Content uploaded by Svetlana Shcherbik
Author content
All content in this area was uploaded by Svetlana Shcherbik on Jul 14, 2014
Content may be subject to copyright.
229
Int. J. Plant Sci. 164(2):229–237. 2003.
䉷 2003 by The University of Chicago. All rights reserved.
1058-5893/2003/16402-0003$15.00
PHYLOGENETIC RELATIONSHIPS OF THE SIBERIAN IRIS SPECIES INFERRED FROM
NONCODING CHLOROPLAST DNA SEQUENCES
Irina Makarevitch,
1
Kseniya Golovnina, Svetlana Scherbik,
2
and Alexander Blinov
Institute of Cytology and Genetics, Lavrentjeva 10, 630090 Novosibirsk, Russia
A study of phylogenetic relationships of Iris species has been complicated because of extreme morpho-
ecological diversity, wide distribution of the genus, multiple hybridizations, and convergent evolution processes
in the genus. In order to get an insight into the evolutionary history of Iris and to clarify some contradictions
of contemporary classifications, we performed a molecular analysis of the phylogenetic relationships of a
heterogeneous group of Iris species occurring in Siberia and covering major taxonomic groups of the genus
Iris. According to contemporary classifications, these species belong to the subgenera Limniris (Tausch) Spach,
Xyridion (Tausch) Spach, Iris, and Pardanthopsis (Hance) Baker. However, the position of Pardanthopsis
within the genus Iris and the position of Xyridion as a distinct subgenus or a part of Limniris are disputable.
Based on an analysis of 56 RAPD markers in 12 Siberian Iris species and comparative analysis of trnL intron
and trnL-trnF intergenic spacer noncoding chloroplast DNA sequences in all 22 Siberian Iris species, we
reconstructed the phylogenetic relationships of Siberian Iris species. In general, the species grouping coincides
with contemporary classifications. According to our results, Pardanthopsis dichotoma forms a separate branch
on the phylogenetic trees based on sequences of chloroplast DNA and RAPD analysis, which supports its
position as a distinct genus. All of the Siberian Iris species are clustered into four phylogenetic groups. Our
results indicate that the phylogeny and taxonomic structure of the genus Iris may need reconsideration.
Keywords: Iris, trnL intron, trnL-trnF intergenic spacer, RAPD analysis, molecular phylogeny, evolution.
Introduction
Iris L. is the largest and most complicated genus of the
Iridaceae. The name of the genus is the Greek word for “rain-
bow.” Rich coloring and extreme diversity are the distinctive
characteristics of more than 300 Iris species. Irises arose in the
eastern and southeastern regions of Asia during the lower Ter-
tiary (Rodionenko 1987). Now the range of the genus extends
to all of the continents of the Northern Hemisphere. Under
diverse ecological conditions in this broad territory, irises grow
as mesophytes, merophyte-cryophytes, psammophytes, and
hydrophytes, and as calcifuges and calcifiles (Rodionenko
1987). This large variety of natural habitats and adaptive traits
makes the phylogenetic relationships and evolutionary history
of the genus Iris worth investigating.
The systematics of the genus Iris is intricate. The methods
of species grouping vary depending on monographer and
choice of characters compared, and, as Mathew (1989, p. 5)
wrote, “there [are] nearly as many different classifications as
there [are] botanists who [have] studied the subject.” Most
classifications of Iris are based on morpho-anatomical features,
1
Author for correspondence; current address: Agronomy and
Plant Genetics, University of Minnesota, 411 Borlaug Hall, 1991 Up-
per Buford Circle, St. Paul, Minnesota 55108, U.S.A.; e-mail
makar003@umn.edu.
2
Current address: Department of Biology, Georgia State University,
24 Peachtree Center Avenue, Atlanta, Georgia 30303, U.S.A.
Manuscript received February 2002; revised manuscript received September
2002.
ecological features, and cytogenetic analysis (Wu and Cutler
1985; Rodionenko 1987; Doronkin 1987, 1990; Mathew
1989). The extreme diversity of morpho-ecological character-
istics, the evidence of convergent evolution in many traits used
for systematic reconstructions, and a particularly complex pat-
tern of chromosomal variation (Ellis 1997; Goldblatt 1997)
make the study of the evolutionary and systematic relation-
ships of Iris species complicated. Consequently, no consensus
classification of Iris species has been constructed to date.
In Siberia, Russia, the genus Iris is represented by 22 species
covering almost all Iris subgenera. Later in this article we will
use “Siberian irises” to mean Iris species occurring in Siberia.
Some of the Siberian irises grow in narrow natural habitats;
others have an extensive geographical distribution. Eleven of
the Siberian Iris species are endemic to Asia. Iris ludwigii is
one of the rarest species in the world, occurring only in eight
natural habitats, including the Altai foothills in Siberia (Do-
ronkin 1987, 1990). In contrast, I. ruthenica is a very common
plant in Eastern Europe and Asia, growing in different eco-
logical conditions and habitats from low altitudes up to 3200
m. Iris setosa grows in eastern Asia and on the North American
continent (Mathew 1989). Many of the Siberian Iris species
are included on national and regional endangered species lists.
All species are of great ornamental importance and have the
potential to serve as a source of germplasm for breeding pro-
grams (Doronkin 1987; Lyte 1997; Maynard 1997).
Two contemporary classifications of the genus Iris differ in
the position of Iris species included in this study. According
to Rodionenko (1987), these species belong to four subgenera:
Limniris (Tausch) Spach, Xyridion (Tausch) Spach, Iris, and
230 INTERNATIONAL JOURNAL OF PLANT SCIENCES
Pardanthopsis (Hance) Baker. Of these, Limniris is the most
ancient group, and Pardanthopsis is nested within other Iris
species. Originally, Lenz (1972) detached I. dichotoma from
the genus Iris to the distinct genus Pardanthopsis, and Mathew
(1989), in the last formal revision of Iris, supported this point.
He did not recognize the subgenus Xyridion and included the
species of Xyridion sensu Rodionenko into the subgenus Lim-
niris (Mathew 1989). Doronkin (1987) recognized the sub-
genus Eremiris as being comprised of I. biglumis and its allies,
which are included in the subgenus Limniris in other systems.
According to Rodionenko (1987), the subgenus Limniris is the
most ancient group of irises. However, almost no information
regarding the phylogenetic relationships of Iris species has been
found to date. Little has changed since Goldblatt (1990) noted
that the circumscription, taxonomic rank, and phylogenetic
position of different groups of the genus Iris are the subjects
of long-standing debates. The question of Iris systematics re-
mains open, and phylogenetic relationships of Iris species re-
quire further careful study.
The most recent phylogeny of Iridaceae was constructed by
Reeves et al. (2001) using molecular data derived from four
coding and noncoding plastid DNA regions. In this study and
in the previous report of intrafamiliar phylogenetic relation-
ships by Souza-Chiez et al. (1997), Pardanthopsis was ana-
lyzed as a distinct genus and was placed as the sister clade to
Belamcanda. To date, no study of the phylogeny and evolution
of the Iris genus at the intrageneric level based on molecular
sequence analysis has been reported.
Recently, the RAPD technique was successfully used in stud-
ies of the systematic relationships between five of the Far East-
ern Iris species (Zhuravlev et al. 1998) and of the genetic
variability of I. setosa populations (Artyukova et al. 2001),
and a set of arbitrary primers suitable for inter- and intraspe-
cific polymorphism analysis in irises was found. RAPD mark-
ers (Welsh and McCleland 1990; Williams et al. 1990), un-
influenced by the environment and the developmental stage of
an organism, allow an easy and fast identification even of
closely related species, a wide characterization of the germ-
plasm used in breeding programs (Ranade et al. 2001), and a
clarification of the systematic relationships between species
(Commicini et al. 1996; Swoboda and Bhalla 1997; Klaas
1998; Svitashev et al. 1998; Sun et al. 1999; Bartish et al.
2000). However, use of RAPD markers in phylogenetic studies
has some limitations. First, with the techniques usually applied,
the validity of interpretation of RAPD data requires the DNA
fragments of the same size to be strictly homologous. Second,
the technique gives stable, reliable results only for comparisons
of closely related species when true nucleotide sequence di-
vergence does not exceed ca. 10% (Clark and Lanigan 1993;
Espinasa and Borowsky 1998). In order to carefully evaluate
the genomic diversity and to get a better resolution of the
phylogenetic relationships of the taxa studied, RAPD analysis
can be combined with the comparative study of nucleotide
sequences (Blattner et al. 2001; Gehrig et al. 2001).
Different genes from nuclear and chloroplast genomes have
proven to be useful for deducing the phylogenetic relationships
of plant species. Sequences of internal transcribed spacers (ITS)
are the most often used nuclear DNA sequences for revealing
relationships at the inter- and intrageneric levels in plants (Cer-
bah et al. 1998; Wen 2000; Gehrig et al. 2001). The chloroplast
genome provides a large amount of basic information sup-
porting comparative evolutionary research at different taxo-
nomic levels. The coding sequences of the plastid genome
(rbcL, rps4, ndhF, matK, and atpB) have been shown to have
a lower rate of nucleotide substitutions than noncoding se-
quences of the plastid and nuclear genome and were more
successfully used in phylogenetic studies at the higher taxo-
nomic level (Freshwater et al. 1994; Fritsch 2001; Lewis and
Doyle 2001; Lia et al. 2001) as opposed to noncoding se-
quences (atpB-rbcL intergenic spacer, trnL intron, and trnL-
trnF intergenic spacer; Taberlet 1991; Soltis and Soltis 2000;
Nozaki et al. 2000; Asmussen and Chase 2001). Nevertheless,
in some plant taxa, coding sequences, such as rbcL (ribulose-
1,5-bisphosphate carboxylase), can provide sufficient infor-
mation for reconstructions of phylogenetic relationships at the
intrageneric level (Chase et al. 2000). In our study, we chose
to use a noncoding trnL-trnF region characterized by higher
nucleotide substitution rate and proven to be useful for elu-
cidating phylogenetic relationships at the intrageneric level
(Cros et al. 1998; Richardson et al. 2000; Fernandez et al.
2001; Fukuda et al. 2001; Soliva et al. 2001).
Our study is the first attempt to resolve the phylogenetic
relationships of Iris species at the intrageneric level using mo-
lecular data. The aim of this analysis was to clarify the sys-
tematic and phylogenetic relationships of Iris species, to resolve
some contradictions of contemporary classifications, and to
enhance our understanding of the phylogenetic and evolution-
ary history of the genus Iris. Based on an analysis of 56 RAPD
markers in 12 Siberian Iris species and the comparative anal-
ysis of trnL intron and trnL-trnF intergenic spacer noncoding
chloroplast DNA sequences in all of the 22 Siberian Iris spe-
cies, we reconstructed the phylogenetic relationships of Sibe-
rian Iris species. According to our results, Pardanthopsis di-
chotoma forms a separate branch on the phylogenetic trees
based on sequences of chloroplast DNA and RAPD analysis.
All of the Siberian Iris species are clustered into four groups.
Bearded irises (subgenus Iris) appear to be the most ancient
group of Iris species.
Material and Methods
Plants
All of the 22 Iris species occurring in Siberia and Pardan-
thopsis dichotoma, which is closely related to irises, were an-
alyzed in this study (table 1). Belamcanda chinensis was in-
cluded in the analysis as an outgroup species. Samples were
collected from natural populations in different regions of Si-
beria or were obtained from the living collection of the Central
Siberian Botanical Garden (SB RAN, Novosibirsk). Samples
of I. loczyi, I. pallasii, and I. tenuifolia were received from the
herbarium of SB RAN.
Genomic DNA Isolation
DNA was isolated from 50–170 mg of frozen leaves using
a standard CTAB method (kit Plant NucleoSpin, Macherey
Nagel). The quality and quantity of DNA were checked by
electrophoresis in a 1% agarose gel containing ethidium bro-
mide (0.5 mg/mL) in 1#TAE.
MAKAREVITCH ET AL.—PHYLOGENY OF SIBERIAN IRIS SPECIES 231
Table 1
Origin of the Investigated Accessions
Species
Cultivated
or wild Origin
GenBank accession
trnL intron trnL-trnF
Iris biglumis Vahl. Wild West Siberia AF480388 AF480366
I. bloudowii Bunge Cultivated
a
Altai AF480389 AF480367
I. ensata Thunb. Wild West Siberia AF480390 AF480368
I. glaucescens Bunge Cultivated
a
Altai AF480391 AF480369
I. halophila Pallas Cultivated
a
Altai AF480392 AF480370
I. humilis Georgi Cultivated
a
West Siberia AF480393 AF480371
I. ivanovae Doronkin Wild East Siberia AF480394 AF480372
I. lactea Pallas Wild West Siberia AF480395 AF480373
I. laevigata Fisher Cultivated
a
Yakutia AF480396 AF480374
I. loczyi Kanitz Wild
b
West Siberia AF480397 AF480375
I. ludwigii Maxim. Wild Altai AF480398 AF480376
I. pallasii Fisch. Wild
b
East Siberia AF480399 AF480377
I. potanini Maxim. Wild West Siberia AF480400 AF480378
I. pseudacorus L. Cultivated
a
Ural AF480401 AF480379
I. ruthenica Ker-Gawler Wild West Siberia AF480402 AF480380
I. sanguinea Hornem ex Donn Wild West Siberia AF480403 AF480381
I. setosa Pallas ex Link Cultivated
a
Yakutia AF480404 AF480382
I. sibirica L. Cultivated
a
Altai AF480405 AF480383
I. tenuifolia Pall. Wild
b
West Siberia AF480406 AF480384
I. tigridia Bunge ex Ledeb. Cultivated
a
Altai AF480407 AF480385
I. uniflora Pallas Wild East Siberia AF480408 AF480386
I. ventricosa Pall. Wild West Siberia AF480409 AF480387
Pardanthopsis dichotoma Lenz Cultivated
a
East Siberia
Belamcanda chinensis Adans. Cultivated
c
Southeast Asia
a
Cultivated at the Exhibition of the Rare Species, Central Siberian Botanical Garden (SB RAN).
b
Herbarium collection, SB RAN.
c
Cultivated in the Greenhouse of Tropical Plants, SB RAN (voucher specimen 5799).
RAPD Analysis
Four RAPD primers (decamer oligonucleotides; Operon
OPD08, OPD11, OPD13, OPB12) were used for the ampli-
fication of the total DNA. The primers for the reactions were
chosen because of their successful application in the previous
study of species within the genus Iris, in which a reproducible
amplification and an appropriate level of polymorphism were
shown. OPD08, OPD11, OPD13, OPB12 gave stable, well-
reproducible, species-specific patterns for Iris species with no
variation between individuals from the same populations
(Zhuravlev et al. 1998). PCR was performed in a volume of
10 mL containing 65 mM Tris-HCl (pH 8.9), 16 mM
(NH
4
)
2
SO
4
, 1.5 mM MgCl
2
, 0.2 M dATP, 0.2 M dGTP, 0.2
M dTTP, 0.2 M dCTP, 10 mM primer, 10–50 ng of genomic
DNA template, and 1 U Taq DNA polymerase. The PCR pro-
gram had an initial strand separation step at 92⬚Cfor3min
followed by 30 cycles of denaturation at 92⬚C for 30 s, an-
nealing at 36⬚C for 45 s, and elongation at 72⬚C for 1 min.
The amplification products were separated by electrophoresis
in 1.5% agarose gel containing ethidium bromide (0.5 mg/mL)
in 0.5#TBE and visualized under UV light. Molecular weights
of amplified products were estimated using HindIII/EcoRI-
digested lDNA as a standard. The DNA profiles were scored
manually, directly from photographs of the gels. Only bands
clearly marked on the photographs and reproducible bands
were scored and used in the analysis. For each genotype, the
presence or absence of fragments, regardless of their intensity,
was recorded as 1 or 0, respectively, and these were treated
as discrete characters. Pairwise comparison of banding pat-
terns was evaluated by calculating an index of genetic simi-
larity (S
AB
) using the matching coefficient method of Nei and
Li (1979): number of shared bands/(the numberS p 2 # the
AB
of bands in number of bands in B) with RAPD programA + the
(Simtel.Net, Eden Prairie, Minn.; ftp://ftp.simtel.net/pub/
simtelnet/msdos/biology). A RAPD phenogram was con-
structed with Phylip software package (Felsenstein 1993) using
the unweighted pair-group method with arithmetical averages
(UPGMA) and neighbor-joining methods based on genetic dis-
tances, calculated as .d p 1 ⫺ S
AB
Sequence Analysis
The part of the rbcL gene, the ITS region, the trnL intron,
and the trnL-trnF intergenic spacer were amplified using pairs
of primers S-523 (5
-AAACCAAAATTGGGATTATCCGCA-
AAAAATTA-3
) and Z-1204R (5
-CCCTAAGGGTGTCCTA-
AAGTTTCTCCACC-3
), ITS1 (5
-TCCGTAGGTGAACCTG-
CGG-3
) and ITS4 (5
-TCCTCCGCTTATTGATATGC-3
), c
and d, and e and f (Taberlet et al. 1991), respectively. All PCR
reactions were performed in a volume of 50 mL containing 65
mM Tris-HCl (pH 8.9), 16 mM (NH
4
)
2
SO
4
, 1.5 mM MgCl
2
,
0.2 M dATP, 0.2 M dGTP, 0.2 M dTTP, 0.2 M dCTP, 10 mM
primers, 20–50 ng genomic DNA template, and 1 U Taq DNA
polymerase. The PCR program had an initial strand separation
step at 92⬚C for 3 min followed by 30 cycles of denaturation
232 INTERNATIONAL JOURNAL OF PLANT SCIENCES
Fig. 1 RAPD patterns of Siberian Iris species with OPD08 primer. Lane 1, Iris laevigata; 2, I. glaucescens; 3, I. tigridia; 4, I. ivanovae; 5,
I. ensata; 6, I. humilis; 7, I. bloudowii; 8, I. ludwigii; 9, I. halophila; 10, I. pseudacorus; 11, I. uniflora; 12, I. ruthenica; 13, P. dichotoma; 14,
B. chinensis. weight marker (HindIII/EcoRI-digested lDNA). The scale represents the size of marker bands. The fragment specificM p molecular
for subgenus Iris species is marked by an arrow.
at 92⬚C for 30 s, annealing at 46⬚C for 45 s, and elongation
at 72⬚C for 1 min. The PCR products were analyzed in agarose
electrophoresis, extracted from gel with a Qiaquick Gel Ex-
traction Kit (Qiagene; according to manufacturer protocol),
and directly sequenced with the primers used for the PCR
reactions by the dideoxy chain termination method of Sanger,
for which the dsDNA Cycle Sequencing Version (Gibco BRL)
was used. PCR products from samples of three species were
polished by Pfu polymerase (Stratagene; according to manu-
facturer protocol), cloned into BlueScript, and sequenced using
universal primers.
The sequences were aligned manually, and only one good
way to align sequences was found. Constructed alignments of
the sequences of the trnL intron and trnL-trnF intergenic
spacer region were analyzed with maximum parsimony, max-
imum likelihood, and neighbor-joining methods using a Phylip
software package (Felsenstein 1993). The Tajima-Nei model
from this package was used to calculate the distances between
sequences for neighbor-joining analysis because it takes into
account differences in nucleotide frequencies. The trnL intron
and the trnL-trnF intergenic spacer region were analyzed sep-
arately as two distinct and potentially phylogenetically infor-
mative regions and as one combined data set. Internal support
was assessed with 1000 bootstrap replicates (Felsenstein
1985). Only groups with frequency of more than 50% were
reported. To explore the phylogenetic utility of indels, we
treated gap sites in two different ways. First, the analysis of
the data with the gap positions treated as missing values was
performed. Second, indels were included in the analysis as
binary characters. As the next step of analysis, the Iris species
were clustered based on the distribution of the indels.
Results and Discussion
Morphological and ecological diversity of Iris species, mul-
tiple convergent processes, and interspecific hybridizations,
which may be common within Iris (Rodionenko 1987;
Mathew 1989), make a phylogenetic analysis based on mor-
phological characters complicated. Molecular data can sup-
plement morphological studies and provide more reliable and
powerful techniques for phylogenetic reconstructions. In our
study we estimated the phylogenetic and systematic relation-
ships of Siberian Iris species based on a molecular RAPD anal-
ysis and a comparative analysis of the chloroplast noncoding
DNA sequences.
RAPD Analysis
We analyzed 12 species representing different systematic
groups of Iris species using the RAPD technique. In order to
clarify the position and rank of Pardanthopsis dichotoma,we
included this species in our study. Belamcanda chinensis was
used as an outgroup species. RAPD assays were performed at
least twice for all of the primer-genotype pairs. RAPD patterns
remained stable when assays were treated at different anneal-
ing temperatures (36⬚ –39⬚C) and with different quantities of
DNA samples (10–50 ng per reaction). Based on the repro-
ducibility of clear banding patterns and band intensity, 36⬚C
was selected as the optimum annealing temperature. We found
no variation of individual RAPD patterns with any of the four
primers.
The optimized conditions were used to study genetic vari-
ability in 12 Iris species covering the major systematic groups
of Siberian Iris species. The total number of polymorphic frag-
ments yielded by the four decamer primers was large, but only
the bands that were clearly marked and distinguishable on
photographs were considered for further study; faint bands
were ignored. On this basis, a total of 56 scorable polymorphic
fragments from all experiments were unambiguously identified
and used for a statistical analysis. The sizes of the scored frag-
ments ranged from 0.2 to 4 kb (fig. 1; data not shown).
RAPD amplification of Iris DNA revealed a high degree of
MAKAREVITCH ET AL.—PHYLOGENY OF SIBERIAN IRIS SPECIES 233
Fig. 2 Systematic relationships of Siberian Iris species based on
RAPD analysis. Clades and clusters formed by Iris species are shown
in letters and numbers, respectively. Bootstrap percentages with a fre-
quency of more than 50% are shown in the nodes of the dendrogram.
genetic variability among species investigated as indicated by
the generation of unique sets of RAPD products. OPD13 gave
the monomorphic band (species-specific marker) for I. ludwigii
(data not shown). All of the arbitrary primers used in this
study showed different amplification patterns in all of the 12
species analyzed, which indicates interspecific and intergeneric
variation. We found several specific markers for different taxa,
namely, for the subgenus Iris (marked by an arrow in fig. 1),
the section Ioniris, and the series Laevigata (data not shown).
OPB 12 gave a fragment specific for all of the Iris species except
for the species of the subgenus Xyridion (data not shown).
RAPD patterns for the pairs of closely related species—I. hu-
milis and I. bloudowii (fig. 1, lanes 6 and 7), I. uniflora and
I. ruthenica (fig. 1, lanes 11 and 12), and I. tigridia and I.
ivanovae (fig. 1, lanes 3 and 4)—had different amplification
patterns when different primers were used that could be used
to identify these species. Recently, the reports of successful use
of RAPD analysis in flower breeding programs (Scovel et al.
1998), Iris cultivar development, hybrid detection (Shimizu et
al. 1999), and studies of the genetic variability of Iris popu-
lations (Artiukova et al. 2001) have begun to appear. The
RAPD analysis of the Siberian Iris species used in our study
revealed an extensive amount of variation, which provided
further evidence that this technique offers a reliable and simple
method for identification of closely related species and marker-
assisted selection in Iris breeding programs.
DNA fingerprints produced by each primer were used to
calculate similarity coefficients in pairwise comparisons of the
accessions, as described by Nei and Li (1979). This similarity
coefficient is based on the presence of shared amplified frag-
ments, assuming that fragments with the same electrophoresis
mobility are allelic and fragments with different electropho-
resis mobility are nonallelic. The similarity coefficients ranged
from 0.84 for two closely related species, I. ruthenica and I.
uniflora,to0.24forI. bloudowii and I. pseudacorus. The
average interspecific similarity coefficient was 0.68.
Using the UPGMA and neighbor-joining method based on
genetic distances, phenograms representing the relationships
within the genus Iris were constructed with B. chinensis as an
outgroup (fig. 2). The phenograms constructed using both
methods were the same. Pardanthopsis dichotoma forms a
separate branch, which indicates its remoteness from the Iris
species and supports its position as a distinct genus. All of the
Iris species analyzed form three clades: A, I. bloudowii, I.
humilis, I. ivanovae, I. tigridia, and I. glaucescens;B,I. ha-
lophila and I. ludwigii;C,I. ruthenica, I. uniflora, I. laevigata,
I. pseudacorus, and I. ensata.
In general, characterization of Iris species based on RAPD
analysis agrees with the classifications by Rodionenko (1987)
and Mathew (1989). However, the power of the technique is
not sufficient to resolve the relationships between species
groups, which is primarily due to a decreasing number of truly
homologous RAPD bands and an increasing number of non-
homologous bands of the same size when distant relatives are
analyzed. The use of RAPD markers in phylogenetic studies
is limited by an insufficient resolution of commonly applied
techniques of electrophoresis (Clark and Laningan 1993). Fur-
thermore, the impossibility of the use of RAPD-PCR as a uni-
versal molecular clock makes phylogenetic analysis based on
only RAPD markers contestable (Espinasa and Borowsky
1998). A comparison of DNA nucleotide sequences is often
used to supplement and extend primary RAPD-PCR analysis.
Nucleotide sequence data provide considerably greater statis-
tical reliability and allow for a better resolution of an evolu-
tionary tree topology (Blattner et al. 2001; Gehrig et al. 2001).
Sequence Analysis
As the next step in our study of the phylogenetic relation-
ships of Iris species, we analyzed chloroplast DNA sequences
of Siberian Iris species. RbcL gene sequences were shown to
be useful in phylogenetic reconstructions at the intrageneric
taxonomic level (Chase et al. 2000). We partially sequenced
the rbcL gene from some Siberian Iris species belonging to
different subgenera (I. halophila, I. ludwigi, I. uniflora, I. pseu-
dacorus, I. glaucescens, I. tigridia, and I. laevigata) and ana-
lyzed these sequences together with rbcL sequences from other
Iris species determined earlier by other researchers (I. ungui-
cularis, I. ensata, and I. germanica; accession numbers
AJ309693, D28332, and L05037, respectively; Reeves et al.
2001). However, partial sequences of the rbcL gene from Iris
species revealed only a few nucleotide substitutions (
!5% poly-
morphic positions), which is insufficient for phylogenetic anal-
ysis. ITS nuclear sequences are widely used in molecular phy-
logenetics at the species and genus levels. However, sequences
of the ITS region in Iris species displayed multiple copies of
different size within one species, which suggests an occurrence
of multiple types of rDNA repeats. Thus, an additional cloning
step is required in order to involve ITSs sequences as phylo-
genetic markers for reconstructions of relationships between
Iris species.
In this study we determined sequences of the trnL intron
234 INTERNATIONAL JOURNAL OF PLANT SCIENCES
Fig. 3 Clusters of Siberian Iris species based on indels distribution. Positions of indels correspond to alignments. The trnL-trnF intergenic
spacer indels are in bold. Signs designate each indel and correspond to those in fig. 4.
and the trnL-trnF intergenic spacer of chloroplast DNA for
all of the 22 Siberian Iris species. In order to clarify the position
of P. dichotoma, we included sequences of the trnL-trnF region
from this species (Reeves et al. 2001; accession number
AJ409612). The majority of PCR fragments were sequenced
directly; PCR products for both regions in I. laevigata, I. po-
tanini, and I. sanguinea were cloned and then sequenced with
universal primers. All the sequences were unique. The se-
quences were aligned manually with all the gaps introduced
in evident positions. No other good alignment could be con-
structed. The alignment of the trnL intron sequences in Iris
species requires the introduction of 13 gaps, from 1 to 101 bp
in length. The alignment of trnL-trnF intergenic spacer se-
quences inferred eight gaps from 1 to 15 bp. The nature of
the gaps (insertion or deletion) was hypothesized based on an
outgroup comparison. Five out of 13 gaps in the trnL intron
sequences were small insertions
!10 bp in length; two others
were large insertions of 25 and 101 bp in length. Eight other
gaps were deletions of 1 to 5 bp, with one larger deletion of
35 bp in length. Six out of eight gaps in trnL-trnF intergenic
spacer sequences were small insertions of up to 5 bp in length,
with one larger insertion of 15 bp; two other gaps were small
deletions ranging from 1 to 5 bp.
The length of trnL intron sequences varied from 371 (I.
loczyi and I. tenuifolia) to 508 (I. sibirica and I. sanguinea)
in individual accessions, the multiple alignment being 557 bp
in length. The length of trnL-trnF intergenic spacer sequences
ranged from 392 (I. ivanovae, I. tigridia, and I. glaucescens)
to 409 (I. biglumis, I. laevigata, I. pallasii, I. pseudacorus, and
I. lactea) in individual species, the multiple alignment being
424 bp in length. Among all Iris species studied, the multiple
alignment of trnL intron sequences resulted in 92 variable
characters (79 nucleotide substitutions and 13 indels) and 53
parsimony-informative positions; 280 aligned nucleotide sites
were invariant. The multiple alignment of trnL-trnF intergenic
spacer sequences resulted in 191 variable characters (183 nu-
cleotide substitutions and eight indels) and 167 parsimony-
informative positions; 190 aligned nucleotide sites were in-
variant. The transition/transversion ratio was 1.57 for trnL
intron sequences and 1.04 for trnL-trnF intergenic spacer se-
quences. Chloroplast DNA sequences of the Iris species studied
were AT rich (nucleotide composition is A p 39.1%, C p
16.1%, G p 19.5%, and T p 25.3% for trnL intron se-
quences and A p 29.6%, C p 20.1%, G p 15.7%, and T
p 34.5% for trnL-trnF intergenic spacer sequences).
Pairwise nucleotide differences of aligned portions excluding
gaps were calculated. Pairwise nucleotide differences for trnL
intron sequences ranged from 0.25% (for I. lactea and I. pal-
MAKAREVITCH ET AL.—PHYLOGENY OF SIBERIAN IRIS SPECIES 235
Fig. 4 Phylogenetic relationships of the Siberian Iris species based on comparative analysis of DNA sequences of the trnL intron and trnL-
trnF intergenic spacer. Clusters formed by Iris species are shown as numbers. Species groups are denoted as Roman numerals. Bootstrap percentages
with a frequency of more than 50% are shown in the nodes of the dendrogram. Signs under branches represent position of indels and correspond
to signs in fig. 3.
lasii) to 15.17% (for I. humilis and I. sibirica), with an average
of 4.65%. Pairwise nucleotide differences for trnL-trnF inter-
genic spacer ranged from 0.74% (for I. lactea and I. pallasii)
to 38.81% (for I. ivanovae and I. sibirica), with an average
of 19.38%.
Phylogenetic Analysis
Because of the different rate of evolution and the different
nucleotide composition, phylogenetic analyses were done using
the sequence data of each region separately. Indels were ex-
cluded from the analysis or were treated as single nucleotide
substitutions. The topologies of the trees constructed by
neighbor-joining, maximum parsimony, and maximum likeli-
hood methods with or without indels were consistent, differing
only in the position of species within clusters. The bootstrap
probabilities calculated based on 1000 replications were high,
even in the analysis of separate sequences. The topologies of
trees constructed based on separate analysis of the trnL intron
and trnL-trnF intergenic spacer sequences were similar, with
several exceptions in the positions of species within clusters.
Based on the phylogenetic trees and the distribution of indels,
assuming that the same indels are less likely to occur several
times in different branches independently, all of the species
could be grouped into eight clusters (fig. 3): 1, I. bloudowii,
I. potanini, and I. humilis;2,I. ivanovae, I. tigridia, and I.
glaucescens;3,I. tenuifolia, I. loczyi, and I. ventricosa;4,I.
halophila and I. ludwigii;5,I. ruthenica and I. uniflora;6,I.
laevigata and I. pseudacorus;7,I. pallasii, I. lactea, and I.
biglumis;8,I. sibirica, I setosa, and I. sanguinea.
The grouping of the species into clusters coincides with re-
cent classifications. The species of all clusters except of cluster
1 have supercluster-specific insertion of 4 bp in the trnL-trnF
intergenic spacer region. The species of clusters 3–8 have two
specific deletions of 6 bp in the trnL intron sequences. The
species of clusters 6–8 have a specific deletion of two nucle-
otides in the trnL-trnF intergenic spacer region sequences. The
species of clusters 2, 3, 5, 6, 7, and 8 each have from one to
six cluster-specific indels in the sequences analyzed. Species I.
tigridia, I. glaucescens, and I. halophila have one to two
species-specific indels. Only two indels were common in the
species of different clusters. The species of clusters 1, 3, and
4 have one specific insertion of one nucleotide in the trnL-trnF
intergenic spacer region; however, the species of cluster 2 do
not have such an insertion. A species from cluster 2, I. glau-
cescens, shares the 6 bp deletion specific for clusters 3–8. These
indels seem to appear independently, probably from the specific
sequence structure of the region. In cases when some sequences
had smaller indels in the regions of larger indels in other se-
quences, species with larger indels were considered to possess
only one larger indel, not both larger and smaller indels.
236 INTERNATIONAL JOURNAL OF PLANT SCIENCES
Only the location of I. ensata is different in trees based on
both sets of data. On the tree constructed based on the trnL
intron sequences, I. ensata is included in cluster 6. On the tree
constructed based on the trnL-trnF intergenic spacer se-
quences, I. ensata forms a separate branch close to clusters 4
and 5. Iris ensata does not share cluster-specific indels for
clusters 4, 5, or 6. On the RAPD tree, I. ensata is nested within
the clade C comprised of clusters 5 and 6 with a low support
coefficient. Therefore, the location of I. ensata seems to be
undetermined.
Since the topology of the trees based on the analysis of
different regions was similar, we combined all of the data and
constructed a phylogenetic tree based on the total alignment,
971 bp in length. The topologies of the combined trees con-
structed by all three methods used were consistent with each
other and with the topologies of trees constructed based on
separate analysis of both trnL intron and trnL-trnF intergenic
spacer. The resulting tree is shown in figure 4. To verify the
reliability of our results we tested different species as an out-
group (B. chinensis, Aristea coerulea, and Isophysis tasmanica;
accession numbers AJ409610, AJ290319, and AJ290317, re-
spectively; Reeves et al. 2001). The topologies of the evolu-
tionary trees constructed with different outgroup species dif-
fered only in the position of several species within clusters,
which kept the position and volume of the clusters and the
place of P. dichotoma the same. The genus B. chinensis closest
to irises (Souza-Chiez et al. 1997; Chase et al. 2000) gave the
highest bootstrap coefficients.
Pardanthopsis dichotoma is closer to the outgroup species
than to Iris species on the trees based on the analysis of both
regions and the combined data set. These results support the
detachment of P. dichotoma into a different genus (Lenz 1972)
and its placement as a distinct genus based on a recent mo-
lecular study of intergeneric relationships within Iridaceae
(Reeves et al. 2001). By Mathew’s classification (1989), all of
the Siberian Iris species form two major branches correspond-
ing to subgenera Limniris (beardless irises) and Iris (bearded
irises). On the combined tree in our study, all of the eight
clusters described above could be observed. These clusters form
four phylogenetic groups corresponding to the clades of species
observed on the tree constructed based on the RAPD analysis:
group I (clade A), bearded irises, corresponding to subgenus
Iris, includes clusters 1 and 2; group II (clade B) consists of
clusters 3 and 4; group III forms cluster 5; and group IV in-
cludes clusters 6, 7, and 8 (fig. 4). Groups III and IV were
clustered together into clade C. Groups II, III, and IV corre-
spond to beardless irises, subgenera Xyridion and Limniris.
The low bootstrap coefficient (28%) in the node corresponding
to group III could be explained by the influence of the unde-
termined location of I. ensata. Iris ensata is included in cluster
5 with a small bootstrap coefficient.
In most recent classification systems (Rodionenko 1987;
Mathew 1989), the groups of species are similar, but the rank
of each group and its position in the system vary. Our results
support Mathew’s classification (1989). Species of cluster 4,
I. halophila and I. ludwigii, and of cluster 7, I. biglumis, I.
pallasii, and I. lactea, although forming separate clusters, are
close to other Limniris species, which makes the detachment
of these species into the subgenera Xyridion sensu Rodionenko
and Eremiris sensu Doronkin unfounded. Further taxonomic
classification and circumscription of subgenera were beyond
the scope of this study, and according to our data, they require
a much broader species sampling covering the entire genus.
This study provides initial insights into the phylogeny of the
diverse and complicated genus Iris. The results of the study
suggest that the phylogeny and taxonomic structure of the
genus Iris may need reconsideration. To reach a robust phy-
logeny, additional DNA regions from plastid and nuclear ge-
nomes should be analyzed. This will allow testing the present
hypothesis based on the noncoding trnL-trnF region of the
chloroplast DNA using several different phylogenetic markers
on a broader sampling of the Iris species.
Acknowledgments
We are grateful to Dr. V. M. Doronkin, Dr. G. P. Semenova,
and Dr. Y. V. Ovchinnikov for providing the plant material
and to Dr. J. Armstrong, Dr. A. Gibbs, Dr. R. Peakall, and Dr.
G. Weiller for the RAPD software.
Literature Cited
Artyukova EV, MM Kozyrenko, MV Ilyshko, YN Zhuravlev, GD Reu-
nova 2001 Genetic variability of Iris setosa. Mol Biol 35:134–138.
Asmussen CB, MW Chase 2001 Coding and non-coding plastid DNA
in palm systematics. Am J Bot 88:1103–1117.
Bartish IV, K Rumpunen, H Nybom 2000 Combined analyses of
RAPDs, cpDNA and morphology demonstrate spontaneous hybrid-
ization in the plant genus Chaenomeles. Heredity 4:383–392.
Blattner FR, K Weising, G Banfer, U Maschwitz, B Fiala 2001
Molecular analysis of phylogenetic relationships among Myrme-
cophytic macaranga species (Euphorbiaceae). Mol Phylogenet Evol
19:331–344.
Cerbah M, T Souza-Chies, MF Jubier, B Lejeune, S Siljak-Yakov-
lev 1998 Molecular phylogeny of the genus Hypochaeris using in-
ternal transcribed spacers of nuclear rDNA: inference for chromo-
somal evolution. Mol Biol Evol 15:345–354.
Chase MW, AY De Bruijn, AV Cox, G Reeves, PJ Rudall, MAT John-
son, LE Eguiarte 2000 Phylogenetics of Asphodelaceae (Aspara-
gales): an analysis of plastid rbcL and trnL-F DNA sequences. Ann
Bot 86:935–951.
Clark AG, CM Lanigan 1993 Prospects for estimating nucleotide di-
vergence with RAPDs. Mol Biol Evol 10:1096–1111.
Comincini S, M Sironi, C Bandi, C Giunta, M Rubini, F Fontana
1996 RAPD analysis of systematics relationships among the Cer-
vidae. Heredity 76:215–221.
Cros J, MC Combes, P Trouslot, F Anthony, S Hamon, A Charrier, P
Lashermes 1998 Phylogenetic analysis of chloroplast DNA varia-
tion in Coffea L. Mol Phylogenet Evol 9:109–117.
Doronkin VM 1987 Iridaceae Juss. Pages 113–125 in Flora of Siberia
(Araceae–Orchidaceae). Nauka, Novosibirsk. (In Russian.)
——— 1990 The synopsis of Siberian species of the genus Iris (Iri-
daceae). Bot Zh 75:409–416. (In Russian.)
Ellis JR 1997 Chromosomes and the genus Iris. Pages 8–10 in A
MAKAREVITCH ET AL.—PHYLOGENY OF SIBERIAN IRIS SPECIES 237
guide to species irises: their identification and cultivation. Cam-
bridge University Press, New York.
Espinasa L, R Borowsky 1998 Evolutionary divergence of AP-PCR
(RAPD) patterns. Mol Biol Evol 15:408–414.
Felsenstein J 1985 Confidence limits on phylogenies: an approach us-
ing the bootstrap. Evolution 39:783–791.
——— 1993 PHYLIP (phylogeny inference package), version 3.5c.
University of Washington, Seattle.
Fernandez IA, JF Aguilar, JL Panero, GN Feliner 2001 A phylogenetic
analysis of Doronicum (Asteraceae, Senecioneae) based on mor-
phological, nuclear ribosomal (ITS), and chloroplast (trnL-F) evi-
dence. Mol Phylogenet Evol 20:41–64.
Freshwater DW, S Fredericq, BS Butler, MH Hommersand, MW
Chase 1994 A gene phylogeny of the red algae (Rhodophyta) based
on plastid rbcL. Proc Natl Acad Sci USA 91:7281–7285.
Fritsch PW 2001 Phylogeny and biogeography of the flowering plant
genus Styrax (Styracaceae) based on chloroplast DNA restriction
sites and DNA sequences of the internal transcribed spacer region.
Mol Phylogenet Evol 19:387–408.
Fukuda T, J Yokoyama, H Ohashi 2001 Phylogeny and biogeography
of the genus Lycium (Solanaceae): inferences from chloroplast DNA
sequences. Mol Phylogenet Evol 19:246–258.
Gehrig H, O Gaussmann, H Marx, D Schwarzott, M Kluge 2001
Molecular phylogeny of the genus Kalanchoe (Crassulaceae) inferred
from nucleotide sequences of the ITS-1 and ITS-2 regions. Plant Sci
160:827–835.
Goldblatt P 1990 Phylogeny and classification of Iridaceae. Ann Mo
Bot Gard 77:607–627.
Goldblatt P, M Takei 1997 Chromosome cytology of Iridaceae: pat-
terns of variation, determination of ancestral base numbers, and
modes of karyotype change. Ann Mo Bot Gard 84:285–304.
Klaas M 1998 Applications and impact of molecular markers on evo-
lutionary and diversity studies in the genus Allium. Plant Breed 117:
297–308.
Lenz LW 1972 The status of Pardanthopsis (Iridaceae). Aliso 7:4.
Lewis CE, JJ Doyle 2001 Phylogenetic utility of the nuclear gene ma-
late synthase in the palm family (Arecaceae). Mol Phylogenet Evol
19:409–420.
Lia VV, VA Confalonieri, CI Comas, JH Hunziker 2001 Molecular
phylogeny of Larrea and its allies (Zygophyllaceae): reticulate evo-
lution and the probable time of creosote bush arrival to North Amer-
ica. Mol Phylogenet Evol 21:309–320.
Lyte C 1997 The Iris in history. Pages 1–4 in A guide to species irises:
their identification and cultivation. Cambridge University Press,
New York.
Mathew B 1989 The Iris. Oregon Timber, Portland, Oreg. 326 pp.
Maynard P 1997 Cultivation and the great genus. Pages 4–8 in A
guide to species irises: their identification and cultivation. Cambridge
University Press, New York.
Nei M, WH Li 1979 Mathematical model for studying genetic var-
iation in terms of restriction endonucleases. Proc Natl Acad Sci USA
76:5269–5273.
Nozaki H, K Misawa, T Kajita, M Kato, S Nohara, MM Watana-
be 2000 Origin and evolution of the colonial volvocales (Chloro-
phyceae) as inferred from multiple, chloroplast gene sequences. Mol
Phylogenet Evol 17:256–268.
Ranade SA, N Farooqui, E Bhattacharya, A Verma 2001 Gene tag-
ging with random amplified polymorphic DNA (RAPD) markers for
molecular breeding in plants. Crit Rev Plant Sci 20:251–275.
Reeves G, MW Chase, P Goldblatt, P Rudall, MF Fay, AV Cox, B
Lejeune, T Souza-Chies 2001 Molecular systematics of Iridaceae:
evidence from four plastid DNA regions. Am J Bot 88:2074–2087.
Richardson JE, MF Fay, QC Cronk, D Bowman, MW Chase 2000 A
phylogenetic analysis of Rhamnaceae using rbcL and trnL-F plastid
DNA sequences. Am J Bot 87:1309–1324.
Rodionenko GI 1987 The genus Iris L.: questions of morphology,
biology, evolution, and systematics. British Iris Society, London. 222
pp.
Scovel G, H Benmeir, M Ovadis, H Itzhaki, A Vainstein 1998 RAPD
and RFLP markers tightly linked to the locus controlling Carnation
(Dianthus caryophyllus) flower type. Theor Appl Genet 96:
117–122.
Shimizu K, Y Miyabe, H Nagaike, T Yabuya, T Adachi 1999
Production of somatic hybrid plants between Iris ensata Thumb.
and I. germanica L. Euphytica 107:105–113.
Soliva M, A Kocyan, A Widmer 2001 Molecular phylogenetics of the
sexually deceptive orchid genus Ophrys (Orchidaceae) based on
nuclear and chloroplast DNA sequences. Mol Phylogenet Evol 20:
78–88.
Soltis ED, PS Soltis 2000 Contributions of plant molecular system-
atics to studies of molecular evolution. Plant Mol Biol 42:45–75.
Souza-Chies TT, G Bittar, S Nadot, L Carter, E Besin, B Lejeune
1997 Phylogenetic analysis of Iridaceae with parsimony and dis-
tance methods using the plastid gene rps4. Plant Syst Evol 204:
109–123.
Sun GL, O Diaz, B Salomon, R von Bothmer 1999 Genetic diversity
in Elymus caninus as revealed by isozyme, RAPD, and microsatellite
markers. Genome 42:420–431.
Svitashev S, TLX Bryngelsson, RR Wang 1998 Genome-specific re-
petitive DNA and RAPD markers for genome identification in Ely-
mus and Hordelymus. Genome 41:120–128.
Swoboda I, PL Bhalla 1997 RAPD analysis of genetic variation in the
Australian fan flower, Scaevola. Genome 40:600–606.
Taberlet P, L Gielly, G Pautou, J Bouvet 1991 Universal primers for
amplification of three non-coding regions of chloroplast DNA. Plant
Mol Biol 17:1105–1109.
Welsh J, M McCleland 1990 Fingerprinting genomes using PCR with
arbitrary primers. Nucleic Acids Res 18:7213–7218.
Wen J 2000 Internal transcribed spacer phylogeny of the Asian and
eastern North American disjunct Aralia sect: Dimorphanthus (Ara-
liaceae) and its biogeographic implications. Int J Plant Sci 161:
959–966.
Williams JGK, AR Kubelik, KJ Livak, JA Rafalski, SV Tingey
1990 DNA polymorphism amplified by arbitrary primers are useful
as genetic markers. Nucleic Acids Res 18:6531–6535.
Wu Q-G, DF Cutler 1985 Taxonomic, evolutionary and ecological
implications of the leaf anatomy of rhizamatous Iris species. Bot J
Linn Soc 90:253–303.
Zhuravlev YN, MM Kozyrenko, EV Artyukova, GD Reunova, MV
Ilyushko 1998 Fingerprinting genomes of the Far Eastern species
of the genus Iris L. by RAPD PCR. Russ J Genet 34:368–372.
