On the origin of European lilies: phylogenetic analysis of Lilium section Liriotypus (Liliaceae) using sequences of the nuclear ribosomal transcribed spacers

Abstract

The sequences of the nuclear ribosomal internal transcribed spacer (ITS) region were analysed for 28 representatives of Lilium, one Nomocharis species and three outgroup taxa of Lilieae (Notholirion, Fritillaria and Cardiocrinum). 17 of the 20 members of Lilium sect. Liriotypus were included. A maximum parsimony analysis was carried out for the phylogenetic reconstruction. The results are not completely congruent with sectional delimitations of L. sect. Liriotypus based on morphological characters. They confirm the previous suggestion that L. sect. Liriotypus is monophyletic only if L. bulbiferum is excluded and placed in L. sect. Sinomartagon. The monophyly of the remaining L. sect. Liriotypus receives good support from bootstrap analysis. It can be divided into two groups, one comprising NE Turkish-Caucasian species and another the European species, L. candidum and the two Turkish endemics L. ciliatum and L. akkusianum. The results also show that L. ponticum cannot be included within the so-called L. carniolicum group of lilies.

Full text

NURSEL IKINCI, CHRISTOPH OBERPRIELER & ADIL GÜNER
On the origin of European lilies: phylogenetic analysis of Lilium sec-
tion Liriotypus (Liliaceae) using sequences of the nuclear ribosomal
transcribed spacers
Abstract
Ikinci, N., Oberprieler, C. & Güner, A.: On the origin of European lilies: phylogenetic analysis of
Lilium section Liriotypus (Liliaceae) using sequences of the nuclear ribosomal transcribed spacers.
– Willdenowia 36: 647-656. – ISSN 0511-9618; © 2006 BGBM Berlin-Dahlem
doi:10.3372/wi.36.36201 (available via http://dx.doi.org/)
The sequences of the nuclear ribosomal internal transcribed spacer (ITS) region were analysed for
28 representatives of Lilium, one Nomocharis species and three outgroup taxa of Lilieae (Notho-
lirion, Fritillaria and Cardiocrinum). 17 of the 20 members of Lilium sect. Liriotypus were in-
cluded. A maximum parsimony analysis was carried out for the phylogenetic reconstruction. The
results are not completely congruent with sectional delimitations of L. sect. Liriotypus based on
morphological characters. They confirm the previous suggestion that L. sect. Liriotypus is mono-
phyletic only if L. bulbiferum is excluded and placed in L. sect. Sinomartagon. The monophyly of
the remaining L. sect. Liriotypus receives good support from bootstrap analysis. It can be divided
into two groups, one comprising NE Turkish-Caucasian species and another the European species,
L. candidum and the two Turkish endemics L. ciliatum and L. akkusianum. The results also show that
L. ponticum cannot be included within the so-called L. carniolicum group of lilies.
Key words: ITS, phylogeny, systematics, DNA sequencing, Turkey, Caucasia.
Introduction
The genus Lilium L. includes approximately 100 species distributed mainly in temperate regions
throughout the northern hemisphere. Infrageneric classification of Lilium is subject of considerable
discussion and the number of sections used differs depending on the author. Based on flower
shape, Baker (1871) divided Lilium into four sections: L. sect. Eulirion Rchb. (funnel-flowered lil-
ies), L. sect. Archelirion Baker (open-flowered lilies), L. sect. Isolirion Baker (erect-flowered lil-
ies) and L. sect. Martagon Rchb. (Turk’s cap lilies). Again using flower shape, but also using the
position of the anthers, Wilson (1925) divided the genus into four sections similar to those of
Baker’s classification. The difference between the classification of the two authors is that Wilson
(1925) includes in L. sect. Archelirion only L. auratum, the type species of this section, while he
placed the other members of Bakers’s (1871) section into L. sect. Martagon. Comber (1949) re-
classified the genus Lilium into seven sections using 15 diagnostic characters, not only of the
flower and the growth habit, but also of the germination and bulb structure.
Willdenowia 36 – 2006 647

According to Comber (1949), Lilium sect. Liriotypus has a total number of 20 species and in-
cludes all European, Turkish and Caucasian species (Fig. 1) except for L. martagon, which has the
widest distribution range of all Lilium species and belongs to L. sect. Martagon Rchb. (Table 1).
Since no lilies are distributed between Asia Minor / Caucasus and E Afghanistan (Stern 1938), the
section thus contains all but one Lilium species occurring west of this gap. With the exception of
L. candidum with widely trumpet-shaped flowers and L. bulbiferum with erect bowl-shaped flow-
ers, all members of L. sect. Liriotypus have scattered leaves and Turk’s cap flowers (McRae
1998).
Baranova (1988) further subdivided the genus into eleven sections and classified the Euro-
pean lilies into four different sections (Table 1). She placed L. bulbiferum in L. sect. Pseudoli-
rium E. H. Wilson with other Asian and American erect-flowering lilies and L. candidum in the
unispecific section Lilium.L. martagon is placed in L. sect. Martagon together with other lilies
having a verticillate leaf arrangement. The remaining European lilies with Turk’s cap flowers
and scattered leaf arrangements were placed in L. sect. Eurolirium Baranova.
All members of the genus Lilium have the same basic chromosome number (x = 12), similar
chromosome morphology and a very large genome size (Siljak-Yakovlev & al. 2003). Smyth &
al. (1989) analysed 20 Lilium species from six sections and showed that C-banding patterns can
provide only little information for deducing the relationships between species.
Recently, several molecular markers have proved useful in developing an improved classifi-
cation. Yamagishi (1995) employed the RAPD technique to differentiate Lilium species and hy-
brids. Hayashi & Kawano (2000) used the coding regions of rbcLandmatK genes of chloroplast
DNA for their phylogenetic analysis of the family Liliaceae sensu stricto and also for the
infrageneric relationships within Lilium, although due to low substitution rates, with the rbcL
gene the resolution was very low. The matK marker proved to be more variable than the rbcL
gene. The study of these latter authors, however, included only four representatives of the L.
sect. Liriotypus and these four species were not resolved as a clade from other closely related
sections.
Dubouzet & Shinoda (1999) employed the sequences of the internal transcribed spacer re-
gion (ITS) of the nuclear ribosomal DNA to resolve the phylogenetic relationships among Japa-
648 Ikinci & al.: On the origin of European lilies
Table 1. Sectional classification according to different authors of the European, NE Turkish and Caucasian
Lilium species included in this study.
Taxon Baker (1871) Wilson (1925) Comber (1949) Baranova (1988)
L. akkusianum R. Gämperle Martagon Martagon Liriotypus Eurolirium
L. albanicum Griseb. Martagon Martagon Liriotypus Eurolirium
L. armenum Grossh. Martagon Martagon Liriotypus Eurolirium
L. artvinense Miscz. Martagon Martagon Liriotypus Eurolirium
L. bosniacum G. Beck Martagon Martagon Liriotypus Eurolirium
L. bulbiferum L. Isolirion Pseudolirium Liriotypus Pseudolirium
L. candidum L. Eulirion Leucolirion Liriotypus Lilium
L. carniolicum Bernh. Martagon Martagon Liriotypus Eurolirium
L. ciliatum P. H. Davis Martagon Martagon Liriotypus Eurolirium
L. jankae A. Kern. Martagon Martagon Liriotypus Eurolirium
L. kesselringianum Miscz. Martagon Martagon Liriotypus Eurolirium
L. martagon L. Martagon Martagon Martagon Martagon
L. monadelphum M. Bieb. Martagon Martagon Liriotypus Eurolirium
L. pomponium L. Martagon Martagon Liriotypus Eurolirium
L. ponticum K. Koch Martagon Martagon Liriotypus Eurolirium
L. pyrenaicum Gouan Martagon Martagon Liriotypus Eurolirium
L. rhodopaeum Delip. Martagon Martagon Liriotypus Eurolirium
L. szovitsianum Fisch. & Avé-Lall. Martagon Martagon Liriotypus Eurolirium

nese Lilium species and suggested further use of this region in a broader sampling of taxa. Their
results were congruent with Comber’s (1949) morphological classification. However, no repre-
sentatives of L. sect. Liriotypus were included because they do not occur in that area. At the same
time Nishikawa & al. (1999) showed that the nrDNA ITS region is a useful marker to unravel the
phylogenetic relationships and section delimitations of Lilium. Their study of 55 species of
Lilium included representatives of all sections according to the classification by Comber (1949)
but only six species of the European L. sect. Liriotypus. The latter section did not form a
monophyletic group, because L. bulbiferum was shown to be more closely related to L. sect.
Daurolirion than to the other members of L. sect. Liriotypus.
The present paper analyses the phylogenetic relationships of Lilium sect. Liriotypus based on
a phylogenetic analysis of sequence variation of the nuclear ribosomal ITS region with a compre-
hensive sampling representing 17 of the section’s 20 members. It aims at a better understanding
of the biogeography and evolution of L. sect. Liriotypus and of the origin of the European lilies.
Following the taxonomic view of Popova (1966) and Muratovi6(2005) rather than of Mathews
(1980), we recognise L. bosniacum, L. jankae and L. albanicum as distinct species and not as va-
rieties of a polymorphic L. carniolicum.
The present authors performed also a preliminary study with the sequences of the trnL
(UAA) intron and the intergenic spacer region between the trnL (UAA) 3’ exon and the trnF
(GAA) gene of the chloroplast DNA on nine species of Lilium sect. Liriotypus. Amplifications
were performed using the universal primers of trnl-c, trnl-d, trn-e and trnl-f of Taberlet (1991).
However, it was seen that there is no variable nucleotide position for a phylogenetic reconstruc-
tion. Therefore these chloroplast sequence data were not included in our analysis.
Material and methods
Plant material. – Sequences of the ITS region of 29 Lilium species (including Nomocharis)
representing all sections of the genus sensu Comber (1949), as well as Notholirion thomsonianum
Willdenowia 36 – 2006 649
Fig. 1. Distribution of the European, Turkish and Caucasian species of Lilium sect. Liriotypus included in this
study according to Davis & Henderson (1970), Stoker (1938, 1939), Synge (1980), Matthews (1980) and per-
sonal observations.

650 Ikinci & al.: On the origin of European lilies
Table 2. List of the taxa and the sources of the plant material or of the sequences, respectively, included in the
analysis.
Taxon Source and EMBL / GenBank accession number
Lilium L.
sect. Archelirion Baker
L. auratum Lindl. Nishikawa & al. (1999) — AB020472
L. japonicum Thunb. Nishikawa & al. (1999) — AB020451
sect. Daurolirion H. F. Comber
L. dauricum Ker Gawl. Nishikawa & al. (1999) — AB020473
sect. Leucolirion Wilson
L. longiflorum Thunb. Yang, C. M. & Chen, R. S. (unpubl.) — AY684927
L. philippinense Baker Nishikawa & al. (1999) — AB020437
sect. Liriotypus Asch. & Graebn.
L. akkusianum R. Gämperle Turkey, Ordu, Akkuî, 23.6.2002, Ikinci 1928 (AIBU) — AM292422
L. albanicum Griseb. Albanien-Montenegro, 29.6.1914, Dörfler 432 (WU) — AM292432
L. armenum Grossh. Turkey, Trabzon, Tonya, 25.6.2002, Ikinci 1933 (AIBU) — AM292425
L. artvinense Miscz. Turkey, Artvin, 2.7.2002, Ikinci 1960 (AIBU) — AM292427
L. bosniacum G. Beck Croatia, Velebit, Karlobag, B. Zollitsch (M 0056366) — AM292423
L. bulbiferum Nishikawa & al. (1999) — AB020468
L. candidum L. Turkey, Mu8la, Köyce8iz, 2.6.2002, Ikinci 1912 (AIBU) — AM292424
L. carniolicum Bernh. Austria, Königsberg, 1988 (M 0056392) — AM292419
L. ciliatum P. H. Da vis Turkey, Giresun, Tamdere, 25.6.2002, Ikinci 1932 (AIBU) — AM292421
L. jankae A. Kern. Romania, Transsilvania, J. Cstaó (B) — AM292431
L. kesselringianum Miscz. Turkey, Artvin, Ardanuç, 3.7.2002, Ikinci 1966 (AIBU) — AM292429
L. monadelphum M. Bieb. Russia, Stawropol, Kislowodsk, 29.5.1970, coll. ignot. (JE) — AM292418
L. pomponium L. Nishikawa & al. (1999) — AB035281
L. ponticum K. Koch Turkey, Trabzon, Çaykara, 28.6.2002, Ikinci 1944 (AIBU) — AM292426
L. pyrenaicum Gouan Nishikawa & al. (1999) — AB020428
L. rhodopaeum Delip. Greece, Rhodopi Mt, 23.7.1981, Strid & al. 19503 (B) — AM292430
L. szovitsianum Fisch. & Avé-Lall. Turkey, Ardahan, Çildir, 5.7.2002, Ikinci 1973 (AIBU) — AM292428
sect. Martagon Rchb.
L. martagon L. Nishikawa & al. (1999) — AB020455
sect. Pseudolirium Endl.
L. canadense L. Nishikawa & al. (1999) — AB020457
L. philadelphicum L. Nishikawa & al. (1999) — AB020432
sect. Sinomartagon H. F. Comber
L. nepalense D. Don Nishikawa & al. (1999) — AB020444
sect. Sinomartagon H. F. Comber
L. pumilum DC Nishikawa & al. (1999) — AB020430
L. davidii Duch. Nishikawa & al. (1999) — AB020461
Nomocharis Franch.
Nomocharis saluenensis Balf. Nishikawa & al. (1999) — AB020449
Outgroups
Cardiocrinum giganteum (Wall.)
Makino
Nishikawa & al. (1999) – AB020466
Fritillaria latifolia Willd. Turkey, Artvin, Borkça, 6.7.2002, Ikinci 1977 (AIBU) — AM292420
Notholirion thomsonianum (Royle)
Stapf
Rønsted & al. (2005) – AY616752

(Royle) Stapf, Fritillaria latifolia Willd. and Cardiocrinum giganteum (Wall.) Makino were used
(Table 2). Leaf material was collected in the field and dried in silica gel; in addition leaf probes
were prepared from herbarium specimens.
DNA isolation, amplification and sequencing. – Using the protocol provided by the manufac-
turer, total DNA was extracted from 20-25 mg of silica-gel dried leaves or leaves from herbarium
material using Qiagen DNeasy Plant Mini Kit (Qiagen, Hilden, Germany). The ITS region was
amplified in two pieces due to usage of herbarium material, since the degraded DNA was easier to
amplify in short pieces. The ITS1 region was amplified by using the primers ITS5 (White & al.
1990) and ITS1P2, which was designed by Ochsmann (2000) for Centaurea L. ITS2 was ampli-
fied using the primers ITS3 and ITS4 designed by White & al. (1990). Amplification conditions
were 94 °C (2 min) initial denaturation, 40 cycles of 94 °C (20 s) denaturation, 55 °C (45 s) an-
nealing, 72 °C (1 min) elongation and 72 °C (10 min) final elongation. PCR products were puri-
fied by Montage PCR Centrifugal Filter Device (Millipore Company). Cycle sequencing of
purified PCR products was performed by using the CEQ Dye Terminator Cycle Sequencing Start
Kit (Beckman Coulter) and sequences were analysed in a CEQ 8000 automated sequencer
(Beckman Coulter). All new nrDNA ITS sequences were submitted to the EMBL sequence data
bank (accession numbers given in Table 2).
Sequence analysis. – Limits of the ITS-1, the 5.8S rRNA gene and ITS-2 were determined by
comparison with published sequences. DNA sequences were aligned using ClustalX (Thompson
& al. 1997) and the alignment was corrected manually, following the guidelines by Kelchner
(2000).
Phylogenetic reconstruction. – Maximum parsimony analyses of the data set were performed us-
ing the heuristic search algorithm of PAUP* version 4.0b10 (Swofford 2002) with ACCTRAN;
MULPARS and tree-bisection-reconnection (TBR) branch swapping in action. 1000 addition se-
quence replicates were performed to locate potential islands of the most parsimonious trees.
Character states were specified unordered and unweighted and gaps in aligned sequences were
treated as missing data. The transitions/transversions ratio was set equal. A bootstrap analysis
was performed for the measurement of clade support with the following settings: 100 bootstrap
replicates with 10 random addition sequence replicates per bootstrap replicate.
Results
The data matrix obtained from nr ITS sequences comprised 32 taxa and contained 663 sites.
Among these, 384 (58 %) were constant, 128 (19 %) were variable but parsimony-uninforma-
tive and 151 (23 %) were parsimony-informative. The heuristic parsimony search generated 816
equally most parsimonious trees with a length of 558 steps, a consistency index (CI, excluding
uninformative characters) of 0.4717, homoplasy index (HI, excluding uninformative characters)
of 0.5283 and a retention index (RI) of 0.6687. The strict consensus tree of these 816 most parsi-
monious trees and bootstrap percentages (BP) for monophyletic groups is shown in Fig. 2.
In this analysis, the ingroup containing all Lilium species and one Nomocharis species is not
monophyletic. North American species of L. sect. Pseudolirium (L. philadelphicum and L. cana-
dense) were not resolved and separated from the rest of Lilium species. L. auratum and L.
japonicum as representatives of L. sect. Archelirion form a clade with a strong support (100 BP).
L. martagon is sister to the clade composed of representatives of L. sect. Daurolirion, sect.
Sinomartagon, sect. Leucolirion and of L. bulbiferum of L. sect. Liriotypus. The clade formed by
Nomocharis saluenensis and Lilium nepalense is sister to all members of L. sect. Liriotypus apart
from L. bulbiferum, but the support for this relationship is low (51 BP). NE Turkish-Caucasian
(99 BP) and European-SE Mediterranean lilies (85 BP) of L. sect. Liriotypus are sisters with a
strong support (95 BP). However, the clade formed by the two NE Turkish endemics L. ciliatum
and L. akkusianum (93 BP) is sister to all other European lilies with a bootstrap percentage of 85.
Willdenowia 36 – 2006 651

Relationships among all other NE Turkish-Caucasian Lilium species (L.armenum,L.kesselrin
-
gianum, L. monadelphum, L. artvinense, L. ponticum and L. szovitsianum) remain unresolved. L.
candidum is sister to a clade of L. rhodopaeum and L. jankae (83 BP) with a low bootstrap sup-
port (57 BP).
Discussion
Systematics.– Based on the resulting phylogenetic trees, Lilium sect. Liriotypus is only mo-
nophyletic if L. bulbiferum is excluded. This confirms the corresponding results of Nishikawa &
al. (1999) and Hayashi & Kawano (2000). Hence, the recent molecular studies, including our
study, do not support Comber’s (1949) classification of L. bulbiferum in L. sect. Liriotypus. Baker
(1871) placed L. bulbiferum in L. sect. Isolirion with other erect-flowering lilies from North
America and E Asia. Wilson (1925) and later Baranova (1988) followed Baker (1871) and placed
L. bulbiferum in L. sect. Pseudolirium, which has the same species composition as Baker’s (1871)
L. sect. Isolirion. According to our results, L. bulbiferum is closely related to L. pumilum of L.
sect. Sinomartagon, although the members of this section have Turk’s cap flowers, and is other-
wise placed in a clade comprising members also of L. sect. Daurolirion, sect. Leucolirion and
652 Ikinci & al.: On the origin of European lilies
Fig. 2. Strict consensus tree of the 816 equally most parsimonious trees based on nrDNA ITS sequence infor-
mation of the 32 taxa. – Percentages given above branches are bootstrap values (100 bootstrap replicates).
The length of trees is 558 steps, the consistency index (CI, excluding uninformative characters) = 0.4717, the
homoplasy index (HI, excluding uninformative characters) = 0.5283, the retention index (RI) = 0.6687. Sec-
tional names of the genus Lilium are given according to the classification of Comber (1949).

sect. Martagon. In the study by Nishikawa & al. (1999) L. bulbiferum is placed in L. sect. Dau-
rolirion, which is closely related to L. sect. Sinomartagon. The study by Hayashi & Kawano
(2000) revealed a similar result. The position of L. bulbiferum is discussed in more detail in the
following section in the light of morphological and other source of evidence.
Classification of the remaining species of Lilium sect. Liriotypus is less controversial apart
from the question whether they should be included in L. sect. Martagon, as proposed by Baker
(1871) and Wilson (1925), or in L. sect. Liriotypus. The type species of L. sect. Martagon, L. mar-
tagon differs from all other European lilies by having a verticillate leaf arrangement. In our phylo-
genetic trees this species is grouped with other E Asian species. Therefore, it is reasonable to
conclude that the European lilies should be placed in L. sect. Liriotypus, not in sect. Martagon.
Baranova’s (1988) section Eurolirium (the type of the section name is L. pyrenaicum) includes all
the species of section Liriotypus except L. candidum, which has trumpet shape flowers. Our results
clearly indicate that L. candidum should not be classified in a unispecific section, because this
would make the remaining group paraphyletic.
Morphological data and other evidence in conjunction with the phylogeny based on ITS. –In
the group of species forming Lilium sect. Liriotypus, only L. bulbiferum has upward-facing flow-
ers. Such a type of floral morphology is seen in other lilies from North America and Asia, where
they have a scattered distribution. Because of their wide distribution range and little specialized
flowers, Stoker (1939) considered these species as primitive. In contrast, Comber (1949) ex-
pressed the view that upward facing flowers appeared later as a result of parallel evolution in re-
sponse to drier environmental conditions with less need to protect the pollen. Lighty (1968)
proposed that L. bulbiferum shares common ancestors with the members of L. sect. Sinomartagon
and L. dauricum, which gave rise to L. bulbiferum through a different evolutionary pathway, rather
than with other European lilies. This last hypothesis gains support from our study, as well as the
hypothesis does that the upward-facing flowers of L. bulbiferum are plesiomorphic in Lilium.
The other morphologically divergent species is Lilium candidum with snow-white, open fun-
nel-shaped flowers, whereas the remainder of L. sect. Liriotypus has yellow or red Turk’s cap
flowers. The inner perianth segments of L. candidum are not papillose and are not distinctly hol-
low and with the stamens shorter than the style. Hibernating basal leaves arising from the bulb
scales are seen only in L. candidum, which also has radical leaves. Lighty (1968) considers L.
candidum as an evolutionary terminus of the route that prior gave rise to the other species of sec-
tion Liriotypus.
Although their floral morphology is quite different from each other, the long-ciliate hairs on the
leaf margins and on the flower buds of Lilium ciliatum and L. akkusianum are unique in the sec-
tion. L. ciliatum has small, complete Turk’s cap flowers, whereas L. akkusianum has large tepals,
which are only slightly recurved at the tips. Karyological studies of L. ciliatum (Özdemir 2003)
show 2n= 24 acrocentric chromosomes, two pairs of which were shorter than the others. There
are no chromosome data for L. akkusianum, which was recently described (Gämperle 1998).
The so-called Lilium carniolicum group includes L. carniolicum, L. albanicum, L. bosniacum, L.
jankae, L. pomponium and L. rhodopaeum. In this group L. albanicum, L. bosniacum and L. jankae
are often classified as subspecies or varieties of L. carniolicum, which is assumed to be a polymor-
phic species because of its variable flower colour and leaf indumentum (Grey-Wilson 1982, Stoker
1938). In this group L. jankae has yellow, strongly reflexed Turk’s cap flowers and evenly pubes-
cent veins on the ventral leaf face (Synge 1980). L. bosniacum has yellow or orange Turk’s cap
flowers with the leaf indumentum varying from glabrous to hairy. L. albanicum has yellow Turk’s
cap flowers and glabrous leaves and L. carniolicum has orange-red Turk’s cap flowers with hairs
on the on the ventral leaf face (Turrill 1953). A distinct species, L. rhodopaeum, is intermediate be-
tween L. carniolicum and the L. monadelphum-szovitsianum group (Synge 1980) and has
lemon-yellow flowers with barely reflexed tepals. L. pyrenaicum has small, yellow Turk’s cap
flowers, L. pomponium has bright red Turk’s cap flowers with papillose inner perianth segments.
Willdenowia 36 – 2006 653

According to our phylogenetic analysis, the Lilium carniolicum group is polyphyletic.
Matthews (1984) classified L. pyrenaicum, L. carniolicum (including L. albanicum, L. bosniacum
and L. jankae)andL. ponticum (including L. artvinense) as subspecies of L. pyrenaicum based on
the morphological similarities of the flowers and leaves. Nevertheless, our results do not support
this hypothesis, because at least L. ponticum (including L. artvinense) is distinctly separated from
the L. carniolicum group and also from L. pyrenaicum. In the light of their studies on genome size
and chromosome organisation of L. pyrenaicum, L. pomponium and L. carniolicum, Siljak-Ya-
kovlev & al. (2003) also suggest retaining them as separate species.
The Caucasian Lilium monadelphum group, which includes L. armenum, L. szovitsianum, L. mo-
nadelphum and L. kesselringianum, has similar large, yellow to pale cream flowers with tepals
only slightly recurved, leaving the tube visible. Although Stewart (1947) examined 48 species
and varieties of Lilium, only L. monadelphum of this group was included in his study. He found
that L. monadelphum has a distinct karyotype with secondary constrictions on the long arms of
the sixth chromosome pair.
The other Caucasian lilies, Lilium ponticum and L. artvinense, have smaller, deep orange to
yellow flowers with strongly recoiled tepals hiding the tube (Davis & Henderson 1970). We con-
sider L. ponticum and L. artvinense as conspecific, judging from their similar morphological
characters and overlapping distribution.
Biogeography. – The Himalayas are considered to be the centre of origin of the genus Lilium with
species having spread into the rest of Eurasia and North America (Patterson & Givnish 2002). Ac-
cording to Lighty (1968), L. martagon was the first species colonising Europe, a second migra-
tion from the north gave rise to the Central and SW European L. bulbiferum and the third route
from the south through Caucasia into Europe resulted in all other European Lilium species. Ac-
cording to our phylogenetic analysis, L. bulbiferum was not descended from the same ancestor as
the other European lilies, supporting the above hypothesis of a evolutionary origin of L.
bulbiferum separate from the other European lilies.
Lilium pomponium and L. pyrenaicum clustered together in the phylogenetic tree. L. pyrenaicum
occurs only in the eastern Pyrenees in both N Spain and in S France, growing in woods and
mountain meadows (Matthews 1980). L. pomponium occurs on rocky hillsides in the Maritime
Alps in S France and in NW Italy (Matthews 1984). They occupy the western limit of European
lilies and there is a distance of more than a thousand miles between the areas of these two species
and the rest of the L. carniolicum group.
Within the L. carniolicum group, L. bosniacum is distributed only in Bosnia and Herzegovina.
L. albanicum has the southernmost distribution of them, occurring in the mountains of Albania,
Y.R.P. Makedonia, Montenegro and N Greece. L. rhodopaeum occurs only in the Rhodope Mts in
N Greece and S Bulgaria where it grows in alpine meadows and on rocky slopes (Matthews 1980).
L. jankae is distributed in E Yugoslavia, Romania and Bulgaria (Matthews 1980).
Lilium candidum, which occurs naturally in SE Europe, Syria, Lebanon and Israel (Synge
1980), clusters together with other Balkan lilies. This is congruent with Lighty’s (1968) view that
aBalkanlinegaverisetoL. candidum.
According to Mandenova (1940), orographical and geographical isolation in the Caucasus played
a critical role in the evolution of new Lilium species. Except for L. martagon, all of the Lilium spe-
cies occurring in the NE Turkey-Caucasus region are confined to just this region. Only L.
monadelphum extends to the northern side of Caucasus (Synge 1980), all the other species are dis-
persed in Georgia, Armenia and NE Turkey. Our results show that two lineages are present in this
region, one giving rise to L. ciliatum and L. akkusianum, which are endemic to NE Turkey, and
the other lineage to all Caucasian lilies.
Conclusion. – Our phylogenetic analysis obtained from the sequence analysis of the nuclear ri-
bosomal ITS region shows that Lilium sect. Liriotypus is a monophyletic group if L. bulbiferum
654 Ikinci & al.: On the origin of European lilies

is excluded. Our results support the hypothesis that European species derived from three differ-
ent routes; first L. martagon colonized Europe, a second route gave rise to L. bulbiferum and the
last route gave rise to all other European Lilium species including L. candidum. Our analysis also
shows that L. ponticum (incl. L. artvinense)isnopartoftheL. carniolicum group.
Acknowledgements
Financial support by the Ali Nihat Gökyi8it Foundation for the field work is gratefully acknowl-
edged. The laboratory work was conducted at the Botanic Garden and Botanical Museum, Freie
Universität Berlin, when the first author received a one-year scholarship from DAAD (German
Academic Exchange Service). Many thanks are due to Margaret Johnson (Kew) for linguistic
and Orkun Ikinci (Ankara) for technical assistance, and to the reviewers Nikolai Friesen (Osna-
brück) and, in particular, Nina Rønsted (Kew) for valuable comments and suggestions.
References
Baker, J. G. 1871: A new synopsis of all the known lilies. – Gard. Chron. 28: 104.
Baranova, M. V. 1988: Konspekt sistemy roda Lilium (Liliaceae). [A synopsis of the system of
the genus Lilium (Liliaceae)]. – Bot. Zurn. (Moscow & Leningrad) 73: 1319-1329.
Comber, H. F. 1949: A new classification of the genus Lilium. – Lily Year-Book 13: 86-105.
Davis, P. H. & Henderson, D. M. 1970: A revision of Turkish lilies. – Lily Year-Book 33:
212-223.
Dubouzet, J. G. & Shinoda, K. 1999: Phylogenetic analysis of the Internal Transcribed Spacer
region of Japanese Lilium species. – Theor. Appl. Genet. 98: 954-960. [CrossRef]
Gämperle, R. 1998: A new species of lily from Turkey. – Quart. Bull. Alp. Gard. Soc. 66:
378-389.
Grey-Wilson, C. 1982: Lilium carniolicum, Liliaceae. – Curtis’s Bot. Mag. 184(1): 9-11.
Hayashi, K. & Kawano, S. 2000: Molecular systematics of Lilium and allied genera (Liliaceae):
phylogenetic relationships among Lilium and related genera based on the rbcLandmatK
gene sequence data. – Pl. Spec. Biol. 15: 73-93. [CrossRef]
Kelchner, S. A. 2000: The evolution of non-coding chloroplast DNA and its application in plant
systematics. – Ann. Missouri Bot. Gard. 87: 482-498. [CrossRef]
Lighty, R. W. 1968: Evolutionary trends in lilies. – Lily Year-Book 31: 40-44.
Mandenova, I. 1940: Lilii Kavkaza. Caucasian lilies. – Trudy Tbilissk. Bot. Inst. 8: 149-208.
Matthews, V. A. 1980: Lilium L. – In: Tutin T. G., Heywood V. H., Burges, N .A., Moore, D.
M., Valentine, D. H., Walters, S. M. & Webb, D. A. (ed.), Flora europea 5: 34-35. – Cam-
bridge, etc.
— 1984: Lilium pyrenaicum: a complex species. – Kew Mag. 1(1): 36-43.
McRae, E. A. 1998: Lilies: A guide for growers and collectors. – Portland, Or.
Muratovi6, E., Boguni6, F., Soljan, D., & Siljak-Yakovlev, S. 2005: Does Lilium bosniacum
merit species rank? A classical and molecular-cytogenetic analysis. – Pl. Syst. Evol. 252:
97-109. [CrossRef]
Nishikawa, T., Okazaki K., Uchino, T., Arakawa, K. & Nagamine, T. 1999: A molecular phy-
logeny of Lilium in the Internal Transcribed Spacer region of nuclear ribosomal DNA. – J.
Mol. Evol. 49: 238-249. [CrossRef]
Ochsmann, J. 2000: Morphologische und molekularsystematische Untersuchungen an der Cen-
taurea stoebe L.-Gruppe (Asteraceae-Cardueae) in Europa. – Diss. Bot. 324.
Özdemir, C. 2003: Morphological, anatomical and cytological characteristics of endemic Lilium
ciliatum P. H. Davis (Liliaceae) in Turkey. – Pakistan J. Bot. 35: 99-110.
Patterson, T. B. & Givnish, T. J. 2002: Phylogeny, concerted convergence, and phylogenetic
niche conservationism in the core Liliales: Insights from rbcLandndhF sequence data. –
Evolution 56: 233-252.
Willdenowia 36 – 2006 655

Popova, M. 1966: Some remarks on the lilies of the Balkans. – The Lily Year-Book 29: 66-70.
Rønsted, N., Law, S., Thornton, H., Fay, M. F. & Chase, M. W. 2005: Molecular phylogenetic
evidence for the monophyly of Fritillaria and Lilium (Liliaceae, Liliales) and the infra-
generic classification of Fritillaria. – Mol. Phylog. Evol. 35: 509-527. [CrossRef]
Siljak-Yakovlev, S., Peccenini, S., Muratovi6, E., Zoldos, V., Robin, O. & Vallés, J. 2003:
Chromosomal differentiation and genome size in the three European mountain Lilium spe-
cies. – Pl. Syst. Evol. 236: 151-163. [CrossRef]
Smyth, D. R., Kongsuwan, K. & Wisudharomn, S. 1989: A survey of C-band patterns in chro-
mosomes of Lilium (Liliaceae). – Pl. Syst. Evol. 163: 53-69. [CrossRef]
Stern, F. C. 1938: A geographical survey of the genus Lilium. – Proc. Linn. Soc. London 150:
29-34.
Stewart, R. N. 1947: The morphology of somatic chromosomes in Lilium.–Amer.J.Bot.34:
9-26. [CrossRef]
Stoker, F. 1938: The carniolicum group of lilies. – Lily Year-Book 7: 86-107.
— 1939: European lilies of the Isolirion section. – Lily Year-Book 8: 15-29.
Swofford, D. L. 2002: PAUP*: Phylogentic Analysis Using Parsimony (* and other methods)
version 4.1b10. – Sunderland, MA.
Synge, P. M. 1980: Lilies. A revision of Elwes’ Monograph of the Genus Lilium and its Supple-
ments. – London.
Thompson, J. D., Gibson, T. J., Plewniak, F., Jeanmougin, F. & Higgins, D. G. 1997: The Clus-
talX windows interface: flexible strategies for multiple sequence alignment aided by quality
analysis tools. – Nucl. Acids Res. 24: 4876-4882.
Turrill, W. B. 1953: The lilies of the Balkan Peninsula. – Lily Year-Book 17: 30-39.
White, T. J., Bruns, T., Lee, S. & Taylor, J. 1990: Amplification and direct sequencing of fungal
ribosomal RNA genes for phylogenetics. – Pp. 315-322 in: Innis, M. A., Gelfand, D. H.,
Sninsky, J. J. & White, T. J. (ed.), PCR protocols: a guide to methods and applications. – San
Diego.
Wilson, E. H. 1925: The lilies of Eastern Asia, a monograph – London.
Yamagishi, M. 1995: Detection of section-specific random amplified polymorphic DNA (RAPD)
markers in Lilium. – Theor. Appl. Genet. 91: 830-835. [CrossRef]
Addresses of the authors:
Nursel Ikinci, Abant Izzet Baysal University, Faculty of Arts & Sciences, Department of Biology,
TR-14280 Bolu, Turkey; e-mail: nurselikinci@yahoo.com.
Christoph Oberprieler, Universität Regensburg, Naturwissenschaftliche Fakultät III-Biologie
und Vorklinische Medizin, Universitätstr. 31, D-93053 Regensburg, Germany; e-mail:
christoph.oberprieler@biologie.uni-regensburg.de
Adil Güner, Nezahat Gökyi8it Botanik Bahçesi, Ataîehir, Atatürk Mah. Fatih Sultan Mehmet
Cad. TEM Otoyolu Anadolu Kavîa8í, Kadíköy, TR-81120 Istanbul, Turkey; e-mail: adilguner@
ttnet.net.tr.
656 Ikinci & al.: On the origin of European lilies