive. Therefore, for genotypes with stigma secretion in Lilium, the optimal pollination time can be determined based on the amount of secretion.