Liliaceous species include many important ornamental crops such as tulips, lilies, hyacinths, crocuses, narcissus and alstroemeria. We have been selecting valuable varieties over a long period of time and crops have been improved remarkably by using different breeding techniques after the rediscovery of Mende's Theory. In tulip, many cultivars have been developed through conventional cross - breeding in Holland since 16th century. In the 1920, tulips were introduced to the northern part of Japan, in Toyama and Niigata prefectures, and Japanese original tulips have been developed. Recently attempts have been made to apply mutation breeding for tulip.
In Japan, a large number of wild native Lilium species grow and are also important as ornamentals. Lilium species became domesticated and have been produced in the prefectures of Japan. The exportation of lily bulbs was an important means of obtaining foreign currency about one hundred years ago. During these years, bulbs were dug in native habitats and exported. Through hybridization among these native species, old cultivars of lilies had developed about three hundred years ago and currently modern important cultivars, 'Asiatic Hybrids' and 'Oriental Hybrids', have been developed. Cut-style pollination and embryo culture played an important role in overcoming interspecific incompatibility and recently, attempts have also been made to apply the mentor pollen technique to overcome interspecific barriers.
In gladioli, bulbs are produced in Ibaraki prefecture. Mutation breeding has been conducted at the Ibaraki Horticultural research station. The culture of corms irradiated with gamma rays leads to a high frequency of variants and production of solid mutants without chimeras.
The objective of this paper is to briefly review the breeding of liliaceous ornamental crops applied by using several techniques, i.e. mutation, interspecific crosses and polyploidy.

General aspects

In tulip, extensive studies on mutation breeding were conducted in Japan, especially in Toyama prefecture (Myodo 1942, Nezu 1962, 1963a,b, 1964, 1965, Nezu and Obata 1964a, b). The effect of acute and chronic irradiation on the growth of tulip plants was examined (Nezu and Obata 1964a). The bulbs were irradiated in two ways; namely, using a total amount of 10 Gy with dose rates ranging from 0.1 Gy to 2 Gy /h, and by chronic 0.068 Gy/h irradiation the total amount ranged from 20 to 80 Gy (Fig.1, Table 1).

The problem is when to irradiate. To answer this question, we need to consider the life cycle of the tulip bulb. When the floral meristem was irradiated at earlier developmental stage after harvest in June, a larger mutated spot appeared on the perianth in the next spring. In summer, the flower meristem gradually grows up in the bulb and the morphogenesis is completed in September. When bulbs were irradiated in September, small splashed spots were consequently observed on the perianth in the next spring (Nezu 1963). The changes in the flower color observed in the first generation after gamma irradiation were due to the presence of mericlinal chimera and this mutated sector in the perianth was not transferred to the next generation. A larger mutated area spreading over 2 perianths was observed after the second generation.
It is important that the mutation of gene, which controls flower color, occurs in the vegetative cell line forming the next new bulb. When the secondary bulblets are formed in the bulb in autumn, the apices in the secondary bulblet are suitable for irradiation and the selection should be started after two or three generations to give a mutated cell the opportunity to express itself in the complete mutant.
Flower color mutation, which can be easily monitored in the subsequent generation after mutagenic treatment, was promoted in Toyama prefecture, but new mutant varieties have not been developed in Japan, yet. Flower color mutation, especially in the top tulip varieties, exerts a considerable impact to customers and growers. However, the ranking of top varieties often changes and it is difficult to anticipate which will be the leading variety in the future. In tulip, since it takes about 15 years to propagate a new mutant practically, the application of mutation breeding is delayed

L. formosanum can be easily propagated through seeds and has the ability to flower within one year after the sowing of seeds. Nishimura, in Nagano, started crossing L. formosanum with L. longiflorum in about 1928 and then developed L. x formolongi which combines the characteristics of flowering within one year after of sowing the seeds with the presence of broad leaves like in L. longiflorum. To obtain the configuration of L. longiflorum, L. x formolongi has been backcrossed to L. longiflorum in the recent varieties. As L. x formolongi is propagated by seeds, viral infection does not occur. The other advantage is that cut flowers of L. x formolongi can be produced from July to November, which it is difficult to produce the cut flowers of L. longiflorum in these months. It has been estimated that about 15 million cut flowers of L. x formolongi are being produced. The trend of lily breeding and production in Japan was reviewed by Okazaki (1996).
Two major horticultural lily groups are referred to as 'Asiatic Hybrids' and 'Oriental Hybrids'. The former are derived from species such as Lilium maculatum, L. dauricum, L. lancifolium, L. maximowiczii, etc. The latter are derived from such species as L. speciosum, L. auratum, etc. Since these hybrid groups were obtained from crosses between closely related species, sexual reproduction barriers between interspecific crosses were relatively low and hybrids could be easily obtained through conventional cross - breeding.
Embryo culture has been found to be useful to overcome interspecific incompatibility caused by insufficient endosperm formation in lilies (Nakajima 1940, Emsweller et al. 1962, North and Wills 1969, Ronald and Ascher 1976). Recently attempts have been made to develop crosses between distantly related species to obtain newer interspecific hybrids (Asano and Myodo 1977a,b, Van Tuly 1991, Okazaki et al. 1992, 1994). Cut-style pollination is suitable for overcoming interspecific pre-zygotic barriers which inhibit the elongation of pollen tubes in the style.
Asano and Myodo (1977) showed that MS medium is suitable for the culture of immature embryos when it is adjusted to pH 5.0 and supplemented with 20 - 40 g/l of10^-4 - 10^-2 mg/l NAA. Okazaki et al. (1994) adjusted the optimum sucrose concentration to 6%. The techniques of nurse culture (Asano 1980) and ovary slice and ovule culture (Hayashi et al 1986, Kanoh et al. 1988, Van Tuyl et al. 1991) are also useful for overcoming interspecific barriers.

Kunishige and Hirata (1972) and van Tuyl et al. (1982) applied the mentor pollen technique to overcome interspecific pre-zygotic barriers in Lilium. The mentor effect is due to pollination with mixtures of compatible (mentor) and incompatible pollen. To avoid prior fertilization by the compatible mentor pollen, mentor pollen is killed by a high radiation dose. The irradiated pollen is unable to induce the development of fertilized eggs, but is capable of elongating the pollen tube. In lily, when pollen was irradiated for ten hours with dose rate ranging from 3 to 20 Gy/h, the occurrence of seed set gradually decreased when the dose increased (Table 2) and the optimum dose was consider to be 100Gy to obtain pollen with a mentor effect. Similarly, Van Tuyl et al. (1982) showed that lily pollen irradiated at 100 Gy was much more effective in stimulating the fruit set than pollen exposed to a dose of 250 Gy. By applying the mentor pollen technique in a cross of L. longiflorum x L. 'Asiatic Hybrid', in which the seed set is absent in a normal cross, interspecific hybrids were obtained by using irradiated L. longflorum pollen as a mentor (Van Tuyl et al, 1982).
Mentor pollen techniques have been successfully applied to overcome incompatibility as reported by Stettler (1968) in Populus, Dayton (1974) in apple, Den Nijs & Oost (1980) in Cucumis, Visser (1981) in apple and pear and Ureshino et al. (2000) in azaleas. Recently, Visser (1981) has observed that the seed set after pollination with compatible pollen was considerably improved by applying the pollen twice, with an interval of 1- 2 days. The pollen in the first application was referred to as 'pioneer pollen' as it appeared to promote the activity of the pollen in the second application. It remains to be determined whether there is a difference in the mentor effect between the mentor and pioneer pollen techniques.

In ornamental plants, a large number of polyploid cultivars have been developed, because they display horticulturally desirable characteristics such as good bulb production and larger flowers. In Easter lily, Lilium longiflorum, successful induction of polypoids by colchicine was reported (Emsweller and Lumsden, 1943; Emsweller ,1949; Emsweller and Uhring, 1960). These reports indicated that tetraploid Easter lily had larger but fewer flowers and thicker leaves and bloomed later than the diploid form. Characters such as reduced flower number were improved in the progeny obtained from crosses among tetraploid individuals. In addition, the range of variation occurring in the tetraploid seedlings was wider than that in any diploid populations. Thus breeding of Easter lily on the tetraploid level is considered to be useful, although tetraploid commercial varieties of Easter lily have not yet been developed.
In the case of Lilium 'Asiatic Hybrid', commercial tetraploid varieties have been distributed to markets recently. However, it remains to be determined what kinds of changes occur with polyploidization. The tetraploid sport, which was derived from natural polyploidization, was compared with the original diploid clones under natural conditions and early forcing culture (Okazaki et al. 2002). It was observed that mean stomatal length of the original clone and its sport which was 80μm and 114μm, respectively, was significantly different by t-test. The increase of the stomatal size in the sport appeared to be due to tetraploidization. Actually, McRae (1987) observed the size of stomata in 40 varieties of L. 'Asiatic Hybrid' and its wild relatives, and demonstrated that the length of the stomatal guard cells was 67 to 83μm in the diploid and 100 to 150μm in the tetraploid clones. The size of the pollen of the sport which was 79.7 ±,0.43μm (ave. ±,SD) did not differ from that of the diploid clone. McRae (1987) reported that the pollen size ranged from 67 to 100μm in diploid lilies and 90 - 150μm in tetraploid ones. In general, a plant is composed of three cell layers, L1 - L3, and reproductive tissue is produced in the L2 layer (Raven et al. 1999). The pollen size of the sport showed that the L2 layer was diploid, and, as mentioned above, the L1 layer corresponding to the epidermis of the sport of 'Kiyotsubeni' was tetraploid. This sport consisted of periclinal chimeras with a 4n chromosome number in the L1 layer and 2n in the L2 layer.
Morphological and physiological changes associated with polyploidization were compared between the diploid clones and the tetraploid ones under natural conditions and early forcing culture. The morphological differences between the diploid and tetraploid clones showed a similar tendency under the natural and forcing conditions. Compared with the diploid clones, the tetraploid clones displayed a reduction in the number of flowers, a shorter stem and broader leaves. Under early forcing culture at a low light intensity, the stems of the tetraploid clones were softer than those of the diploid ones. Leaf scorch frequently occurred in the diploid clones (72.2%), while the rate of leaf scorch was 2.3% in the tetraploid clones. It is concluded that breeding of Lilium 'Asiatic Hybrid' at the tetraploid level is highly suitable.
This sport, which consisted of periclinal chimeras with a 4n chromosome number in the L1 layer and 2n in the L2 layer, propagated without reverting to the diploid form for a long period of time. A periclinal chimera with different polyploidy levels between layers is always stable in plants (Tilney-Bassett 1986).

In mutation breeding, the mutated sector should to be transformed into a solid mutant or a periclinal chimera. The flower color depends mainly on the genetic constitution of the L-1 layer of the flower petals and, in chrysanthemum, a mutated flower color was often obtained in different genetic constitutions between L1 and L2 layers (Shibata and Kawata 1986, Shibata et al.1998). In such periclinal mutants, it is important that periclinal chimeras are maintained as long as vegetative propagation takes place. In general, mutated cell lineage is stable in the periclinal chimeras, but not in the mericlinal chimeras. This situation is the same as that of chimeric polyploidy.
Chimeric condition will be terminated when chimeric plants propagate by seeds. For instance, when a mutation exists in L1 and is lacking in L2 and L3, this mutation cannot be transmitted to the next generation via sexual reproduction, because of the formation of gametes in the L2 layer. Actually, when the above-mentioned chimeric plant, consisting of periclinal chimeras with a 4n chromosome number in the L1 layer and 2n in the L2 layer, was crossed with a diploid parent, most of the progenies were diploid and the tetraploid level in L1 layer was not transmitted to the next generation (Okazaki unpublished), indicating that the L1 layer does not contribute to the production of gametes.
It should be noted that the radiosensitivity of cells in various layers vary in the periclinal chimeras in which the layers differ in the ploidy level (van Harten, 2002). Tetraploid cells are likely to be more resistant to irradiation than diploid cells, because doubled chromosomes in tetraploid cells compensate for chromosomal damages associated with irradiation. As a result of such effects, diploid cell layers may be replaced by tetraploid cell layers when shoot apices of periclinal chimeras in which the layers differ in ploidy level are irradiated. Actually some authors have demonstrated that radiation treatment can be used to transfer a mutation that is present only in the cells of one layer, to another layer (van Harten, 2002).

In gladioli, when the meristems were cut from the corm irradiated with gamma rays and then cultured in vitro, the plantlets were regenerated via embryogenesis and showed a high frequency of variants without chimeras (Kasumi et al. 2001). The total dose of 100 - 200 Gy at 10Gy/hr caused a 50% reduction in the rate of callus formation and somatic embryogenesis. The pink flower color in the original cultivars 'Traveler' changed to a deeper or paler pink colors in the mutated plants and a large number of plants with a deep flower color were observed compared with those with a pale flower color.
This propagation technique in combination with gamma irradiation is an effective way of producing solid mutants, since the regenerated plantlets of gladioli develop from one original cell via embryogenesis, as described in chrysanthemum (Broertjes and Roest 1976, Nagatomi et al. 1998).