American Journal of Botany 88(7): 1230–1239. 2001.

BIOGEOGRAPHY AND ORIGIN OF LILIUM LONGIFLORUM

AND L. FORMOSANUM (LILIACEAE) ENDEMIC TO THE
RYUKYU ARCHIPELAGO AND TAIWAN AS DETERMINED
BY ALLOZYME DIVERSITY1

MICHIKAZU HIRAMATSU,2,5 KAORI II,3 HIROSHI OKUBO,3

KUANG LIANG HUANG,4 AND CHI WEI HUANG4
2
Laboratory of Agricultural Ecology, Faculty of Agriculture, Graduate School, Kyushu University, Kasuya 811-2307 Japan;
3
Laboratory of Horticultural Science, Faculty of Agriculture, Graduate School, Kyushu University, Fukuoka 812-8581, Japan; and
4
Department of Horticulture, National Chiayi University, Chiayi, Taiwan, Republic of China

Allozyme diversity on 13 isozyme loci was investigated for two bulbous species, Lilium longiflorum and L. formosanum, endemic
to the subtropical archipelago of continental origin located in East Asia. Degrees of allozyme variability and divergence for L.
longiflorum were very high for insular endemic species, indicating relatively longtime persistence of the present widespread distribution
across many islands in this phenotypically little-changed species. Lilium formosanum exhibited rather lower variability and divergence
than did L. longiflorum and was genetically close to the southern peripheral populations of L. longiflorum with 0.978 as its highest
genetic identity value. Combined with other biological and insular geohistorical information, our results suggest that L. longiflorum
was established around the end of the Pliocene when the current distribution area was still a continuous part of the ancient Asian
continent, and L. formosanum was derived from southern populations of L. longiflorum around the late Pleistocene when the mainland
of Taiwan was completely separated from the adjacent islands and the main continent. Depauperization of allozyme variability in some
L. longiflorum populations was found on islands with lower altitudes. This reflects bottleneck effects after the complete or almost
complete submergence of such low islands during the archipelago’s development.

Key words: allozyme diversity; biogeography; continental island; insular endemic; Liliaceae; Lilium; Taiwan; the Ryukyu Ar-
chipelago.

Island biotas have attracted evolutionary biologists since the
time of Darwin (see Adsersen, 1995 as a review). A basic
theory holds that genetic depauperization of organisms’ pop-
ulations and genetic differentiation among them, respectively,
are promoted by ‘‘bottlenecks’’ created by diminishing popu-
lation size and restricted gene flow from and into outer pop-
ulations (Wright, 1931, 1943; Hartl, 1981; Kimura, 1983; Nei,
1987). Thus, it can be easily assumed that disjunct geographic
distributions in islands, along with factors regarding migration
(the founder principle) and isolation, greatly affect an organ-
ism’s genetic structure in terms of its association with histor-
ical events. In fact, information on genetic structure obtained
using molecular techniques has accumulated recently in regard
to flowering plants, further clarifying the way current biogeo-
graphic structures have been established in relation to histor-
ical events in islands (e.g., Witter and Carr, 1988; Inoue and
Kawahara, 1990; Crawford et al., 1992; DeJoode and Wendel,
1992; Westerberg and Saura, 1994; Kim et al., 1996; Weller,
Sakai, and Straub, 1996; Affre, Thompson, and Debussche,
1997; Ito, 1998; Baldwin et al., 1999; Kim et al., 1999).
These intensive studies simultaneously generated important
findings, which are now widely accepted regarding the popu-
lation and evolutionary biology of insular plants. First, within
an isolated island environment, evolution expressing extraor-
dinary morphological and ecological divergence, namely,
1
Manuscript received 25 April 2000; revision accepted 21 December 2000.
The authors thank to M. Maki, Tohoku University, and Y. Ozaki, Kyushu
University, for their comments on an early version of the manuscript. This
work was supported in part by a Grant-in-Aid for Encouragement of Young
Scientists from the Ministry of Education, Science, Sports and Culture of
Japan (No. 09760036).
5
Author for correspondence.
1230
adaptive radiation, can frequently occur with relatively little
molecular divergence within a short time (see Crawford, 1990,
for a review). Second, endemic island taxa often possess rel-
atively limited amounts of genetic variation (see DeJoode and
Wendel, 1992; Weller, Sakai, and Straub, 1996; Frankham,
1997, Gemmill et al., 1998 for reviews), though some re-
searchers recently exhibited notable exceptions (Weller, Sakai,
and Straub, 1996; Francisco-Ortega et al., 2000). These find-
ings, however, have been concerned mostly with the taxa en-
demic to oceanic islands, upon which organisms are estab-
lished only through migration events from remote continents.
The archipelago running from Ryukyu to Taiwan consists
of nearly 200 islands forming an arc-array in the subtropical
area between Kyushu, the southwestern district of mainland
Japan, and the southeastern part of China. Since this archi-
pelago is considered geologically to be of continental origin
(Kimura, 1996), unlike oceanic islands such as Hawaii, its bi-
ota is largely comprised of relict taxa, which presumably dif-
ferentiated from their relatives in the adjacent continent or
mainland (Kizaki and Oshiro, 1977). Thus, comparative phy-
logeographic study between biota in this archipelago and that
in oceanic islands may potentially provide significant new in-
sights into island biology. Studies focusing on combinations
of genetic population structures in relation to the biogeography
of relict organisms in the Ryukyu Archipelago and Taiwan,
however, have almost entirely been devoted to animals such
as Plecoglossus altivelis (Nishida, 1985), Iriomote cats (Ma-
suda and Yoshida, 1995), Gekko hokouensis (Toda, Hikida,
and Ota, 1997), wood-feeding cockroaches (Maekawa et al.,
1999), Indian rice frogs (Toda, 1999), and pit vipers (Toda et
al., 1999). To our knowledge no such studies have been con-
ducted regarding plants.
July 2001] HIRAMATSU ET AL.—BIOGEOGRAPHY AND ORIGIN OF
Fig. 1. Geographic distribution of the Lilium longiflorum (V) and L. formosanum (□) populations studied. Closed areas of a small map are enlarged.
Abbreviations of population names are the same as those in Table 1.
Lilium longiflorum and L. formosanum are bulbous plants
of the Liliaceae endemic to this archipelago and are widely
regarded as species of great importance for world horticulture
(Miller, 1993; Okazaki, 1996; McRae, 1998). Both species
have been taxonomically classified into the subsection of the
section Leucolirion by Comber (1949), and their interspecific
fertile hybrid cultivars, L. 3 formolongi imply that the two
species are genetically close. Lilium longiflorum is geograph-
ically distributed from the northernmost islands of the Ryukyu
Archipelago to the mainland seacoast and to small islands in
the eastern part of Taiwan, exhibiting a disjunct distribution
following a pattern of arc-arrayed steppingstones (Wilson,
1925; Shimizu, 1987). On the other hand, L. formosanum is
natively distributed solely, but widely, within the mainland of
Taiwan (Wilson, 1925; Shii, 1983). The combined distribution
of the two species thus covers the entire archipelago across
many islands within an ;1300 km range.
Based on the abovementioned characteristics regarding
study sites and plants, we expected that an analysis of the
genetic structure of these species would be significant not only
in terms of what it might reveal regarding the phylogenetic
relationship of the two species, but also for verifying this re-
lationship’s association with the historical geography in the
Ryukyu to Taiwan archipelago arc and the widely accepted
generalization regarding insular evolutionary biology assessed
LILIUM 1231
mostly in oceanic islands. Allozyme analysis is often em-
ployed for studying types of microevolution such as speciation
and conspecific population differentiation (Crawford, 1990).
Thus, we estimated allozyme diversity of L. longiflorum and
L. formosanum in order to address the following questions: (1)
When and how are the species established? (2) Does the ge-
netic structure of their present populations reflect their ecolog-
ical nature and/or the historical geography of the archipelago?
(3) Are there any properties regarding allozyme diversity ex-
pressed by other insular plant taxa?
MATERIALS AND METHODS
Plant materials—The almost entire native distribution of 19 natural pop-
ulations of L. longiflorum and eight natural populations of L. formosanum
were studied (Fig. 1, Table 1). These two species are relatively similar in
terms of their appearance but are easily distinguishable by their leaf mor-
phology: Lilium formosanum has willowy leaves, which are narrower and
longer than those of L. longiflorum (Wilson, 1925; M. Hiramatsu, unpublished
data). Capsules were collected in the study sites, and plant materials for en-
zyme analysis were grown in a greenhouse from air-dried seeds. A single
progeny individual from each maternal genotype was used for the analysis.
Samples of each population comprised 10–55 individuals (33 for L. longiflo-
rum and 25 for L. formosanum, on average).
Isozyme electrophoresis—Approximately 200 mg of young fresh leaf sam-
1232 AMERICAN JOURNAL OF BOTANY [Vol. 88

TABLE 1. Study sites of Lilium longiflorum and L. formosanum populations indicated with number of individuals examined, population size (S,
,50; M, ,500; L, .500), population altitude (m), and maximum island peak altitude (m).

No. of Maximum
Population individuals Population Population island peak
Species abbreviation Locality examined size altitude (m) altitude (m)

L. longiflorum LYA Yaku Shima, Ryukyu 30 S ,10 1935

LKI Kikai Jima, Ryukyu 43 L ,10 224
LAM1 Amami O Shima, Ryukyu 23 L ,20 694
LAM2 Amami O Shima, Ryukyu 27 M ,20 694
LTO Tokuno Shima, Ryukyu 40 M ,10 645
LOE Okino Erabu Jima, Ryukyu 34 M ,20 246
LYR Yoron To, Ryukyu 51 L ,20 97
LOK1 Okinawa, Ryukyu 11 S ,20 498
LOK2 Okinawa, Ryukyu 13 S ,10 498
LOK3 Okinawa, Ryukyu 10 S ,10 498
LKU Kume Shima, Ryukyu 43 L ,10 310
LMI Miyako Jima, Ryukyu 29 L 20 109
LIS1 Ishigaki Jima, Ryukyu 55 L 110 526
LIS2 Ishigaki Jima, Ryukyu 53 L ,10 526
LIR Iriomote Jima, Ryukyu 37 M ,10 469
LYO Yonaguni Jima, Ryukyu 41 L ,20 231
LPI Pitouchiao, mainland of Taiwan 37 L 60 3997
LFU Fulung, mainland of Taiwan 29 S ,20 3997
LLA Lanyu, Taiwan 30 L 50 552
L. formosanum FWU Wulai, mainland of Taiwan 21 M 600 3997
FLI Lishan, mainland of Taiwan 18 M 1600 3997
FHO Hohuanshan, mainland of Taiwan 33 L 3000 3997
FWS Wushe, mainland of Taiwan 20 M 1100 3997
FSH between Shuili and Ershui, mainland of Taiwan 12 S 400 3997
FTA Tatachia, mainland of Taiwan 33 L 2600 3997
FPA Paolai, mainland of Taiwan 41 L 450 3997
FCH near Chihpen, mainland of Taiwan 22 L 700 3997

ples was placed into cooled mortars and homogenized with a pestle in 2 mL
of the Tris-HCl grinding buffer (Soltis et al., 1983) with a sprinkle of poly-
vinylpolypyrolidone and sea sand. Crude extracts were soaked by paper wicks
(11 3 3 mm), and the wicks were inserted into a slit in a starch gel.
Horizontal starch gel electrophoresis was carried out according to the pro-
cedures described by Wendel and Weeden (1989). Two combinations of gel
and electrode buffers were used to resolve 11 enzymes: aspartate aminotrans-
ferase (AAT), catarase (CAT), diaphorase (DIA), glucose-6-phosphate isom-
erase (GPI), glutamate dehydrogenase (GDH), and malic enzyme (ME) were
resolved using System 6, and fluorescent esterase (FEST), isocitrate dehydro-
genase (IDH), malate dehydrogenase (MDH), phosphoglucomutase (PGM),
and phosphogluconate dehydrogenase (6PGD) were determined using System
2. Staining protocols were also carried out according to the method of Wendel
and Weeden (1989), except for a modification for FEST by dilution of the
substrate with 1/20th volume of acetone.
Statistical analysis—Allele frequencies in each population of the species
were calculated for 13 loci encoding the 11 enzyme systems. The following
parameters concerning genetic diversity were estimated at the population and
species level in the manner described by Hamrick and Godt (1990): the pro-
portion of polymorphic loci (Pp) at a 95% criterion, the number of alleles per
polymorphic loci (Ap), the number of alleles per locus (A), and expected het-
erozygosity (h), where h was an unbiased estimate (Nei and Roychoudhury,
1974; Nei, 1978).
To estimate genetic differentiation among populations, Nei’s (1973) gene
diversity statistics, namely, total genetic diversity (HT), genetic diversity with-
in populations (HS), and proportion of the total diversity among populations
(GST), were determined. In addition, Wright’s (1951) fixation index (Fis) was
estimated at each polymorphic locus as unbiased following the method of Nei
and Chesser (1983). Chi-square analyses were performed to determine the
heterogeneity of allelic frequencies among populations (Workman and Nis-
wander, 1970) and to determine deviations from genotypic frequencies ex-
pected under the Hardy-Weinberg equilibrium (Li and Horvitz, 1953).
Unbiased genetic identity and genetic distance were calculated based on
allele frequencies in accordance with the formula derived by Nei (1978). The
resulting distance matrix among all populations of the two species was then
used to construct a neighbor-joining tree (Saitou and Nei, 1987) using the
NEIGHBOR and DRAWTREE programs of PHYLIP (Felsenstein, 1993).
RESULTS
Genetic variability at the species level—Thirteen loci listed
in Table 2 were consistently resolved. Of a total of 48 alleles
detected across the two species, 14 (Aat-2c, Aat-3c, Cat-1b,
Fest-2a, Fest-2d, Gdh-1a, Gpi-2f, Mdh-1a, 6Pgd-1a, 6Pgd-1d,
6Pgd-2a, Pgm-1a, Pgm-1b, and Pgm-1e) and 3 (Fest-2b, Fest-
2g, and Gpi-2e), respectively, were unique for L. longiflorum
and L. formosanum. In other words, 31 alleles (65%) were
common to both species. The most frequent alleles of each
species were the same for ten loci (Aat-2b, Aat-3b, Cat-1a, Dia-
1b, Gdh-1c, Idh-2a, Mdh-1c, Me-1a, 6Pgd-1c, and 6Pgd-2c),
whereas they were different for the remaining three loci (Gpi-
2d, Fest-2c, and Pgm-1c in L. longiflorum and Gpi-2c, Fest-2f,
and Pgm-1d in L. formosanum). The two species were not dis-
tinguishable by the different alleles, because they were not
species specific.
All (100%) of 13 resolved loci were polymorphic in at least
one population in L. longiflorum, whereas 10 (77%) of 13 loci
were polymorphic in L. formosanum (Table 3), in which Aat-
2, Cat-1, and Idh-2 were judged monomorphic (Table 2). The
other genetic diversity parameters, A, Ap, and h for L. longi-
florum were ;1.2–2.2 times larger than those for L. formo-
sanum.
Genetic variability at the population level—Mean values of
Pp, A, Ap, and h at the population level for L. longiflorum were
also substantially higher than those for L. formosanum (Table
July 2001] HIRAMATSU ET AL.—BIOGEOGRAPHY AND ORIGIN OF LILIUM 1233

TABLE 2. Allele frequencies for 13 isozyme loci summarized in Lilium heterozygotes was observed (Table 3). Fifty-five loci (44%)
longiflorum and L. formosanum. out of 119 loci tested for L. longiflorum showed significant
Locus Allele L. longiflorum L. formosanum
deviation from 0. Relatively high frequencies of the deviated
loci within a population occurred in LYA, LAM1, LAM2,
Aat-2 a 0.044 0.003 LOE, LYR, LOK2, and LOK3, which are located in the rel-
b 0.947 0.997 atively northern part of the archipelago. For L. formosanum,
c 0.008 0.000
Aat-3 a 0.097 0.024 ten loci (30%) out of 33 loci tested were significant.
b 0.902 0.976 Chi-square analyses for heterogeneity indicated significant
c 0.002 0.000 (P , 0.01) allele frequency differences among populations in
Cat-1 a 0.996 1.000 all and seven loci for L. longiflorum and L. formosanum, re-
b 0.004 0.000 spectively (Table 4). On average, the indices of genetic dif-
Dia-1 a 0.015 0.010 ferentiation (GST) were prominently different between the two
b 0.839 0.959
c 0.146 0.031 species. The total gene diversity was moderately (35%) ap-
Gdh-1 a 0.143 0.000 portioned among populations of L. longiflorum, whereas the
b 0.763 0.990 majority (92%) was apportioned within populations of L. for-
c 0.095 0.010 mosanum.
Gpi-2 a 0.020 0.016 Nei’s (1978) unbiased genetic identity (I) and standard ge-
b 0.287 0.343 netic distance (D) values within and between species are sum-
c 0.101 0.628
d 0.530 0.002 marized in Table 5. The I values between populations within
e 0.000 0.011 L. longiflorum ranged widely from 0.592 to 1.000 with the
f 0.062 0.000 mean of 0.850, whereas those within L. formosanum ranged
Fest-2 a 0.167 0.000 much more narrowly from 0.946 to 0.997 with the mean of
b 0.000 0.019 0.977.
c 0.479 0.113 The high correlation between genetic and geographic dis-
d 0.172 0.000
e 0.093 0.289 tance among all populations (r 5 0.791; P , 0.001) was de-
f 0.088 0.551 tected. Thus, branches of a neighbor-joining tree were com-
g 0.000 0.028 bined nearly in the geographic order, showing that the popu-
Idh-2 a 0.974 0.994 lations of L. formosanum was clustered with southernmost L.
b 0.026 0.006 longiflorum populations in Taiwan with the highest I value
Mdh-1 a 0.200 0.000
b 0.013 0.011
(0.978) between LFU and FSH (Fig. 2, Table 5). The neighbor-
c 0.787 0.989 joining tree roughly generated four major clusters: (1) LYA
Me-1 a 0.981 0.969 and LKI, (2) LOK1, LAM1, LAM2, and LTO, (3) LOK2,
b 0.019 0.031 LOK3, LOE, LYR, LKU, LMI, LIS1, LIS2, LIR, and LYO,
6Pgd-1 a 0.024 0.000 and (4) the remainder.
b 0.107 0.086
c 0.846 0.909
d 0.021 0.000 DISCUSSION
e 0.002 0.005
6Pgd-2 a 0.186 0.000 Allozyme diversity and origin of the two species—Lilium
b 0.069 0.063 longiflorum—Frankham (1997) demonstrated that insular en-
c 0.638 0.923 demic species tend to have lower genetic variability than the
d 0.107 0.014 continental taxa. DeJoode and Wendel (1992) summarized the
Pgm-1 a 0.102 0.000
b 0.080 0.000
allozyme variability of 55 insular endemic plant taxa, which
c 0.647 0.115 in general was seen to express relatively limited amounts of
d 0.155 0.885 variability at the species or infraspecific taxon with Pp 5 25.0
e 0.016 0.000 (0.0–57.0), A 5 1.32 (1.00–1.93), and h (HT) 5 0.064 (0.000–
0.195) as an average (range) across them. On the other hand,
notable exceptional cases indicating relatively high estimated
3). The variations of those values within L. longiflorum pop- values were reported recently. For example, in Hawaiian
ulations, however, exhibited considerably wide ranges from Schiedea and Alsinidendron, respective Pp, A, and h values
23.1 to 76.9 for Pp, 1.31 to 2.15 for A, 2.00 to 3.33 for Ap, (and ranges) averaged across 25 populations from 20 species
and 0.077 to 0.315 for h. Thus, the range of each population were 43.1 (0–88.9), 1.80 (1.00–3.00), and 0.183 (0–0.371)
diversity parameter for L. formosanum overlapped with that (Weller, Sakai, and Straub, 1996); in Hawaiian Metrosideros,
for L. longiflorum. The most highly diverged populations were the latter two values (and ranges) averaged across 14 popu-
present in Okinawa and Ishigaki Jima. It was noticeable that lations from three species were 3.0 (2.7–3.3) and 0.371
the diversity values, conspicuously those of h, for L. longiflo- (0.296–0.470), respectively (Aradhya, Mueller-Dombois, and
rum populations located on islands with relatively lower peak Ranker, 1991), and in endemics of the Canary Islands, the h
altitudes (LKI, LOE, LYR, LMI, and LYO) and on the eastern (HT) value (and range) averaged across 69 species from 18
satellite island of the mainland of Taiwan (LLA) were lower genera was 0.186 (0.000–0.456) (Francisco-Ortega et al.,
than those on their adjacent islands with higher peak altitudes. 2000). Compared with the variability values of the reported
insular taxa, those of L. longiflorum (Pp 5 100, A 5 3.46, and
Genetic population structure and intraspecific h 5 0.312 at the species level, and Pp 5 48.2, A 5 1.72, and
differentiation—Fixation indices (Fis) varied greatly among h 5 0.187 at the population level) are comparable to higher
populations of each species, although no significant excess of or the highest ones.
1234 AMERICAN JOURNAL OF BOTANY [Vol. 88

TABLE 3. Genetic diversity estimates and fixation indices of Lilium longiflorum and L. formosanum at the population and species level based on
13 loci examined. N 5 mean sample size per locus, Pp 5 percentage polymorphic loci at 95% criterion, A 5 mean number of alleles per locus,
Ap 5 mean number of alleles per polymorphic locus, h 5 mean expected heterozygosity (Nei’s [1978] unbiased estimate), Fis 5 fixation index
(unbiased estimate of Nei and Chesser [1983]). Results of tests for deviations from Hardy-Weinberg (H-W) equilibrium genotypic frequencies
are also shown (tested 5 number of loci tested out of a total of 13, ns 5 number of loci without significant [P , 0.05] deficiency from H-W
equilibrium, 1 5 number of loci with significant [P , 0.05] deficiency of homozygotes).

H-W deviations
Population name N Pp A Ap h Fis Tested ns 1

LYA 29.8 46.2 1.46 2.00 0.205 0.571 6 0 6

LKI 42.5 46.2 1.62 2.33 0.159 0.188 6 4 2
LAM1 22.8 46.2 1.62 2.33 0.170 0.465 6 1 5
LAM2 26.8 53.8 1.85 2.57 0.221 0.451 7 3 4
LTO 39.5 46.2 1.77 2.67 0.152 0.108 6 5 1
LOE 33.9 69.2 1.85 2.22 0.186 0.473 9 2 7
LYR 50.8 61.5 2.00 2.63 0.197 0.372 8 3 5
LOK1 11.0 76.9 1.92 2.20 0.299 0.291 10 8 2
LOK2 13.0 69.2 2.15 2.67 0.315 0.645 9 2 7
LOK3 9.9 69.2 1.85 2.22 0.284 0.460 9 4 5
LKU 42.9 46.2 1.77 2.67 0.200 0.119 6 4 2
LMI 28.3 30.8 1.38 2.25 0.160 0.086 4 4 0
LIS1 54.4 46.2 2.08 3.33 0.208 0.156 6 5 1
LIS2 53.0 53.8 2.00 2.86 0.204 20.057 7 6 1
LIR 36.8 38.5 1.69 2.80 0.162 0.219 5 3 2
LYO 40.5 30.8 1.38 2.25 0.104 0.077 4 3 1
LPI 36.9 30.8 1.46 2.50 0.134 0.042 4 4 0
LFU 28.8 30.8 1.54 2.75 0.123 20.014 4 4 0
LLA 30.0 23.1 1.31 2.33 0.077 0.291 3 2 1
Mean of L. longiflorum 33.3 48.2 1.72 2.50 0.187
L. longiflorum at the species level 100.0 3.46 3.46 0.312
FWU 21.0 7.7 1.08 2.00 0.044 0.901 1 0 1
FLI 17.9 23.1 1.23 2.00 0.071 0.483 3 2 1
FHO 33.0 30.8 1.46 2.50 0.114 0.347 4 1 3
FWS 20.0 30.8 1.38 2.25 0.149 0.331 4 3 1
FSH 12.0 46.2 1.54 2.17 0.122 0.032 6 6 0
FTA 33.0 46.2 1.77 2.67 0.170 0.228 6 4 2
FPA 40.6 38.5 1.54 2.40 0.175 0.043 5 5 0
FCH 22.0 30.8 1.46 2.50 0.128 0.365 4 2 2
Mean of L. formosanum 24.9 31.7 1.43 2.31 0.121
L. formosanum at the species level 76.9 2.46 2.90 0.142

The both species-level and population-level allozyme vari-

ability of L. longiflorum exceeded that averaged across the
species in various ecological categories, which mainly com-
prised continental species, such as monocotyledonous species
TABLE 4. Gene diversity statistics (Nei and Chesser, 1983) for 13 iso-
zyme loci in Lilium longiflorum and L. formasanum. HT 5 total
(Pp 5 59.2, A 5 2.38, and h 5 0.181 at the species level for
gene diversity, HS 5 gene diversity within population, GST 5 pro- 111 species, and Pp 5 40.3, A 5 1.66, and h 5 0.144 at the
portion of the total gene diversity among populations. population level for 80 species), endemic species (Pp 5 40.0,
A 5 1.80, and h 5 0.096 at the species level for 81 species,
L. longiflorum L. formosanum and Pp 5 26.3, A 5 1.39, and h 5 0.063 at the population
Locus HT HS GSTa HT HS GSTa level for 100 species) and outcrossing animal-pollinated spe-
Aat-2 0.100 0.034 0.662** 0.006 0.006 0.000 cies (Pp 5 50.1, A 5 1.99, and h 5 0.167 at the species level
Aat-3 0.178 0.107 0.397** 0.048 0.047 0.018 for 172 species, and Pp 5 35.9, A 5 1.54, and h 5 0.124 at
Cat-1 0.009 0.008 0.039** 0.000 0.000 0.000 the population level for 164 species) (Hamrick and Godt,
Dia-1 0.274 0.210 0.233** 0.080 0.074 0.072**
Fest-2 0.697 0.436 0.374** 0.600 0.445 0.259**
1990), though the wide range of variations was observed
Gdh-1 0.390 0.245 0.370** 0.021 0.020 0.056** among population-level values of L. longiflorum. Among Lil-
Gpi-2 0.623 0.373 0.402** 0.489 0.448 0.084** iaceous species, the species-level allozyme variability in L.
Idh-2 0.050 0.045 0.102** 0.011 0.011 0.021 longiflorum is almost comparable to the highest one reported
Mdh-1 0.340 0.203 0.405** 0.021 0.021 0.013 in Hemerocallis hakuunensis (Pp 5 83.0, A 5 6.08, and h 5
Me-1 0.038 0.025 0.337** 0.061 0.048 0.206**
6Pgd-1 0.272 0.220 0.192** 0.167 0.157 0.060**
0.279), which is native to the Korean Peninsula and a bulbous
6Pgd-2 0.543 0.202 0.628** 0.144 0.116 0.199** plant with similar life history to L. longiflorum (Kang and
Pgm-1 0.540 0.334 0.381** 0.204 0.198 0.029 Chung, 1997). Thus, L. longiflorum is a plant species with
Mean 0.312 0.188 0.348 0.142 0.122 0.078 remarkably high allozyme variability.
a Chi-square analysis of heterogeneity of allelic frequency among Comparison in genetic identity value (I) within a species
populations: ** P , 0.01. also revealed that L. longiflorum is highly diverged as a single
July 2001] HIRAMATSU ET AL.—BIOGEOGRAPHY AND ORIGIN OF
TABLE 5. Summaries of Nei’s (1978) unbiased genetic identity (I) and standard genetic distance (D) within Lilium longiflorum and L. formosanum
and between them.
I D
Species or pair of species Mean
Within species
L. longiflorum 0.850
L. formosanum 0.977
Between species
L. longiflorum vs. L. formosanum 0.816
species (see Crawford, 1990, for a review). The minimum I
value between populations within L. longiflorum (0.592),
which was recorded between northernmost (LYA) and south-
ernmost (LLA) populations, has rarely been reported within
the flowering plant. For example, extensive minimum I values
have been reported within a selfing insular species, Bidens
discoidea (0.688; Roberts, 1983), in Limnanthes floccosa
(0.575; McNeill and Jain, 1983), and between subspecies pairs
of Lens culinaris (0.65; Pinkas, Zamir, and Ladizinsky, 1985).
Generally, among insular plants, lower minimum I values
have been rarely recorded in very limited ‘‘congeneric’’ or
‘‘intergeneric’’ population pairs in a large complex of mor-
phologically and ecologically highly radiating taxa; e.g., Al-
sinoideae in Hawaii (0.242; Weller, Sakai, and Straub, 1996),
silversword alliance in Hawaii (0.426; Witter and Carr, 1988),
woody Sonchus alliance in the Canary Islands (0.490; Kim et
al., 1999), and Robinsonia in the Juan Fernandez Islands
(0.560; Crawford et al., 1992). Unlike morphological and eco-
logical phenotypes, protein molecules such as allozymes are
assumed to evolve much more consistently because of their
neutral relationship to natural selection, as described by the
neutral theory of molecular evolution (Kimura, 1983). The fact
that the amount of allozyme divergence in L. longiflorum is
close to that in the maximally radiating insular plant taxa in-
dicates that they originated, roughly, at the same time. Nev-
ertheless, they show a great contrast in terms of phenotypic
divergence.
A large number of reports for insular plants have been con-
cerned with the highly radiating taxa in their morphological
and ecological phenotypes in combination with very little mo-
lecular divergence (see Crawford, 1989, 1990; Gemmill et al.,
1998; Ito, 1998, for reviews). The opposite pattern, in which
Fig. 2. Phenogram for 19 populations of Lilium longiflorum and eight
populations of L. formosanum constructed using a neighbor-joining method
based on Nei’s (1978) standard genetic distance. Abbreviations of population
names are the same as those in Table 1.
LILIUM 1235
Range Mean Range
0.592–1.000 0.169 0.000–0.524
0.946–0.997 0.024 0.003–0.056
0.638–0.978 0.208 0.022–0.450
genetically highly diverged insular taxa showed little diver-
gence in their morphological and ecological phenotypes, was
rarely observed until the present study on L. longiflorum. Al-
though it may be difficult to conclude the reason why such
contrasting evolution occurs under ‘‘insular’’ environments, it
could be attributed to differences in the environmental and
ecological properties of oceanic and continental islands. Di-
verse ecologically unoccupied niches, which is presumably a
major factor affecting radiating evolution, are expected to be
rather few and small even at the birth of continental islands
such as the Ryukyu Archipelago and Taiwan.
Nei (1987) demonstrated a method for estimating diver-
gence time based on allozyme data, given certain assumptions
about mutation rates and the operation of a molecular clock:
time (t) 5 D/2a, where D is the standard genetic distance and
a is the substitution rate per locus per year. Usually, a is as-
sumed to be ;1027 per locus per year. Then, t may be cal-
culated as (5 3 106)D. Using this formula, the initiation of
divergence has been estimated in the aforementioned highly
radiating insular taxa, e.g., 3.6 3 106 yr ago (MYA) for the
woody Sonchus alliance in the Canary Islands (Kim et al.,
1999) and 2.9 MYA for Robinsonia spp. in the Juan Fernandez
Islands (Crawford et al., 1992). Since the maximum D within
L. longiflorum was 0.524, initiation of divergence is assumed
to be 2.62 MYA. As pointed out by Nei (1987), however, this
value may sometimes be an overestimate because the genetic
distance tends to increase when the population experiences
‘‘bottlenecks.’’ This is likely to occur under insular environ-
ments and for colonizing plants like L. longiflorum.
From the geological point of view, Kimura (1996) described
that at the end of the Tertiary Period (1.7–2.0 MYA), the area
around Ryukyu and Taiwan was a continuous coastal margin
in East Asia. The archipelago had developed during the Pleis-
tocene Era. It is, therefore, realistic to presume that L. longi-
florum existed as early as around at the end of the Tertiary
Period, and then experienced the Quaternary dynamics that
generated the current Ryukyu to Taiwan arc. The extremely
high allozyme variability and divergence in L. longiflorum pre-
sumably reflect the relict endemism with the relatively long-
time persistence of the present distribution in this species.
Lilium formosanum—Lilium formosanum possessed a subset
of L. longiflorum alleles for 11 (85%) loci and exhibited less
allozyme variability than L. longiflorum (76.9 vs. 100 for Pp
2.46 vs. 3.46 for A, and 0.142 vs. 0.312 for h at the species
level). These facts agree with previous data describing pro-
genitor-derivative species pairs (e.g., Gottlieb, 1973, 1974;
Crawford, Ornduff, and Vasey, 1985; Rieseberg et al., 1987;
Loveless and Hamrick, 1988; Pleasants and Wendel, 1989;
Maki et al., 1999). The three species-specific alleles (Fest-2b,
Fest-2f, and Gpi-2e) were detected for L. formosanum. How-
1236 AMERICAN JOURNAL
ever, such numbers of unique alleles are not uncommon in
recently derived species (0–8 in the 11 species; see Pleasants
and Wendel, 1989). The mean I value of population pairs be-
tween L. formosanum and L. longiflorum (0.816) is close to
the lowest values among those 11 progenitor-derivative species
pairs listed by Pleasants and Wendel (1989). This result can
be undoubtedly attributed to the unusually extreme genetic dif-
ferentiation within a progenitor species, L. longiflorum. Re-
stricted among populations within the mainland of Taiwan,
where the speciation between the two species presumably oc-
curred, estimation of the mean I between L. formosanum and
L. longiflorum was 0.954 (0.925–0.978). This indicates that
the two species are genetically very close in the manner typical
of progenitor-derivative species pairs. Selfing is considered a
key characteristic necessary for the rapid expansion of a spe-
cies (Maki et al., 1999), and the occurrence of selfing natures
shown by several recently derivative species such as Stephan-
omeria malheurensis (Gottlieb, 1973), Polygonella articulata,
P. americana (Lewis and Crawford, 1995), and Tricyrtis nana
(Maki et al., 1999) are very similar to the case of derivation
of self-compatible L. formosanum from self-incompatible L.
longiflorum. The accumulated evidence above demonstrates
that L. formosanum could be a recent local derivative from the
southern peripheral populations of L. longiflorum.
Naturalized populations of L. formosanum are found in veg-
etation dominated by relatively tall grasses with wide geo-
graphical ranges in often-disturbed inland areas of the main-
lands of Japan, and sometimes they comprised thousands of
individuals (M. Hiramatsu, personal observation). In contrast,
L. longiflorum populations are never seen in such contexts. In
South Africa, L. formosanum is widely naturalized under veg-
etation similar to that found in the mainlands of Japan (Wal-
ters, 1983). These facts imply that unlike L. longiflorum, L.
formosanum can be distributed rapidly and widely in adaptable
competitive and disturbed environments. Similar adaptable en-
vironments develop in adjacent regions such as the Ryukyu
Archipelago and the Chinese continent. Nevertheless, native
populations of L. formosanum persist solely within the main-
land of Taiwan. Thus, it is assumed that L. formosanum has
been prevented from migrating to adjacent regions because of
its isolation on the mainland of Taiwan prior to species initi-
ation. The isolation of the mainland of Taiwan is assumed to
have occurred during the late stage of the archipelago’s de-
velopment as early as the last glaciation at the end of the
Pleistocene Era (Kimura, 1996). Examples of recent derivative
species whose initiation times are assumed to be around the
Pleistocene glaciation include Cirsium pitcheri (Loveless and
Hamrick, 1988), Erythronium propullans (Pleasants and Wen-
del, 1989), Polygonella articulata, and P. americana (Lewis
and Crawford, 1995).
The progenitor and derivative relationship between L. lon-
giflorum and L. formosanum based on our results contradicts
the speculation made by Dubouzet and Shinoda (1999), who
demonstrated only a sister relationship between the two spe-
cies based on the internal transcribed spacer sequences of the
species’ nrDNA and regarded L. longiflorum as a species de-
rived from L. formosanum. Since our results demonstrate that
L. longiflorum is highly diverged as a single species, the ac-
curacy of resolution for the two species’ phylogenetic rela-
tionships will depend on the sample size used in the study.
Thus, the very small number of samples (presumably one for
each species) in the study by Dubouzet and Shinoda (1999) is
OF BOTANY [Vol. 88
assumed to be a cause of their inaccurate determination of the
phylogenetic relationship between the two species.
The biogeographic structure of L. longiflorum involving
insular historical events—Detailed comparisons of allozyme
diversity among L. longiflorum populations revealed biogeo-
graphic structures highly associated with the historical geog-
raphy of the Ryukyu to Taiwan archipelago arc, which has
been assumed based on geology and the biogeography of other
organisms.
First, the depauperization in allozyme variability for some
populations closely correlated with the maximum altitude of
their islands, i.e., the populations that did not exhibit as much
allozyme variability as did the adjacent island populations,
LKI, LOE, LYR, LMI, and LYO, were located on islands low-
er than 231 m (Tables 1, 3). Because the sea level was at one
time 200 m higher than that at present, lower islands were
submerged largely or completely and then pushed upward dur-
ing the late Pleistocene Era (0.4–1.0 MYA) (Kimura, 1996).
This evidence suggests that those L. longiflorum populations
had recently experienced very severe bottlenecks either by di-
minishing population size or by subsequent migration from
relict populations on adjacent islands not highly submerged
and with higher altitudes. Similarly, another substantially ge-
netically eroded population on the small but high volcanic is-
land southeast of the mainland of Taiwan (LLA) seems to have
also experienced severe bottlenecks, although the initiating
time of this island is not known. A similar geohistory-asso-
ciated biogeographic hypothesis regarding this archipelago has
been proposed to explain the mosaic distribution pattern of pit
vipers (Trimeresurus spp.), whether they are present or absent
in each island (Takara, 1962). However, no evidence has been
based on population genetic diversity until the present study.
Secondly, by excluding the islands with genetically eroded
populations (LKI, LOE, LYR, LMI, LYO, and LLA), three
major vicariant splits generating large genetic differentiation
on the neighbor-joining tree correspond to interisland splits
between Yaku Shima and Amami O Shima, between Tokuno
Shima and Okinawa, and between Iriomote Jima and the main-
land of Taiwan, though a northernmost population in Okinawa
(LOK1) is included as an exception in the geographically dif-
ferent cluster group over the splits (Figs. 1 and 2). The north-
ernmost split has long been recognized as the first strait
formed in the archipelago land bridge (Kizaki and Oshiro,
1977; Kimura, 1996) and as a vicariant border called Watase’s
Line (Kuroda, 1931; Hotta, 1974; Ono, 1989), since it corre-
sponds to distribution borders dividing Japanese biota (Kuro-
da, 1931; Inger, 1950; Hotta, 1974; Ono, 1989; Ota, 1998) and
to the Tokara Tectonic Strait (Kimura, 1996). Thus, due to the
persistence of such an old strait, the population of L. longiflo-
rum on Yaku Shima (LYA) seems to have been isolated from
the other southern populations for a long time. Whereas the
vicariant border between Tokuno Shima and Okinawa had
scarcely been recognized based on the distribution patterns of
organisms until the accumulation of recent molecular data re-
garding such species as Japanese newts, Cynops ensicauda
(Hayashi and Matsui, 1988), semi-aquatic annual ferns, Cer-
atopteris thalictroides (Watano and Masuyama, 1994), wood-
feeding cockroaches, Salganea taiwanensis (Maekawa et al.,
1999), and pit vipers, Trimeresurus flavoviridis (Toda et al.,
1999) exhibits considerable genetic differentiation between
Okinawa and Tokuno Shima. Our results together with those
regarding other organisms may suggest the possibility that an-
July 2001] HIRAMATSU ET AL.—BIOGEOGRAPHY AND ORIGIN OF
other noticeable vicariant border limiting gene flow has long
persisted between Okinawa and Tokuno Shima. At present, we
do not know why the remaining major vicariant split exists
between Iriomote Jima and the mainland of Taiwan, only that
the separation of this region is assumed to have originated
during a relatively late stage of the archipelago’s development
(Kimura, 1996).
Population structure with respect to its relation to
reproductive and breeding system, and geographic
distribution of the two species—In general, the breeding sys-
tem of flowering plant species greatly affects their GST values,
e.g., outcrossed and mixed animal-pollinated species have 39
and 42% GST values of selfing species, respectively (Hamrick
and Godt, 1990). Most Lilium species including L. longiflorum
secrete nectaries to attract pollinating insects (McRae, 1998;
M. Hiramatsu, personal observation), and L. longiflorum is
generally regarded as self-incompatible (Miller, 1993). These
facts imply that L. longiflorum is an obligate outcrossed, in-
sect-pollinated species. Nevertheless, the GST value of L. lon-
giflorum (0.348) was 77% higher than that of the mean across
124 outcrossed, animal-pollinated species (0.197) and 61%
higher across 60 mixed, animal-pollinated species (0.216)
(Hamrick and Godt, 1990). This shows conspicuously limited
gene flow between L. longiflorum populations. The distribu-
tion of L. longiflorum ranges ;1300 km between its northern-
most and southernmost populations, but is disconnected in the
manner of steppingstones (Fig. 1). Since the 19 populations
treated in the present study are located widely across 14 dif-
ferent islands, it is highly unlikely that pollen transfer by in-
sects and seed dispersal across the sea occurs for these pop-
ulations.
The frequent occurrence of the loci with significant excesses
of homozygous genotypes in some northern populations from
Okinawa (LYA, LAM1, LAM2, LOE, LYR, LOK2, and
LOK3) is also an unexpected result, because L. longiflorum is
a putative self-incompatible, outcrossed species. Because of
the lack of additional evidence, this result is difficult to inter-
pret. For the moment, it could only be said that this is either
because of the random drift of a small specimen as seen in
Okinawa (LOK2 and LOK3), the relatively restricted gene
flow of the metapopulation structure within a large population,
or possibly the lack of random mating within a population,
namely, the selfing of self-compatible individuals.
Likewise, the frequency of loci deviating significantly to-
ward an excess of homozygotes varied between populations
within L. formosanum and tended to be high in populations
with relatively low percentages of polymorphic loci (Table 3).
Lilium formosanum is generally recognized as self-compatible
(Shii, 1983; M. Hiramatsu, unpublished data). These facts thus
indicate that facultative breeding occurs in L. formosanum, i.e.,
within some populations, outcrossing dominates, while selfing
dominates within others. Selfing must play an important role,
particularly in rapidly establishing new colonies from only sin-
gle introductions, as described by Baker’s law (Baker, 1955,
1967; Stebbins, 1957).
The GST value of L. formosanum (0.078) was 60% smaller
than the mean of 124 outcrossed, animal-pollinated species
and 64% smaller than the mean of 60 mixed, animal-pollinated
species (Hamrick and Godt, 1990). This shows frequent gene
flow between populations. Unlike in L. longiflorum, L. for-
mosanum populations are distributed solely on the mainland
of Taiwan. Further, L. formosanum produces thinner seeds with
LILIUM 1237
a wider winged margin than those of L. longiflorum and has
an advantage in natural seed dispersal by wind (Shii, 1983;
McRae, 1998). Gene flow between the populations of the spe-
cies, therefore, seems to be maintained by frequently repeated
pollen flow by insects and seed dispersal by wind or human
activities within Taiwan without a major restriction by the sea.
Conservational aspects—We are confident that natural pop-
ulations of L. longiflorum and L. formosanum are gradually
diminishing because of such human activities as robbery for
horticultural purposes and developmental destruction on is-
lands, even though they are exempted from inclusion in the
red data book at present. From the viewpoint of conservation
biology, our present study is quite educational; the diminish-
ment of natural populations and genetic assimilation caused
by reckless human activities will eventually erase the natural
history written within the genes of these attractive lilies.
LITERATURE CITED
ADSERSEN, H. 1995. Research on islands: classic, recent, and prospective
approaches. In P. M. Vitousek, L. L. Loope, and H. Adsersen [eds.],
Islands: biological diversity and ecosystem function, 7–21. Springer-Ver-
lag, Berlin, Germany.
AFFRE, L., J. D. THOMPSON, AND M. DEBUSSCHE. 1997. Genetic structure of
continental and island populations of the Mediterranean endemic Cycla-
men balearicum (Primulaceae). American Journal of Botany 84: 437–
451.
ARADHYA, K. M., D. MUELLER-DOMBOIS, AND T. A. RANKER. 1991. Genetic
evidence for recent and incipient speciation in the evolution of Hawaiian
Metrosideros (Myrtaceae). Heredity 67: 129–138.
BAKER, H. G. 1955. Self-incompatibility and establishment after ‘‘long dis-
tance’’ dispersal. Evolution 9: 347–349.
———. 1967. Support for Baker’s Law as a rule. Evolution 21: 853–856.
BALDWIN, B. G., D. J. CRAWFORD, J. FRANCISCO-ORTEGA, S. C. KIM, T.
SANG, AND T. F. STUESSY. 1999. Molecular phylogenetic insights on the
origin and evolution of oceanic island plants. In P. M. Hollingsworth, R.
M. Bateman, and R. J. Gornall [eds.], Molecular systematics and plant
evolution, 400–441. Taylor and Francis, London, UK.
COMBER, H. F. 1949. A new classification of the genus Lilium. Lily Yearbook,
Royal Horticultural Society 13: 85–105.
CRAWFORD, D. J. 1989. Enzyme electrophoresis and plant systematics. In D.
E. Soltis and P. S. Soltis [eds.], Isozymes in plant biology, 146–164.
Dioscorides Press, Portland, Oregon, USA.
———. 1990. Plant molecular systematics: macromolecular approaches.
John Wiley, New York, New York, USA.
———, R. ORNDUFF, AND M. C. VASEY. 1985. Allozyme variation within
and between Lasthenia minor and its derivative species L. maritima (As-
teraceae). American Journal of Botany 72: 1177–1184.
———, T. F. STUESSY, D. W. HAINES, M. B. COSNER, M. SILVA, AND P.
LOPEZ. 1992. Allozyme diversity within and divergence among four spe-
cies of Robinsonia (Asteraceae: Senecioneae), a genus endemic to the
Juan Fernandez Islands, Chile. American Journal of Botany 79: 962–
966.
DEJOODE, D. R., AND J. F. WENDEL. 1992. Genetic diversity and origin of
the Hawaiian Islands cotton, Gossypium tomentosum. American Journal
of Botany 79: 1311–1319.
DUBOUZET, J. G., AND K. SHINODA. 1999. Phylogenetic analysis of the in-
ternal transcribed spacer region of Japanese Lilium species. Theoretical
and Applied Genetics 98: 954–960.
FELSENSTEIN, J. 1993. PHYLIP (phylogeny inference package) version 3.5c.
Distributed by author. Department of Genetics, University of Washing-
ton, Seattle, Washington, USA.
FRANCISCO-ORTEGA, J., A. SANTOS-GUERRA, S.-C. KIM, AND D. J. CRAW-
FORD. 2000. Plant genetic diversity in the Canary Islands: a conservation
perspective. American Journal of Botany 87: 909–919.
FRANKHAM, R. 1997. Do island population have less genetic variation than
mainland populations? Heredity 78: 311–327.
GEMMILL, C. E. C., T. A. RANKER, D. RAGONE, S. P. PERLMAN, AND K. R.
WOOD. 1998. Conservation genetics of the endangered endemic Hawai-
1238 AMERICAN JOURNAL
ian genus Brighamia (Campanulaceae). American Journal of Botany 85:
528–539.
GOTTLIEB, L. D. 1973. Genetic differentiation, sympatric speciation, and the
origin of a diploid species of Stephanomeria. American Journal of Bot-
any 60: 545–533.
———. 1974. Genetic confirmation of the origin of Clarkia lingulata. Evo-
lution 28: 244–250.
HAMRICK, J. L., AND M. J. GODT. 1990. Allozyme diversity in plant species.
In A. H. D. Brown, M. T. Clegg, A. L. Kahler, and B. S. Weir [eds.],
Plant population genetics, breeding, and genetic resources, 43–63. Sin-
auer, Sunderland, Massachusetts, USA.
HARTL, D. L. 1981. A primer of population genetics. Sinauer, Sunderland,
Massachusetts, USA.
HAYASHI, T., AND M. MATSUI. 1988. Biochemical differentiation in Japanese
newts, genus Cynops (Salamandridae). Zoological Science 5: 1121–1136.
HOTTA, M. 1974. Evolutionary biology in plants. III. History and geography
of plants (in Japanese). Sanseido Co., Tokyo, Japan.
INGER, R. F. 1950. Distribution and speciation of the amphibians of the Ryu-
kyu Islands. American Naturalist 84: 95–115.
INOUE, K., AND T. KAWAHARA. 1990. Allozyme differentiation and genetic
structure in island and mainland Japanese populations of Campanula
punctata (Campanulaceae). American Journal of Botany 77: 1440–1448.
ITO, M. 1998. Origin and evolution of endemic plants of the Bonin (Ogasa-
wara) Islands. Researches on Population Ecology 40: 205–212.
KANG, S. S., AND M. G. CHUNG. 1997. Genetic variation and population
structure in Korean endemic species: IV. Hemerocallis hakuunensis (Lil-
iaceae). Journal of Plant Research 110: 209–217.
KIM, S. C., D. J. CRAWFORD, J. FRANCISCO-ORTEGA, AND A. SANTOS-GUER-
RA. 1996. A common origin for woody Sonchus and five related genera
in the Macronesian Islands: molecular evidence for extensive radiation.
Proceedings of the National Academy of Sciences, USA 93: 7743–7748.
———, ———, ———, AND ———. 1999. Adaptive radiation and genetic
differentiation in the woody Sonchus alliance (Asteraceae: Sonchinae) in
the Canary Islands. Plant Systematics and Evolution 215: 101–118.
KIMURA, M. 1983. The neutral theory of molecular evolution. Cambridge
University Press, Cambridge, UK.
KIMURA, M. 1996. Quaternary paleogeography of the Ryukyu Arc. Journal
of Geography 105: 259–285 (in Japanese with English abstract).
KIZAKI, K., AND I. OSHIRO. 1977. Paleogeography of the Rytkyu [sic] Islands.
Marine Science Monthly 9: 543–549 (in Japanese with English abstract).
KURODA, N. 1931. Evaluation of the Watase’s line from the view point of
vertebrate distribution. Zoological Magazine 43: 172–175 (in Japanese).
LEWIS, P. O., AND D. J. CRAWFORD. 1995. Pleistocene refugium endemics
exhibit greater allozyme diversity than widespread congeners in the ge-
nus Polygonella (Polygonaceae). American Journal of Botany 82: 141–
149.
LI, C. C., AND D. G. HORVITZ. 1953. Some methods of estimating the in-
breeding coefficient. American Journal of Human Genetics 5: 107–117.
LOVELESS, M. D., AND J. L. HAMRICK. 1988. Genetic organization and evo-
lution history in two North American species of Cirsium. Evolution 42:
254–265.
MAEKAWA, K., N. LO, O. KITADE, T. MIURA, AND T. MASUMOTO. 1999.
Molecular phylogeny and geographic distribution of wood-feeding cock-
roaches in Eastern Asian islands. Molecular Phylogenetics and Evolution
13: 360–376.
MAKI, M., H. MORITA, S. OIKI, AND H. TAKAHASHI. 1999. The effect of
geographic range and dichogamy on genetic variability and population
genetic structure in Tricyrtis section Flavae (Liliaceae). American Jour-
nal of Botany 86: 287–292.
MASUDA, R., AND M. C. YOSHIDA. 1995. Two Japanese wildcats, the Tushima
cat and Iriomote cat, show the same mitochondrial lineage as the leopard
cat Felis bengalenses. Zoological Science 12: 655–659.
MCNEILL, C. I., AND S. K. JAIN. 1983. Genetic differentiation studies and
phylogenetic inference in the plant genus Limnanthes (section Inflexae).
Theoretical and Applied Genetics 66: 257–269.
MCRAE, E. A. 1998. Lilies—a guide for growers and collectors. Timber
Press, Portland, Oregon, USA.
MILLER, W. B. 1993. Lilium longiflorum. In A. De Hertogh and M. Le Nard
[eds.], The physiology of flower bulbs, 391–422. Elsevier, Amsterdam,
The Netherlands.
NEI, M. 1973. Analysis of gene diversity in subdivided populations. Pro-
ceedings of the National Academy of Sciences, USA 70: 3321–3323.
OF BOTANY [Vol. 88
———. 1978. Estimation of average heterozygosity and genetic distance
from a small number of individuals. Genetics 89: 583–590.
———. 1987. Molecular evolutionary genetics. Columbia University Press,
New York, New York, USA.
———, AND R. K. CHESSER. 1983. Estimation of fixation indices and gene
diversities. Annals of Human Genetics 47: 253–259.
———, AND A. K. ROYCHOUDHURY. 1974. Sampling variances of hetero-
zygosity and genetic distance. Genetics 76: 379–390.
NISHIDA, M. 1985. Substantial genetic differentiation in Ayu Plecoglossus
altivelis of the Japan and Ryukyu Islands. Bulletin of the Japanese So-
ciety of Scientific Fisheries 51: 1269–1274.
OKAZAKI, K. 1996. Lilium species native to Japan, and breeding and pro-
duction of Lilium in Japan. Acta Horticulturae 414: 81–92.
ONO, M. 1989. Phytogeography in Ogasawara and Ryukyu (Nansei) Islands
(in Japanese). In A. Miyawaki [ed.], Vegetation of Japan, vol. 10, Oki-
nawa and Ogasawara, 127–137. Shibundo Co., Tokyo, Japan.
OTA, H. 1998. Geographic patterns of endemism and speciation in amphib-
ians and reptiles of the Ryukyu Archipelago, Japan, with special refer-
ence to their paleogeographical implications. Researches on Population
Ecology 40: 189–204.
PINKAS, R., D. ZAMIR, AND G. LADIZINSKY. 1985. Allozyme divergence and
evolution in the genus Lens. Plant Systematics and Evolution 151: 131–
140.
PLEASANTS, J. M., AND J. F. WENDEL. 1989. Genetic diversity in a clonal
narrow endemic, Erythronium propullans, and in its widespread progen-
itor, Erythronium albidum. American Journal of Botany 76: 1136–1151.
RIESEBERG, L. H., P. M. PETERSON, D. E. SOLTIS, AND C. R. ANNABLE. 1987.
Genetic divergence and isozyme variation among four varieties of Allium
douglasii (Alliaceae). American Journal of Botany 74: 1614–1624.
ROBERTS, M. L. 1983. Allozyme diversity in Bidens discoidea (Compositae).
Brittonia 35: 239–247.
SAITOU, N., AND M. NEI. 1987. The neighbor joining method: a new method
for reconstructing phylogenetic trees. Molecular Biology and Evolution
4: 406–425.
SHII, C. T. 1983. The distribution and variation of Lilium formosanum Wall.
and L. longiflorum Thunb. in Taiwan. Lily Year Book of the North Amer-
ican Lily Society 36: 48–51.
SHIMIZU, M. 1987. Lilium species, subspecies, and cultivars (in Japanese). In
M. Shimizu [ed.], The lilies of Japan—species and hybrids, 47–97. Sei-
bundoshinkosha Ltd., Tokyo, Japan.
SOLTIS, D. E., C. H. HAUFLER, D. C. DARROW, AND G. J. GASTONY. 1983.
Starch gel electrophoresis of ferns: a compilation of grinding buffers, gel
and electrode buffers, and staining schedules. American Fern Journal
73: 9–27.
STEBBINS, G. L. 1957. Self fertilization and population variability in the high-
er plants. American Naturalist 9: 337–354.
TAKARA, T. 1962. Studies on the terrestrial snakes in the Ryukyu Archipel-
ago. Science Bulletin of Agriculture and Home Economy Division, Uni-
versity of the Ryukyus 9: 1–102 (in Japanese with English abstract).
TODA, M. 1999. Historical biogeography of East Asian populations of Rana
limnocharis (Amphibia: Anura): a review. In H. Ota [ed.], Tropical island
herpetofauna: origin, current diversity, and conservation, 299–315. El-
sevier Science B. V., Amsterdam, The Netherlands.
———, T. HIKIDA, AND H. OTA. 1997. Genetic variation among insular pop-
ulations of Gekko hokouensis (Reptilia: Squamata) near the northeastern
borders of the Oriental and Palearctic zoogeographic regions in the north-
ern Ryukyus, Japan. Zoological Science 14: 859–867.
———, M. NISHIDA, M.-C. TU, T. HIKIDA, AND H. OTA. 1999. Genetic
variation, phylogeny and biogeography of the pit vipers of the genus
Trimeresurus sensu lato (Reptilia: Viperidae) in the subtropical East
Asian islands. In H. Ota [ed.], Tropical island herpetofauna: origin, cur-
rent diversity, and conservation, 249–270. Elsevier Science B. V., Am-
sterdam, The Netherlands.
WALTERS, G. 1983. Naturalization of Lilium formosanum in South Africa.
Lily Year Book of the North American Lily Society 36: 44–47.
WATANO, Y., AND S. MASUYAMA. 1994. Genetic differentiation in popula-
tions of the polymorphic fern Ceratopteris thalictroides in Japan. Journal
of Plant Research 107: 139–146.
WELLER, S. G., A. K. SAKAI, AND C. STRAUB. 1996. Allozyme diversity and
genetic identity in Schiedea and Alsinidendron (Caryophyllaceae: Alsi-
noideae) in the Hawaiian Islands. Evolution 50: 23–34.
July 2001] HIRAMATSU ET AL.—BIOGEOGRAPHY AND ORIGIN OF
WENDEL, J. F., AND N. F. WEEDEN. 1989. Visualization and interpretation of
plant isozymes. In D. E. Soltis and P. S. Soltis [eds.], Isozymes in plant
biology, 5–45. Dioscorides Press, Portland, Oregon, USA.
WESTERBERG, A., AND A. SAURA. 1994. Genetic differentiation in endemic
Silene (Caryophyllaceae) on the Hawaiian Islands. American Journal of
Botany 81: 1487–1493.
WILSON, E. H. 1925. The lilies of eastern Asia. Dulau and Company, London,
UK.
WITTER, M. S., AND G. D. CARR. 1988. Adaptive radiation and genetic dif-
LILIUM 1239
ferentiation in the Hawaiian silversword alliance (Compositae: Madi-
inae). Evolution 42: 1278–1287.
WORKMAN, P. L., AND J. D. NISWANDER. 1970. Population studies on south-
western Indian tribes. II. Local genetic differentiation in the Papago.
American Journal of Human Genetics 22: 24–49.
WRIGHT, S. 1931. Evolution in Mendelian populations. Genetics 16: 97–159.
———. 1943. Isolation by distance. Genetics 28: 114–138.
———. 1951. The genetical structure of populations. Annals of Eugenics
15: 323–354.