Lilium longiflorum and L. formosanum are closely related species endemic to subtropical islands in the Ryukyu Archipelago and Taiwan. Stem leaf morphology, flowering rate within population and individual net production during the first year seedling growth were determined to clarify whether they can differentiate these two species, and reflect adaptive strategy during species and local population establishment. Four experimental populations of L. longiflorum and five of L. formosanum with different locality covering almost their native distribution were grown under greenhouse condition in the University Farm located in northern Kyushu, Japan. L. longiflorum showed distinct lower ratio of leaf length/width than L. formosanum. Flowering rate and net production tended to decrease along the population location, northward across the archipelago arc and toward higher altitude within the mainland of Taiwan, showing obvious geographic cline (geographically continuous variation). Low degrees of flowering rate and net production in northern L. longiflorum were associated with high frequency of individuals that obtained little net production during spring to summer, indicating an acquired dormancy status. L. formosanum native to about 3000m altitude showed higher resource allocation to the mother bulb. These variations of the growth habits within the species reflect region-specific adaptive strategy of L. longiflorum and L. formosanum for climate during the glacial period and that in highlands of the mainland of Taiwan, respectively. Higher annual net production and an earlier shift to the reproductive phase of L. formosanum is highly likely more advantageous in population establishment under disturbed and competitive environment, where L. longiflorum is rarely found. INTRODUCTION Lilium longiflorum Thunberg and L. formosanum Wallace are Liliaceous species endemic to subtropical arc-arrayed islands in the Ryukyu Archipelago and Taiwan. These species are taxonomically classified into the section Leucolirion by Wilson (1925) and Comber (1949). A study on the allozyme diversity for natural populations, which cover entire distributions of the two species, indicated that L. formosanum seems to be a recently established species from southern peripheral populations of L. longiflorum (Hiramatsu et Proc. 8th Int. Symp. on Flowerbulbs Eds. G. Littlejohn et al. Acta Hort. 570, ISHS 2002 332 al., 2000). Whereas, it has been generally considered that L. formosanum can be distinguished from L. longiflorum by its morphological and ecological characteristics such as narrow and long leaves, and rapid growth from seed-sowing to flowering (Wilson, 1925; McRae, 1998). The allozyme analysis of the two species also exhibited that L. longiflorum is genetically highly dive rged as a single plant species and that high correlation between genetic and geographic distance of two species populations (Hiramatsu, et al., 2000). From these results, we assumed that there might be great intraand interspecific variations in its morphological and ecological traits accompanied by speciation and local population establishment of L. longiflorum and L. formosanum. Thus, we investigated intraand interspecific variations in stem leaf morphology, flowering rate within population and individual net production during the first year seedling growth to verify whether they can discriminate the two species, and reflect adaptive strategy during speciation and local population establishment of them. MATERIALS AND METHODS Seeds used in the present study were collected from four and five natural populations for L. longiflorum and L. formosanum in different locations respectively, and preserved at 5°C until use. The collection sites were Pitouchiao (LPI) in the mainland of Taiwan, and Ishigaki Jima (LIS), Kume Shima (LKU) and Kikai Jima (LKI) in the Ryukyu archipelago for L. longiflorum, and Fukuoka (FFU), a naturalized population in Kyushu district, Japan, and Paolai (FPA), Wulai (FWU), Tatachia (FTA) and Hohuanshan (FHO) in the mainland of Taiwan for L. formosanum. The sites were along the population location, northward across the archipelago from 25° 08' to 28° 21' N for L. longiflorum, and toward higher altitude from 20 to 3000m for L. formosanum. The selected sites in the present study covered almost an entire natural distribution of the two species. The experimental field was located in the University Farm of northern Kyushu, Japan. The seeds were sown into plastic trays with 200 of 2 x 2cm cells filled with commercial compost in early November, 1998. The trays were placed in a nursery greenhouse where the minimum temperature was maintained higher than 10°C. In late April, the seedlings were transplanted in a soil supplemented with 150gm slow-release fertilizer comprising 10% of N, P and K with 15 x 15cm spaces in a greenhouse. More than 100 plants for each experimental population were grown. Ten and 20 plants were randomly harvested 40, 100 and 160 days after transplanting, respectively. Each harvested plant was separated into the upper part including stems, leaves and reproductive organs and the lower part including bulbs, bulblets and roots. Dry weight of each part was determined after drying for 72 h at 70°C. Number of flowered individuals was counted for each experimental population 160 days after transplanting. Length and width of a stem leaf at about 5 to 10cm above the ground were measured for maximally 20 plants per population, except for Kikai Jima of L. longiflorum, in which bolting was not observed. RESULTS AND DISCUSSION Length and width of stem leaf in L. longiflorum ranged from 53 to 185mm and 6.6 to 18.3mm, respectively; whereas, those in L. formosanum ranged from 160 to 299mm and 5.5 to 14.5mm, respectively (Fig. 1). Thus, the values overlapped the two species. The ratio of leaf length/width, however, distinctly separated the two species, permitting to distinguish them by leaf morphology. L. longiflorum possessed relatively wide and short leaves as compared with L. formosanum. Flowering rate decreased from 26 to 0% along the population location, northward across the archipelago in L. longiflorum, and from 100 to 5% toward higher altitude in L. formosanum (Fig. 2). Particularly, high flowering rate (>88%) was observed in three L. formosanum populations, FFU, FPA and FWU. The values of mean individual net production in 160 days after transplanting were associated with flowering rate within a population. Those in L. formosanum populations