Vous êtes sur la page 1sur 17

American Journal of Botany 99(4): 778–794. 2012.

SYSTEMATICS AND EVOLUTION OF ARCTIC-ALPINE ARABIS


ALPINA (BRASSICACEAE) AND ITS CLOSEST RELATIVES IN
THE EASTERN MEDITERRANEAN1

ROBERT KARL2, CHRISTIANE KIEFER3, STEPHAN W. ANSELL4, AND MARCUS A. KOCH2,5


2Department
of Biodiversity and Plant Systematics, Centre for Organismal Studies (COS) Heidelberg, Heidelberg University
69120 Heidelberg, Germany; 3Max-Planck-Institute for Plant Breeding 50829 Cologne, Germany; and 4Natural History Museum
London, Department of Botany, London, SW7 5BD, UK

• Premise of the study: The high mountains in southern Anatolia and the eastern Mediterranean are assumed to play a major role
as a primary center of genetic diversity and species richness in Eurasia. We tested this hypothesis by focusing on the wide-
spread perennial arctic-alpine Arabis alpina and its sympatrically distributed closest relatives in the eastern Mediterranean.
• Methods: Plastid (trnL intron, trnL-F intergenic spacer) and nuclear (ITS) DNA sequence analysis was used for phylogenetic
reconstruction. Broad-scale plastid haplotype analyses were conducted to infer ancestral biogeographic patterns.
• Key results: Five Arabis species, identified from the eastern Mediterranean (Turkey mainland and Cyprus), evolved directly
and independently from A. alpina, leaving Arabis alpina as a paraphyletic taxon. These species are not affected by hybridiza-
tion or introgression, and species divergence took place at the diploid level during the Pleistocene.
• Conclusions: Pleistocene climate fluctuations produced local altitudinal range-shifts among mountain glacial survival areas,
resulting not only in the accumulation of intraspecific genotype diversity but also in the formation of five local species. We also
show that the closest sister group of Arabis alpina consists exclusively of annuals/winter annuals and diverged prior to Pleis-
tocene climatic fluctuations during the colonization of the lowland Mediterranean landscape. These findings highlight that
Anatolia is not only a center of species richness but also a center for life-history diversification.

Key words: annual species; Arabis alpina; Brassicaceae; eastern Mediterranean; evolutionary history; perennial species;
Pleistocene.

In the last 15 years, phylogeographic studies have character- distributed throughout the alpine habitats of Europe and into the
ized the evolutionary history of numerous Eurasian species arctic zones of Greenland and North America, the high moun-
(Taberlet et al., 1998; Hewitt, 1999, 2000; Willis and Whittaker, tains of northern and eastern Africa, Anatolia, the Caucasus,
2000). Many of these studies have focused on time scales span- and the Near East (Meusel et al., 1965; Koch et al., 2006). This
ning the late Tertiary and Pleistocene (Comes and Kadereit, broad distribution makes this species an ideal system to study
1998, 2003; Hewitt, 2004; Birks, 2008) and provided new in- the effect of Pleistocene climatic fluctuations on speciation, ge-
sights into the biogeography of various taxa during past climate netic differentiation, and adaptation over different geographic
change. Organisms were forced either to escape, migrate, and scales, including those regions important for its long-term sur-
eventually recolonize or to stay, adapt, and survive the glacia- vival and speciation. For example, studies on various organ-
tions (Petit et al., 2003; Huck et al., 2009; Schmitt, 2009). isms have highlighted the importance of southern Anatolia
Whereas most studies have used a single species approach or a (Taurus mountain systems) for Pleistocene survival and specia-
comparative analysis of taxonomically unrelated organisms tion (e.g., Font et al., 2009; Parolly et al., 2010). Presently, Arabis
(Taberlet et al., 1998, Hewitt 2004), comparative phylogeogra- alpina is mostly treated as a well-defined species without sub-
phy focusing on closely related taxa should be informative for species or other segregates, although often the segregrate A.
addressing evolutionary questions, especially concerning trait caucasica Willd. (A. alpina subsp. caucasica) has been recog-
and adaptive variation (Hickerson et al., 2010). nized. This is despite few clear discriminating morphological
In this paper, we focus on the arctic-alpine plant Arabis al- characters (Plantholt, 1995) and a lack of genetic evidence to
pina L. and its close annual and perennial relatives (see below). consistently separate it from A. alpina (Plantholt, 1995; Koch
Arabis alpina, the alpine rock cress (Brassicaceae), is widely et al., 2006; Ansell et al., 2011).
Currently, the closest related sister taxa to A. alpina have not
yet been identified. Candidates for closest relatives to A. alpina
1 Manuscript received 14 September 2011; revision accepted 8 February 2012. based on morphological characters include the Central Asian
In addition to the various collections mentioned, we especially endemic A. tianschanica Pavlov and the Cyprian endemics A.
acknowledge receipt of plant material from D. German (South-Siberian cypria Holmboe and A. purpurea Sm. The latter species has
Botanical Garden, Barnaul, Russia), M. Hoffmann (IPK Gatersleben, pink to purple petals, a character that is also shared by the
Germany), B. Mutlu (Inönü University, Malatya, Turkey), and C. Kadis southern Anatolian endemic A. aubrietioides Boiss. Further
(Nature Conservation Unit, Frederick University, Cyprus). Funding was candidates for close relatives include A. deflexa Boiss., A. fari-
provided by the DFG (Deutsche Forschungsgemeinschaft) to M.A.K.
(KO2302/1-1, KO2302/11-1 and KO2302/12-1).
nacea Rupr., and A. ionocalyx Boiss. Understanding the evolu-
5 Author for correspondence (e-mail: marcus.koch@cos.uni-heidelberg.de) tionary and geographic relationships of this species group will
provide a framework for future studies focusing on different
doi:10.3732/ajb.1100447 and fundamental aspects of plant evolution and adaptation.

American Journal of Botany 99(4): 778–794, 2012; http://www.amjbot.org/ © 2012 Botanical Society of America

778
April 2012] KARL ET AL.—EVOLUTIONARY HISTORY OF ARABIS ALPINA 779

Various evolutionary studies have recently focused on the species or a species complex, we will explore the hypothesis
evolution of Arabis alpina in more detail, addressing global that the related species evolved in parallel in these regions from
phylogeographic issues (Koch et al., 2006; Ansell et al., 2011) different A. alpina populations. Furthermore, we will determine
or population differentiation on various spatial scales (Assefa whether any of the eight Mediterranean annual Arabis species
et al., 2007; Ehrich et al., 2007; Ansell et al., 2008). These stud- are sister to the perennial A. alpina group and test the hypothe-
ies revealed that this species originated in a defined area in cen- sis that the divergence of annuals and perennial taxa predates
tral Anatolia (Ansell et al., 2011), from where major waves of the Pleistocene and that they followed different evolutionary
emigration have shaped the current distribution range (Koch pathways in the Mediterranean in response to the Pleistocene
et al., 2006; Ansell et al., 2011): (1) migration to central and temperature oscillations.
northwestern Europe and further into the arctic, (2) westward to
Spain and north Africa, (3) migration into the Caucasus and the
Near East, (4) two independent migration events to the east MATERIALS AND METHODS
African high mountains. These waves of immigration span a
time scale of divergence of around 2 million years (Koch et al., Plant material—For the phylogenetic analysis, we included one outgroup
2006; Assefa et al., 2007; Ansell et al., 2008), which is consid- accession and 72 accessions of Arabis alpina and relatives of the tribe Ara-
bideae. These selected accessions comprised 29 species that have close mor-
ered to be the minimum age of Arabis alpina. phological similarity to Arabis alpina (e.g., Arabis tianschanica). The data set
Several studies have started to harness the potential of A. alpina also included all annuals assigned to Arabis [Arabis aucheri Boiss., Arabis au-
to investigate trait variation and adaptation in mountain habi- riculata Lam., Arabis erikii Mutlu, Arabis kennedyae Meikle, Arabis mont-
tats. This started 50 years ago with the pioneering work of bretiana Boiss., Arabis nova Vill., Arabis parvula Dufour, and Arabis verna
Hedberg (1962), who indicated that there are reproductive bar- (L.) R. Br.]. Six individuals of A. alpina were chosen as representatives of the
riers between the European and East African material of A. al- three major chloroplast DNA (cpDNA) haplotype lineages, which were identi-
fied earlier (Koch et al., 2006; Ansell et al., 2011). Three accessions were cho-
pina. More recently, Ansell et al. (2008) suggested putative sen from both the genus Draba (in its broadest sense, see Jordon-Thaden et al.,
changes in the breeding system coincided with historical differ- 2010) and the “core-Arabis” clade (see Koch et al., 2010). We selected Pseudo-
ences in glacial survival. This is supported by mating analysis, turritis turrita (L.) Al-Shehbaz as the outgroup because this taxon was previ-
which has identified a sporophytic self-incompatibility (SI) sys- ously identified as belonging to the sister tribe of the Arabideae: the Stevenieae
tem in some Italian populations and the loss of SI function in (German et al., 2009, Al-Shehbaz et al., 2011). Plant material was obtained
several Alps populations (Tedder et al., 2011). Elsewhere AFLP from the Botanical Garden and Botanical Museum Berlin (B), the Natural His-
tory Museum London (BM), University of Copenhagen (C), Royal Botanical
analysis of Alps populations has identified a remarkable num- Garden Edinburgh (E), Botanical Garden and Herbarium Heidelberg (HEID),
ber of polymorphic loci that are correlated to ecological varia- Botanische Staatssammlung Munich (M), Real Jardin Botanico Madrid (MA),
tion (Manel et al., 2010; Poncet et al., 2010). There is also the Missouri Botanical Garden (MO), IPK Gatersleben, Inönü University
increasing knowledge on the molecular basis of life-history Malatya in Turkey, the Frederick University Cyprus and the South-Siberian
variation in Arabis alpina. One such trait is the shift from an- Botanical Garden Barnaul in Russia.
nualism to perennialism and the regulation of perennial flower- For the biogeographic analyses, we added a compilation of additional cpDNA
data from 483 accessions representing 296 locations (Koch et al., 2006; Assefa
ing by PEP1 (PERPETUAL FLOWERING 1) in Arabis alpina et al., 2007; Ansell et al., 2008, 2011), giving a combined total of 518 acces-
(Wang et al., 2009), which may be considered a “key character sions from 329 locations.
trait” in the adaptation processes of an arctic-alpine plant. Detailed accession information, including GenBank accession numbers, are
Our study is part of a larger project to unravel the evolution- provided in the Appendix 1.
ary history and systematics of the whole tribe Arabideae, which
consists of more than 500 species. A series of studies has al- DNA extraction, PCR amplification and sequencing—DNA extractions
ready provided evidence for (1) the paraphyly of numerous taxa were carried out using the NucleoSpin Plant II Kit (Macherey-Nagel, Düren,
within the tribe Arabideae (including Arabis) and that (2) the Germany) or according to an isopropyl alcohol variant of the CTAB method
(Doyle and Doyle, 1987). The innuPREP Plant DNA Kit (Analytik Jena) was
tribe can be split into two subclades (Koch et al., 1999a, 2000, used for DNA extractions from seed material with the addition of an extra pre-
2001, 2007, 2010; German et al., 2009; Al-Shehbaz et al., filter and washing step. The internal transcribed spacer of nuclear ribosomal
2011). The Arabideae consist mostly of taxa from the two RNA including the 5.8S rRNA (ITS), chloroplast trnL intron (trnL) and trnL-
closely related genera Draba with more than 390 species and trnF intergenic spacer (trnL-F) were selected as markers and were amplified by
Arabis with more than 65 species. The fully unexplored genus PCR. Reactions were performed in a final volume of 25 µL, using 10 µM of
Aubrieta represents another 10–20 species (Koch et al., ongo- each primer, a total of 2.0 mmol/L MgCl2 and 0.5 U of MangoTaq polymerase
(Bioline). The primers used for ITS amplification were the 18F primer (5′-GGA
ing research). Preliminary phylogenetic analyses have focused AGG AGA AGT CGT AAC AAG G-3′) as modified by Mummenhoff et al.
on establishing the tribal and generic relationships (Jordon- (1997a) and the 25R primer (5′-TCC TCC GCT TAT TGA TAT GC-3′) de-
Thaden et al., 2010; Koch et al., 2010). Consequently, in this signed by White et al. (1990). The trnL intron was amplified using the universal
study we have excluded many taxa previously shown to be un- primers c and d of Taberlet et al. (1991) (c: 5′-CGA AAT CGG TAG ACG CTA
related or only distantly related to Arabis alpina. CG-3′, d: 5′-GGG GAT AGA GGG ACT TGA AC-3′), while the trnL-F inter-
Presently, it is unclear as to whether Arabis alpina represents genic spacer was amplified with primer e of Taberlet et al. (1991) (5′-GGTTCA
AGT CCC TCT ATC ATC CC-3′) and a primer designed by Dobeš et al. (2004)
a single well-defined species or a species complex. However, (5′-GAT TTT CAG TCC TCT GCT CTA C-3′). The amplifications were run on
there are various species that are obviously morphologically a PTC 200 Peltier Thermal Cycler (MJ Research) under the following condi-
very similar to Arabis alpina and are often distributed sympatri- tions: 3 min initial denaturation at 95°C; followed by 30 cycles of 30 s at 95°C,
cally. On the basis of morphology and distribution, we have 30 s at 48°C and 1 min at 72°C; and 10 min final elongation at 72°C. PCR
selected candidate taxa that might be members of an Arabis products were purified by centrifugation in a NucleoFast 96 PCR Plate (Mach-
alpina species complex and asked how the morphologically erey-Nagel) and directly sequenced using the above primers by the commercial
sequencing service GATC (Konstanz, Germany). The ITS (Appendix S1, see
similar species relate to the evolutionary framework of the A. Supplemental Data with the online version of this article) and the combined
alpina lineages, especially in connection to those found in the east- trnL intron and trnL-F intergenic spacer sequences (Appendix S2, see online
ern Mediterranean, Anatolia and Central Asia. In answering the Supplemental Data) were checked and trimmed using the program SeqMan II
question as to whether A. alpina represents a single well-defined v.6.1 (DNAStar, Madison, Wisconsin, USA) (hereafter designated trnLF) and
780 AMERICAN JOURNAL OF BOTANY [Vol. 99

TABLE 1. Parameters and results of the three divergence-time estimate approaches based on the nrDNA ITS marker. Mean values are indicated in boldface;
the lower and upper bounds of the 95% highest posterior density (HPD) interval are given above and below that value, respectively. Enforced bounds
are indicated in italics.

Using secondary calibration of Like previous + fixed


Fossil based calibration of node age of node age of tribe Arabideae (acc. mutational rates
Parameter tribe Thlaspideae (acc. Beilstein et al., 2010) Couvreur et al., 2009) (Kropf, 2002; Kay et al., 2006)

Runs / Generations 4 independent runs with 4 independent runs with 4 independent runs with
30 million generations each 20 million generations each 20 million generations each
Log likelihood −3103.058 −3103.198 −3103.217
−3090.494 −3090.620 −3090.576
−3077.935 −3078.138 −3078.295
tmrca of Arabis alpina aggregate and annual 2.84 0.94 0.94
sister group: clade A (Fig. 1) (Ma) 7.13 2.24 2.25
12.3 3.94 3.98
tmrca of Arabis alpina aggregate (Ma) 1.49 0.51 0.52
4.02 1.26 1.27
7.02 2.25 2.25
tmrca of the core-Arabis (Ma) 4.58 1.54 1.63
12.5 3.91 3.94
21.5 6.95 7.04
tmrca of Draba and its segregates (Ma) 8.41 2.98 3.03
19.3 6.09 6.09
31.7 10.2 10.2
tmrca of tribe Arabideae (Ma) 22.1 10.0 10.0
44.4 14.1 14.1
69.5 21.5 21.5
ucld.mean 0.85 × 10−9 2.66 × 10−9 2.69 × 10−9
1.60 × 10−9 5.10 × 10−9 5.08 × 10−9
2.47 × 10−9 7.37 × 10−9 7.41 × 10−9
meanRate 0.88 × 10−9 2.75 × 10−9 2.75 × 10−9
1.59 × 10−9 5.12 × 10−9 5.10 × 10−9
2.41 × 10−9 7.35 × 10−9 7.38 × 10−9
Bayes factors — / 1.09 / 1.38 0.92 / — / 1.26 0.73 / 0.79 / —

then manually aligned using the program GeneDoc v2.6.002 (Nicholas, 1997). The parsimony heuristic search was performed under the following settings:
In addition, a combined data set was formed by merging the ITS and the trnLF gaps were treated as missing data, the “mstaxa” option was set as “uncertain”,
sequences (online Appendix S3). Identical sequences were excluded from the tree construction was via stepwise addition, tree-bisection-reconnection (TBR)
final alignments prior to phylogenetic analysis. was used as the branch-swapping algorithm, the MaxTrees limit was set to
10 000, and the “MulTrees” option was in effect (saving all minimal trees found
Phylogenetic analyses—Bayesian Markov chain Monte Carlo (MCMC) during branch swapping). A strict consensus tree was calculated from most
analyses (Yang and Rannala, 1997) were performed with the mpi (message parsimonious trees retained. For bootstrapping (Felsenstein, 1985), 100 repli-
parsing interface) version of the program MrBayes v.3.1.2 (Ronquist and cates with a maximum of 250 retained trees were run.
Huelsenbeck, 2003). To substantiate the results of the Bayesian analyses, we under- For the maximum likelihood (ML) analyses, we ran two replicates per data
took additional analyses using maximum parsimony (MP) running the program set, and each replicate was terminated under the condition that the ln likelihood
PAUP* 4.0b10 (Swofford, 2002) and maximum likelihood using the program value did not increase significantly (0.01) for the previous 1 000 000 genera-
Garli v0.96 (Zwickl, 2006). The best-fitting nucleotide substitution models tions. Analyses were conducted using 1000 bootstrapping replicates with the
were chosen using the program Modeltest v3.7 (Posada and Crandall, 1998), lowered termination condition of 10 000 generations without significant ln L
which relied on the Akaike information criterion (AIC) and the Bayesian infor- change. The corresponding alignments to the ITS, trnLF, and combined data
mation criterion (BIC). For the combined analyses, the data set was divided into are provided with online Appendices S1, S2, and S3, respectively. Graphical
nrDNA and cpDNA partitions, so that each partition could be run under the edition of all trees was executed in Adobe Illustrator CS4 (San Jose, California,
respective model. In the Bayesian analyses, four simultaneous runs with four USA).
chains each were run for 20 (ITS and trnLF data set) or 30 (combined data set)
million generations. The temperature of the heated chain was set to 0.01, as this Congruency test—A Shimodaira-Hasegawa test (Shimodaira and Hasegawa,
facilitated the most efficient chain-swapping. For each run, 1001 trees were 1999) was performed to identify congruencies between the tree topologies
sampled, and the first 25% of the trees sampled were discarded (burn-in = 250) of the ITS, trnLF, and combined marker data sets using PAUP* 4.0b10
when computing the consensus tree (50% majority rule) Completion of the (Swofford, 2002). Before the tests, the number of accessions was reduced to the
Bayesian runs was determined using the web-based program AWTY (Nylander 40 common to all three data sets. As a consequence, the number of tree to-
et al., 2007). Comparison of the posterior probabilities of all splits between the pologies was very small (n = 2) and might have biased the levels of signifi-
respective runs and cumulative split frequency plots showed that convergence cance. We therefore added 100 random topologies to the congruency tests to
between the Bayesian MCMC runs had been reached. minimize this problem.


Fig. 1. Phylogenetic tree of the Bayesian analysis based on the nrDNA ITS data set. Posterior probability values (first-mentioned) and bootstrap values
of the maximum-likelihood analysis (middle-mentioned), and the maximum parsimony analysis (last-mentioned) of the nodes are given along the corre-
sponding branches. Pseudoturritis turrita was used as the outgroup taxa. Accessions with a prefixed asterisk had the same sequence as the top-listed acces-
sion and were therefore excluded from the tree calculation. Purple- or pink-flowered species are indicated with a black spot following the taxon label;
species that occasionally produce pink petals are marked with a gray spot. Accessions without any spot correspond to primarily white-flowered species.
Annual/winter-annual taxa are labeled with the corresponding sign for annualism, also following the taxon label.
April 2012] KARL ET AL.—EVOLUTIONARY HISTORY OF ARABIS ALPINA 781
782 AMERICAN JOURNAL OF BOTANY [Vol. 99

Network analyses—Network analysis was performed with the program 2002; Mummenhoff et al., 2004; Kay et al., 2006). The most appropriate evolu-
TCS v.1.21 (Clement et al., 2000), which uses statistical parsimony (Templeton tionary model was estimated using the program Modeltest v.3.7 (Posada and
et al., 1992) as the underlying method of estimation. Taxa identified to be dis- Crandall, 1998), based on the Bayesian information criterion (BIC).
tantly related to Arabis alpina were excluded from the alignment previously Both ITS and trnLF divergence-time estimations were run under the lognor-
used for phylogenetic analaysis. We utilized accessions of Arabis brachycarpa mal uncorrelated relaxed clock method and the birth–death speciation model
Rupr., A. nordmanniana Rupr., A. graellsiiformis Hedge, and A. christiani (Gernhard, 2008), because the calculated Bayes factors (Nylander et al., 2004)
N.Busch to root the network. For the ingroup, we included all cpDNA haplotypes favored their use over a strict clock and a pure birth model (Yule, 1924). Num-
published earlier by Koch et al. (2006) (DQ060112–DQ060138, DQ060140– ber of runs and generations are given in Table 1 (ITS) and 2 (trnLF). In each
DQ060142, DQ060144–DQ060145), Assefa et al. (2007) (EF449508–EF449514), run, 10 001 generations were sampled. Each run was checked in the program
Ansell et al. (2008) (EU403083–EU403084), and Ansell et al. (2011) (JF705219– Tracer v.1.4.1 (Rambaut and Drummond, 2007) for having reached stationary
JF705252), giving a combined data set of 115 haplotypes from of 518 acces- phase and a sufficient effective sample size (ESS over 200, as suggested by the
sions, covering 329 locations and 10 taxa (A. cypria, A. purpurea, A. kennedyae, authors). All runs (log- and trees-files, respectively) were then combined with
A. alpina, A. aubrietoides, A. tianschanica, A. deflexa, A. ionocalyx, A. mont- the LogCombiner program included in the Tracer package.
bretiana, A. nova subsp. iberica Rivas Mart. ex Talavera).
The alignment was adjusted further by excluding a microsatellite-like region Chromosome counts and ploidy level estimates—Most of the leaf material
at the 5′-end of the trnL-F intergenic spacer (alignment position 365–380, was obtained directly from wild sources (e.g., preserved herbarium vouchers)
Appendix S2), and we also excluded the trnL gene from the analysis, so that all with supplementary material grown from seed. Root tips were incubated in
sequences had comparable information content. This led to the exclusion of a 2 mmol/L 8-hydroxyquinoline for 3 h at 15°C and subsequently transferred to
series of nearly identical haplotypes shown in Ansell et al. (2011) (haplotypes freshly prepared ice-cold ethanol–acetic acid (3 : 1). After 2 h of incubation at
06, 22, 23, 24, 27, 32, 48, 52, 53, 55, 57, 64, 65, 66, 67, 71, B, and D). Redun- 4°C, the root tips were transferred to 70% ethanol and stored at −20°C for sub-
dant haplotypes from newly analyzed accessions of our study were also removed sequent analysis. Root tips were washed in water, equilibrated in 10 mmol/L
(MG003, RK240, RK244, RK249, and RK252), as were haplotypes 16 and 31 citrate buffer, and then hydrolyzed with an enzyme solution (0.1% cellulase,
from Koch et al. (2006). Furthermore, to be conservative when interpreting 0.1% pectolyase, 0.1% cytohelicase, and 1 mmol/L citrate buffer) at 37°C fol-
nucleotide variation, we also excluded sequence types that differed by indels in lowed by squashing and fixing in 45% acetic acid and staining with a 1 : 20 dilu-
the beginning and end portions of the PCR fragments, where sequencing qual- tion of DAPI/Vectashield (Vector Laboratories, Burlingame, California, USA).
ity may be insufficient. This excluded haplotypes E and F (Assefa et al., 2007) Additional information on chromosome numbers from various Arabis species
and several newly analyzed accessions of our study (A53M, MA001, MA004, was obtained from published accounts (Warwick and Al-Shehbaz, 2006; Koch
RK001, RK002, RK004, RK006, RK086, RK087, RK088, RK167, and et al., 2010).
RK250). Finally, indels of more than one base pair in length were recoded with
binary characters (online Appendix S4). This resulted in 73 unique haplotypes
for minimum spanning network reconstruction.
RESULTS
Divergence-time estimates—Divergence-time estimates in the Brassicaceae
have been discussed widely, because calculations led to highly diverging re- Nuclear ITS region— The ITS alignment comprised 45 se-
sults, depending on the markers and calibration points used (refer to Franzke quences (representing 71 accessions) and 631 characters, of
et al. [2011] for further discussion). Using the software package BEAST v.1.4.8 which 448 were constant. Of the 183 variable characters, 132
(Drummond and Rambaut, 2007), we performed three courses of divergence-
time estimate calculations employing two different calibration points. (1) The were phylogenetically informative, and the remaining 51 were
primary calibration was based on an Oligocene fruit fossil (Thlaspi primae- autapomorphic positions (Appendix S1). The SYM+Γ evolu-
vum), which was assigned to the tribe Thlaspideae (Manchester and O’Leary, tionary model was applied to the Bayesian and ML analyses,
2010). This fossil, dated at 29.2−30.8 Ma (Wing, 1987), has been recently used the results are shown in Fig. 1. The resulting Bayesian trees had
for age estimates of the Brassicales (Beilstein et al., 2010). (2) Secondary cali- a harmonic mean for the ln likelihood of −2794.51 and an aver-
bration used the estimated age of the most recent common ancestor of the tribe age standard deviation of split frequencies of 0.010. The corre-
Arabideae (mean age 16.8 Myr) as shown by Couvreur et al. (2009). (3) We
additionally applied fixed limits for the rates of variation on the secondary cali- sponding tree of the ML analysis had a ln likelihood of −2678.34.
bration point approach, following Koch et al. (2010). Two representatives of The maximum parsimony analysis resulted in 571 most parsi-
the tribe Thlaspideae [Thlaspi arvense L. AF336152 (ITS1), AF336153 (ITS2), monious trees with a tree length of 326 (with “maximum length =
AY122461 (trnLF); Alliaria petiolata (M.Bieb.) Cavara & Grande AF283492 0” rule applied), consistency index (CI) of 0.66 (autapomor-
(ITS1), AF 283493 (ITS2), JN189740 (trnL), JN189781 (trnF)] were added to phies excluded) and retention index (RI) of 0.90.
both the ITS and cpDNA alignments, used for the phylogenetic analyses. Both The ITS tree based on Bayesian analysis (Fig. 1) is subdivided
taxa were also used by Beilstein et al. (2010) to define the tribe Thlaspideae.
Both original alignments were modified only by the introduction of additional into seven major clades, which are very well supported by both
gaps. Taxon subsets were specified for some clades as designated in the phylo- posterior probabilities and bootstrap values. Clade E comprises
genetic analysis, including a denotation for the Thlaspideae (Thlaspi arvense the three species of the genus Aubrieta and the circum-Mediterra-
and Alliaria petiolata) and the Arabideae (excluding Pseudoturritis turrita and nean Arabis verna as sister to the genus. While this clade is
both members of the Thlaspideae). For the fossil-based primary calibrations, basal to all other clades, it lacks strong statistical support. Clade
the boundaries for the age of the most recent common ancestor (tmrca) of the D consists of Draba and its segregates and is sister to clade L,
taxon set “Thlaspideae” were set between 29.2 and 30.8 Myr. For the secondary
calibration point, the boundaries for the tmrca of the taxon set “Arabideae” were which contains four Arabis species from the Caucasus moun-
set between 10.0 and 23.9 Myr. Additionally, for the calculations with fixed tains (A. brachycarpa, A. christiani, A. graellsiiformis, and A.
mutational rates these rates were fixed within the range of 1.72 × 10−9 to 1.71 nordmanniana). This clade (hereafter called A. nordmanniana
× 10−8 (ITS) and 3.6 × 10−9 and 7.7 × 10−9 (trnLF), respectively. These values group) shows strong congruence with the former section Allari-
reflect the published ranges for mutation rates in herbaceous plants (Kropf, aopsis, as described by Busch (1906) and Schulz (1936). Clade


Fig. 2. Phylogenetic tree from the Bayesian analysis based on the cpDNA trnLF data set. Posterior probability values (first-mentioned) and bootstrap
values of the maximum-likelihood analysis (middle-mentioned) and the maximum-parsimony analysis (last-mentioned) of the nodes are given along the
corresponding branches. Pseudoturritis turrita was used as the outgroup taxa. Accessions with a prefixed asterisk had the same sequence than the top-listed
accession and were therefore excluded from the tree calculation. Purple- or pink-flowered species are indicated with a black spot following the taxon label;
species that occasionally produce pink petals are marked with a gray spot. Accessions without any spot correspond to primarily white-flowered species.
Annual/winter-annual taxa are labeled with the corresponding sign for annualism, also following the taxon label.
April 2012] KARL ET AL.—EVOLUTIONARY HISTORY OF ARABIS ALPINA 783
784 AMERICAN JOURNAL OF BOTANY [Vol. 99

M is formed by the three representatives of the core Arabis, a and perennial as shown in the ITS analyses but rather formed a
group that has been shown to contain the majority of Arabis single polytomy. The predominantly east African “red-coded”
species (see Koch et al., 2010). Clade P comprises the Ibero- haplotype group (Fig. 3) is represented by accessions A22JO
Moroccan Arabis taxa, A. parvula, and its east-Mediterranean and A40JO2 of A. alpina, to which one accession of the Ana-
sister species A. aucheri. Clades D, L, M, and P form a weakly tolian endemic A. aubrietioides (A01M) and the Syrian ac-
supported monophyletic group, as indicated by posterior prob- cession of A. ionocalyx (RK254) are basal. The mainly
ability or bootstrap values. Arabis auriculata and A. nova subsp. western Asian “green-coded” haplotype group (Fig. 3) is rep-
nova are sister to each other in clade B. The accession of A. alpina resented by accessions A01JO2 and AW-30 of A. alpina and
and all its morophologically similar species are confined to by three accessions of A. aubrietioides (MG003, RK244,
clade A, which is further divided into subclades of annual and RK261) and one accession of each A. deflexa (RK249) and A.
perennial taxa, respectively. The annual subclade consists of ionocalyx (RK253). Additionally, the Central Asian A. tians-
one accession of the Spanish A. nova subsp. iberica (MA022), chanica (RK007) appears in this group. Several southern
one accession of the endangered Cypriote endemic A. kenne- Anatolian accessions of A. deflexa (RK088, RK089, RK250,
dyae (RK215), and four accessions of A. montbretiana from and RK251) and A. ionocalyx (RK117, RK118, RK252, and
Western and Central Asia (A32MO, A59M, RK043, and RK121). RK255) plus the westernmost accessions of A. aubrietioides
Two of these accessions (A32MO and A59M) were originally (RK240) are combined together. A group containing all
wrongly determined as A. auriculata and A. nova, respectively. accession of A. cypria is amended by two accessions of
The remaining accessions of A. auriculata (western Europe to A. aubrietioides (RK241, RK243), both being sampled on
central Asia) and A. nova subsp. nova form a discrete clade the Turkish mainland opposite of the Kyrenia mountain
(clade B) not related to other annuals within clade A. Finally, range. Finally, species-specific subgroups are formed by the
the perennial subclade of clade A comprises the various A. alpina Cypriote endemic A. purpurea and by the annual A. mont-
accessions and six other perennial species: Arabis aubrietioides bretiana. The accessions of A. nova subsp. iberica (MA022)
(A01M, MG003, RK240, RK241, RK243, RK244, RK261), A. and A. kennedyae (RK215) are not directly connected to
deflexa (RK088/89, RK248–251), A. ionocalyx (RK117/18, A. montbretiana.
RK252–255), A. tianschanica (RK007), A. cypria (RK086/
87, RK167, RK245/46), and A. purpurea (RK004, RK006, Combined data set— The alignment of the combined data set
A53M, RK247, RK257–260). Arabis cypria and A. purpurea comprised 66 sequences (representing 70 accessions) and 1390
are characterized by discrete ITS sequences with the exception characters, of which 1066 were constant. Of the 324 variable
of one accession of A. cypria (RK246), which shared its ITS- characters, 212 were phylogenetically informative and the re-
type with A. purpurea. Most accessions of the remaining taxa maining 112 were autapomorphic positions (Appendix S3). The
(Arabis aubrietioides, A. deflexa, A. ionocalyx, and A. tians- SYM+Γ evolutionary model was applied to the nrDNA parti-
chanica) had an ITS sequence identical to the widespread ITS- tion, and the K81uf+Γ evolutionary model was applied to the
type A of A. alpina (Koch et al., 2006). cpDNA partition in the respective ML and Bayesian analyses.
The result of the Bayesian analysis is shown in Fig. 4. The re-
Chloroplast trnLF region— The combined trnLF align- sulting trees had a harmonic mean for the ln likelihood of
ment comprised 61 unique sequences (representing 72 acces- −5422.36 and an average standard deviation of split frequencies
sions) and 759 characters, 617 of which were constant. Of of 0.011. The corresponding tree of the ML analysis had a ln
the 142 variable characters, 75 were phylogenetically infor- likelihood of −5104.81. The maximum parsimony analysis re-
mative, and the remaining 67 were autapomorphic positions sulted in 6134 most parsimonious trees with a tree length of 534
(Appendix S2). (with “maximum length = 0” rule applied), CI of 0.68 (autapo-
The K81uf+Γ evolutionary model was applied to the ML and morphies excluded), and RI of 0.89.
Bayesian analyses, the results of which are shown in Fig. 2. The The phylogenetic tree based on the combined data set (Fig. 4)
resulting Bayesian trees had a harmonic mean for the ln likeli- shows the same seven major clades as recovered by the ITS
hood of −2581.62 and an average standard deviation of split analysis (Fig. 1), reflecting the high amount of parsimony infor-
frequencies of 0.011. The corresponding tree of the ML analy- mative characters in the ITS data set. All of the clades were
sis had a ln likelihood of −2287.00. The maximum parsimony very strongly supported by their posterior probabilities (ppr =
analysis resulted in 10 000 most parsimonious trees with a tree- 1.0) and bootstrap values (bootstrap > 98.5%). Clade δ (Draba
length of 195 (with “maximum length = 0” rule applied) and an and its segregates) and clade ε (Aubrieta and Arabis verna)
ensemble CI of 0.77 (autapomorphies excluded) and an ensem- remained unsupported.
ble RI of 0.90. Similar to the ITS tree (Fig. 1), the A. alpina containing
The chloroplast trnLF Bayesian phylogenetic tree (Fig. 2) clade (clade α) is subdivided into statistically supported annual
shows virtually the same seven major clades as the ITS tree and perennial subclades. The perennial subclade comprises
(Fig. 1). Among these are the following: clade p (A. aucheri and several subgroups recovered by both ITS and the trnLF analy-
A. parvula), clade m (core Arabis), clade d (Draba and its seg- ses (Figs. 1, 2).
regates), and clade e (Aubrieta and Arabis verna). The latter is
sister to clade d and has high posterior probability support (ppr = Congruency test— The best tree topology was defined for
0.99), but only limited bootstrap support (maximum-likelihood each of the three data sets (ITS, trnLF, and combined) accord-
bsML = 58.4, and maximum-parsimony, bsMP = 45.6). Arabis ing to the –lnL scores (ITS, trnLF, and 100 random topologies).
auriculata (clade b1) and A. nova subsp. nova (clade b2) are not This was either the ITS topology (for the ITS and combined
grouped together as in the ITS analyses, but form discrete data set) or the trnLF topology (for the trnLF data set). For all
clades. Clade l (A. nordmanniana group) forms a well-supported other topologies including all random topologies, the associated
clade sister to the clade comprising A. alpina and its closest p-value was used as a measure for the probability that this tree
related species. This clade was not subdivided into the annual topology can also explain the data. All random tree topologies
April 2012] KARL ET AL.—EVOLUTIONARY HISTORY OF ARABIS ALPINA 785

Fig. 3. (A) Network of 73 unique cpDNA halpotypes, and B) regional distribution of haplotypes in the eastern Mediterranean. Shapes and colors cor-
respond in both illustrations. The A. nordmanniana group (colorless circles) was used to root the network (see asterisk). Previously published A. alpina
haplotypes are represented by filled circles; newly described haplotypes are represented either by triangles (for A. tianschanica, A. cypria, and A. purpurea),
empty circles (for A. aubrietioides, A. deflexa, and A. ionocalyx) or gray diamonds (for the annual taxa A. montbretiana, A. nova subsp. Iberica, and A.
kennedyae). (B) Some accessions are further specified by an A (for A. aubrietioides), D (for A. deflexa) or I (for A. ionocalyx), if they belonged to one of
these species. (C) shows the global distribution of Arabis alpina’s three major cpDNA trnLF haplotype groups. The gray-shaded area in central Turkey
contains members of all three haplotype groups. The locality sampled for A. tianschanica is indicated with a dark green triangle.
786 AMERICAN JOURNAL OF BOTANY [Vol. 99

had a p-value of 0.00, rejecting the hypothesis that any of to other haplotypes of A. alpina. One accession of A. deflexa
them explain the data equally well. Conversely, the trnLF to- (RK248, haplotype 77) had a haplotype derived from central
pology could explain the ITS (p = 0.22) and combined data blue haplotype 01, while two accessions of A. aubrietioides
set (p = 0.50), and the ITS topology could explain the trnLF (RK241 and RK243) had haplotype 81, which is derived from
data set (p = 0.30), as these values were above the threshold A. alpina haplotype 45 via two mutational steps. The annual
of p = 0.05. species A. montbretiana (haplotypes 86, 87) and A. nova subsp.
iberica (haplotypes 85 and 88) were also connected to haplo-
Chloroplast DNA network— The modified trnLF alignment type 01, whereas related A. nova subsp. iberica held an iso-
for phylogeographic analysis resulted in 666 characters (660 lated position in the network (Fig. 3). The three haplotypes
regular bases, plus six positions used for gap-coding), from recovered from A. montbretiana were connected to haplotypes
which 73 unique haplotypes were discriminated among the 518 20 and 03 of A. alpina, although an alternative mutational
accessions of A. alpina and close relatives. The resulting mini- pathway linking to haplotype 41 was also proposed by the
mum spanning network is shown in Fig. 3. TCS program. Both of these pathways seem unrealistic based
The network was almost identical to those previously pre- on the clear morphological and biogeographical differences
sented based solely on accessions of A. alpina (e.g., Koch between the annual A. montbretiana and the perennial species
et al., 2006; Assefa et al., 2007; Ansell et al., 2011), especially of the A. alpina group. Finally, the Cypriote A. kennedyae
in recovering the three major cpDNA haplotype groups (blue, (haplotype 89) was connected with three mutational steps to
red, green). The two haplotypes of the A. nordmanniana group A. montbretiana haplotype 86.
(haplotypes 90, 91) were used to root the network; these types
were connected through six mutational steps to central haplo- Divergence-time estimates— The results of the divergence-
type 01 (blue). Four of the five newly included A. alpina ac- time calculation are given in Table 1 (ITS) and Table 2 (trnLF).
cessions (MA001, MA004, RK001, and RK002) had the The results varied greatly among the three different ap-
common blue haplotype (haplotype 01), and one new blue proaches, but two main conclusions could be drawn. (1) Esti-
haplotype (haplotype 72) was identified from the western Ana- mates calculated from the trnLF marker normally exceeded
tolian accession RK001b. Two new haplotypes (83, 84) were those from the ITS marker by a factor of two. When the rates
added to the base of the red haplotype group. Haplotype 84 of variation were fixed, the estimates were congruent. Fixing
from A. aubrietioides (accession A01M) formed a side branch these rates in a reasonable interval seems important, because
to the red haplotypes 62 and 63 of A. alpina. The A. ionocalyx the calculations run without strict limits on the rates of varia-
accession from Syria (RK254, haplotype 83) formed another tion showed nonconstant rates that dropped below all reported
side branch within the red haplotype group and was closely values for this chloroplast marker (see Table 2). Therefore, the
connected to the most basal haploytype 60 of A. alpina from corresponding divergence time values might be overestimated.
northern Lebanon. Only one new haplotype was added to the (2) The calculated estimates using the Thlaspideae fossil as
green haplotype group: the accession of A. tianschanica the primary calibration point according to Beilstein et al. (2010)
(RK007, haplotype 82) from Kyrgyzstan, which occupied a tip were more than three times higher than those calculated with
position in the green haplogroup (Fig. 3). In addition, several the secondary calibration point published by Couvreur et al.
of the remaining relatives had green haplotypes that were (2009). Although confidence intervals are large, the overlap
identical to those recovered from A. alpina: A. aubrietioides between fossil based calibration and secondary calibration is
(MG003) had haplotype 04, A. deflexa (RK249) had haplotype small. Notably, the log likelihood values remain relatively
47, while accessions of A. aubrietioides (RK244, RK261) and constant for all dating approaches per marker, and therefore
A. ionocalyx (RK253) held haplotype 51. All remaining acces- the calculated Bayes factors do not favor one approach over
sions of A. alpina relatives were affiliated with the blue haplo- another.
type group of Europe. The Cypriote species A. purpurea
(haplotype 73) and A. cypria (haplotypes 75 and 76) along Chromosome numbers and ploidy levels— Known chromo-
with one accession of A. purpurea (described as A. cypria some numbers and ploidy levels are summarized in Table 3.
[haplotype 74] formed a distinct “purpurea haplotype group”, Here we present counts for A. purpurea (accessions RK004,
restricted to the Kyrenia mountains in Cyprus (Fig. 3). Several RK006) and A. cypria (accession RK167) for the first time;
accessions of the A. alpina relatives from the Taurus Moun- both are 2n = 16. A hybrid plant of unknown origin (not in-
tains shared haplotype 41, which is also common in accessions cluded in the analyses) between A. alpina and A. aubrietioides
of A. alpina from this region (A. deflexa [RK088, RK250], A. was also counted as 2n = 16, indicating a hybridization event at
ionocalyx [RK117, RK252, RK255] and A. aubrietioides the diploid level. According to the data available, there are no
[RK240]). Other accessions of the relatives from the Taurus tetraploid or higher polyploid counts for any species closely re-
region also have haplotypes that are closely related to haplo- lated to A. alpina. Although there are some species (A. deflexa,
type 41 (A. deflexa [RK089/haplotype 78 and RK251/haplotype A. tianschanica) whose chromosome numbers have not yet
80] and A. ionocalyx [RK118/haplotype 79]). Other acces- been determined. Hence it seems likely the “A. alpina clade”
sions of the A. alpina relatives were connected genologically consists exclusively of diploids.


Fig. 4. Phylogenetic tree of the Bayesian analysis based on the combined data set (ITS plus trnLF). Posterior probability values (first-mentioned) and
Bootstrap values of the maximum-likelihood analysis (middle-mentioned) and the maximum-parsimony analysis (last-mentioned) of the nodes are given
along the corresponding branches. Pseudoturritis turrita was used as the outgroup taxa. Accessions with a prefixed asterisk had the same sequence than the
top-listed accession and were therefore excluded from the tree calculation. Purple- or pink-flowered species are indicated with a black spot following the
taxon label; species that occasionally produce pink petals are marked with a gray spot. Accessions without any spot correspond to primarily white-flowered
species. Annual/winter-annual taxa are labeled with the corresponding sign for annualism, also following the taxon label.
April 2012] KARL ET AL.—EVOLUTIONARY HISTORY OF ARABIS ALPINA 787
788 AMERICAN JOURNAL OF BOTANY [Vol. 99

TABLE 2. Parameters and results of the three divergence-time estimate approaches based on the cpDNA trnLF marker. Mean values are indicated in
boldface; lower and upper bounds of the 95% highest posterior density (HPD) interval are given above and below that value, respectively. Enforced
bounds are indicated in italics. Bayes-Factors with positive evidence against M0 (i.e., both trees explain the data equally well) are marked with an
asterisk.

Fossil based calibration of node age of Using secondary calibration of node age of Like previous + fixed mutational
Parameter tribe Thlaspideae (acc. Beilstein et al., 2010) tribe Arabideae (acc. Couvreur et al., 2009) rates (Mummenhoff et al., 2004)

Runs / Generations 4 independent runs with 4 independent runs with 4 independent runs with
80 million generations each 20 million generations each 25 million generations each
Log likelihood −2604.064 −2603.344 −2602.146
−2588.331 −2588.106 −2586.818
−2574.108 −2573.179 −2471.985
tmrca of Arabis alpina agg. and annual 4.32 1.88 1.01
sister group: clade a (Fig. 2) (Ma) 16.8 4.82 2.23
34.2 8.61 3.84
tmrca of core-Arabis (Ma) 3.99 1.65 0.53
19.3 5.50 2.42
40.0 10.2 4.75
tmrca of Draba and its segregates (Ma) 8.86 3.75 1.22
31.4 8.94 4.17
61.6 15.5 7.66
tmrca of tribe Arabideae (Ma) 16.8 10.0 mya 10.0
48.8 14.1 10.8
92.2 21.5 12.7
ucld.mean 0.15 × 10−9 0.75 × 10−9 3.60 × 10−9
0.48 × 10−9 1.57 × 10−9 3.97 × 10−9
0.87 × 10−9 2.44 ×10−9 4.77 × 10−9
meanRate 0.16 × 10−9 0.84 × 10−9 2.41 × 10−9
0.49 × 10−9 1.63 × 10−9 3.15 × 10−9
0.86 × 10−9 2.43 × 10−9 3.91 × 10−9
Bayes factors — / 0.93 / 0.24 1.08 / — / 0.26 4.10* / 3.81* / —

DISCUSSION it appears that the cpDNA results are more biologically meaning-
ful, as all four species are confined to the Caucasus Mountains,
Congruency between nuclear and cpDNA derived phyloge- which are themselves near to the center of origin and diversity
nies and divergence-time estimates— In this study, which fo- for A. alpina (Ansell et al., 2011).
cuses on Arabis alpina and its closest relatives, the combined The phylogenetic position of the genus Aubrieta (including
tree and the derived phylogenetic hypothesis (Fig. 4) help to Arabis verna) formed a second incongruency between the two
resolve phylogentic relationships within the tribe Arabideae markers. For the trnLF -based analyses, Aubrieta was placed
and identify new divergent lineages (Figs. 4, 5). We show that close to Draba (Fig. 2). This relationship was not recognized by
Arabis verna is more closely related to the genus Aubrieta than ITS based analyses (Fig. 1). The low resolution of the ITS-based
to genus Arabis (Figs. 2, 4) and that it forms a monophyletic analysis has previously been shown with two large-scale data
clade (clade ε in Fig. 4), despite the different markers creating sets focusing on Draba and core-Arabis (Jordon-Thaden et al.,
uncertainty in the phylogenetically position of Aubrieta (e.g., 2010; Koch et al., 2010). Extensive hybridization and reticulate
sister to Arabis alpina: Koch et al., 1999a, 2001; sister to core evolution within these groups has resulted in chloroplast cap-
Arabis: Koch et al., 2000; sister to all Arabideae s.s.: German ture and incomplete lineage sorting of both nuclear and plastid
et al., 2009). Although the phylogenetic relationship of A. verna markers. So far we have found no evidence for these complexi-
and Aubrieta is based on limited species sampling for Aubrieta ties in the Aubrieta-Arabis verna and Arabis brachycarpa-A.
(n = 3), we assume that this pattern is significant. Future work nordmanniana-A. christiani-A. graellsiiformis clades. Instead,
will test whether Arabis verna is nested within the genus Aubrieta, it seems more likely that the mode of marker molecular evolu-
which consists of ca. 15 species or whether it is a sister taxon to tion has contributed to the phylogenetic incongruencies; con-
the genus Aubrieta. certed evolution for nuclear ITS regions (Koch et al., 2003) and
The congruency test revealed no major incongruencies between the low mutation rates for plastidic genomes (trnLF sequences).
both DNA markers. However, several minor differences exist Consequently, experimental work adding single-copy nuclear
between the nrDNA ITS (Fig. 1) and the cpDNA trnLF (Fig. 2) genes is needed to further resolve these phylogenetic relation-
sequence analyses, and these should be mentioned. On the basis ships at the tribal level (e.g., chs, adh; M. A. Koch and R. Karl,
of the cpDNA analysis, four Arabis species (A. brachycarpa, A. unpublished manuscript).
nordmanniana, A. graellsiiformis, and A. christiani) are sister Our various estimates of divergence times varied greatly de-
to A. alpina and its close relatives (Fig. 2). However, this relation- pending on the approaches applied, making it difficult to deter-
ship was not supported by the ITS analyses, which placed all mine the most reliable methodology. The primary fossil-based
four species close to Draba. This difference probably reflects calibrations following the approach of Beilstein et al. (2010)
the variation in the mode of marker inheritance with the plastidic seem to have resulted in overly high divergence time values.
trnLF being maternally inherited (Harris and Ingram, 1991) and These estimates (ITS: 7.1 Ma with a 95% CI from 2.8–12.3,
the nuclear-encoded ITS being biparentally inherited. Overall, trnLF: 16.8 Ma with a 95% CI from 4.3–34.2) would scale most
April 2012] KARL ET AL.—EVOLUTIONARY HISTORY OF ARABIS ALPINA 789

TABLE 3. Compilation of published chromosome numbers and ploidy The closest relatives of Arabis alpina: Parallel evolution of dis-
levels of the taxa included in the study. tinct morphological characters—This study identified Arabis
alpina as a paraphyletic taxon, because six Arabis species (A.
Taxon 2n Ploidy level Reference
aubrietoides, A. cypria, A. deflexa, A. ionocalyx, A. purpurea, and
Arabis alpina 16 2x # A. tianschanica, Figs. 2–5) are nested within the widely distributed
Arabis aubrietioides 16 2x this study A. alpina. All six species are perennials with small distribution
Arabis aucheri 16 2x # areas: Kyrgyzstan in the case of A. tianschanica and the eastern
Arabis auriculata 16 2x # Mediterranean for the remaining five species. The eastern Mediter-
Arabis blepharophylla 16 2x #
Arabis brachycarpa 16 2x # ranean, more precisely Anatolia and the adjacent Levantine region,
Arabis christiani 32 4x # has been shown to be the center of origin and genetic diversity of
Arabis cypria 16 (14*) 2x Hand (2006); this study A. alpina (Ansell et al., 2011). As mentioned, we assume Pleisto-
Arabis deflexa n.d. n.d. cene radiation events within A. alpina. The mountainous areas of
Arabis erikii n.d. n.d. Anatolia in comparison to the European Alps and Fennoscandia,
Arabis graellsiiformis n.d. n.d. are very diverse in terms of altitude range and geology and were
Arabis hirsuta 32 4x #
Arabis ionocalyx 16 2x # never covered by a continuous ice-shield during the Pleistocene
Arabis kennedyae n.d. n.d. glaciation periods. Only the higher mountain peaks were glaciated
Arabis montbretiana 16 2x Hoffmann et al. (2010) (Erinç, 1978; Atalay, 1996), and the vegetation belts shifted by up
Arabis nordmanniana 16 2x # to 500 m of elevation (Parolly et al., 2010). The existence of several
Arabis nova subsp. iberica n.d. n.d. small-scale refugia in this area (Médail and Diadema, 2009;
Arabis nova subsp. nova 16 2x # Parolly et al., 2010) allowed for the survival of various biota during
Arabis parvula 32 4x #
Arabis purpurea 16 2x #; this study the periods of temperature fluctations. In the warmer periods, a dry
Arabis sachokiana 16 2x # steppe-like climate prevailed in the area (Webb and Bartlein, 1992),
Arabis tianschanica n.d. n.d. and higher elevations would have provided suitable habitats for
Arabis verna 16/32 2x/4x # temperate biota due to higher precipitations and more moderate
Aubrieta canescens subsp. n.d. n.d. temperatures. For cold-tolerant biota, these microrefugia would be
macrostyla further limited to the high alpine habitats of each mountain system,
Aubrieta deltoidea 16 2x #
Arabis parviflora 16 2x #
thereby creating gene flow limitations and promoting local lin-
Draba hederifolia 30 4x # eages accumulation. Ansell et al. (2011) emphasized the roles of
Draba hispanica 16 4x # local survival centers in the coastal Taurus Mountains, which pro-
Draba reptans 30/32 4x # vided moister conditions compared to the drier interior climates
# Warwick and Al-Shehbaz (2006); * Yildiz and Gücel (2006); n.d.: no of the Anatolian (Davis, 1971; Ekim and Güner, 1986). Accord-
data available ingly, the Turkish Arabis species with close relationships to A.
alpina occur mainly within the distinct mountain ranges of the
Taurus systems: the red-flowering A. aubrietioides in the Central
Taurus (Cilician sector) and Arabis deflexa and A. ionocalyx in the
of the radiation of A. alpina to the late Miocene/Pliocene, a western parts (Lycian and Pisidian-Isaurian sectors) (Davis, 1965).
period with relatively stable, predominantly warm and moist A very similar spatial distribution of taxa from Heldreichia Boiss.
climate (Dowsett et al., 2005; Salzmann et al., 2009), which is has been recently reported along the Taurus systems (Parolly et al.,
not in agreement with the biology of the various species. Fur- 2010). Although there are clear morphological characters that
thermore, the uplift of mountain chains during the alpide orog- differentiate between the Taurus Arabis and A. alpina, so far it is
eny was still in progress at that time (Atalay, 2002). The Taurus not possible to discriminate between them using the molecular
and the adjacent East Anatolian Plateau are considered to have markers employed by our investigation. Rather, A. aubrietioides,
arisen in the late Miocene at the earliest or more likely origi- A. deflexa, and A. ionocalyx often bear local A. alpina haplotypes,
nating in the late Pliocene (Güldali, 1979; Franzke et al., or closely related ones (see Fig. 3A, B). Thus, for the mainland
2011). Consequently, the most likely divergence time esti- species of A. aubrietioides, A. deflexa, and A. ionocalyx, the
mates might be close to the lower boundaries of the primary geographical footprint of A. alpina genotypes is stronger than
fossil-based calibration method or the upper boundaries of the the morphological differentiation.
secondary calibration method (Couvreur et al., 2009). Under Similar geographical separation of species flock members can
these approaches, the split between Arabis alpina and its an- be seen on Cyprus, where endemites A. purpurea and A. cypria
nual sister group would be dated between 2.84–3.98 (for ITS are restricted to the Troodos and Kyrenia mountain ranges, re-
data) or between 3.84–4.32 (–8.6 Ma) (for trnLF data), respec- spectively. In contrast to the mainland A. alpina relatives, these
tively. We recognize that the Bayes factors are insufficiently island relatives are genetically distinct from A. alpina in bear-
high to support or reject either of the approaches. However, ing private chloroplast and ITS haplotypes. Together, these ob-
the divergence times calculated under the secondary calibra- servations suggest that A. purpurea and A. cypria evolved
tion approach correspond well to a previous cpDNA-based independently in disjunct mountain systems of Cyprus in the
coalescence estimate of 0.80–6.43 Ma (mean of 2.69 Ma) for local absence of A. alpina populations. Hadjisterkotis and Bider
the most distantly related Arabis alpina cpDNA haplotypes (1997) observed that many Cypriote species are closely related
(Ansell et al., 2011). Furthermore, summary data from four to mainland taxa, despite Cyprus never being directly connected
DNA markers indicates a split between the African and Eur- to the mainland during its geological history (Robertson, 1990;
asian Arabis alpina of 0.28–1.39 Ma (mean of 0.70 Ma) (Koch Hadjisterkotis, 2007). Hence, the modest distance to southern
et al., 2006). These findings place the subsequent radiation Anatolia (ca. 68 km) has probably allowed for rare long-distance
events within the Arabis alpina clade into the Late Pliocene/ dispersal, but represents a sufficient barrier to have promoted
early Pleistocene. allopatric speciation.
790 AMERICAN JOURNAL OF BOTANY [Vol. 99

Within the red-flowering species, there is some interesting the habitat fragmentation during Pleistocene glaciations, and
vicariance of A. purpurea on Cyprus and A. aubrietoides on the possibly reinforced by breeding system shift.
Turkish mainland. We speculate that the red flowers may be In contrast to most members of the Arabis alpina species
indicative of outbreeding. Because the phylogenetic trees do flock, A. tianschanica is located in Uzbekistan and western
not support a shared origin of the Cypriote and Turkish main- Kyrgyzstan, several hundred kilometers east of the nearest lo-
land populations within the predominantly white-flowering calities for A. alpina in northeast Iran. German et al. (2009)
Arabis alpina, this phenological change may have occurred in have already suggested a close relationship between this en-
parallel. This observation is supported also by the diploid chro- demic and A. alpina based on morphological characters. Geneti-
mosome numbers of these taxa (Table 3), indicating an absence cally, A. tianschanica has a private haplotype that is derived
of hybridization or polyploidy, in contrast to the other major from the common green haplotype 04, which suggests this taxa
clades in the Arabideae for which various ploidy level series might have evolved from a remnant population of a former
indicate extensive hybridization and reticulation (Jordon-Thaden more easternly distributed A. alpina.
and Koch, 2008; Koch et al., 2010). Further ecological and
molecular research may unravel new and interesting aspects of Southern refuges as centers of origin and evolutionary
parallel evolution among sister-species pairs that have been starting points for different life-history strategies— The occur-
derived from white-flowering Arabis alpina. More widely rence of species-poor annual clades being sister to species-rich
red-flowering taxa within the genus Arabis are typically not perennial clades is a common feature within the tribe Arabideae.
sympatrically distributed with other red-flowering species: e.g., This trend is also found in many other tribes and genera of the
in Europe the red-flowering endemic of Arabis stenocarpa Brassicaceae (e.g.: Arabidopsis/tribe Sisymbrieae [Koch and
Boiss. & Reut. from Spain, Arabis cretica Boiss. & Heldr. from Matschinger 2007]; Noccaea, Microthlaspi/tribe Coluteocarpeae
Crete (Koch et al., 2010), and Arabis rosea DC. from southern [Mummenhoff et al., 1997a, b; Koch et al., 1998]; Cochlearia,
Italy. Furthermore, all European red-flowering Arabis species Ionopsidium/tribe Cochlearieae, [Koch et al., 1999b]).
with the exception of Arabis alpina have failed to developed Notably, this is pattern is also found in Draba (Arabideae),
red-flowering sister species pairs (M. A. Koch and R. Karl, where we found that the ca. 20 annual species are restricted to a
unpublished data). These multiple changes in flower color single clade that is sister to the clade comprising the remaining
might be indicative of genetic/geographic isolation caused by 300 perennial taxa, and which underwent a dramatic Pleistocene

Fig. 5. Diagram of systematic relationships within the tribe Arabideae based on data summarized from various sources (German et al., 2009; German
and Al-Shehbaz, 2008; Koch et al., 1999a, 2000, 2001, 2007, 2010; Jordon-Thaden et al., 2010; Warwick et al., 2010; Al-Shehbaz et al., 2011) and comple-
mented with the data presented in this study. l.s. = low significance.
April 2012] KARL ET AL.—EVOLUTIONARY HISTORY OF ARABIS ALPINA 791

radiation (Jordon-Thaden and Koch, 2008; Jordon-Thaden et al., mountains as the cradle of global diversity in Arabis alpina, a key
2010). This pattern is also true for the Arabis alpina species arctic-alpine species. Annals of Botany 108: 241–252.
group, although less pronounced, whereby the seven perennial ASSEFA, A., D. EHRICH, P. TABERLET, S. NEMOMISSA, AND C. BROCHMANN.
species are sistered by three annual taxa: Arabis montbretiana, 2007. Pleistocene colonization of afro-alpine ‘sky islands’ by the
arctic-alpine Arabis alpina. Heredity 99: 133–142.
A. nova subsp. Iberica, and A. kennedyae (Fig. 4). The five re-
ATALAY, I. 1996. Palaeosoils as indicators of the climatic changes during
maining annual species included in this study (Arabis aucheri, Quaternary period in S. Anatolia. Journal of Arid Environments 32:
Arabis auriculata, Arabis erikii, Arabis nova subsp. nova, Ara- 23–35.
bis parvula) show no close relationships to any other perennial ATALAY, I. 2002. Mountain ecosystems of Turkey. In M. F. Buchroithner [ed.],
Arabis species but form two distinct clades (Fig. 4). Future phy- Kartographische Bausteine, vol. 28, 29–38. Institute for Cartography,
logenetic research may help clarify their systematic and life his- Dresden University of Technology, Dresden, Germany.
tory relationships. BEILSTEIN, M. A., N. S. NAGALINGUM, M. D. CLEMENTS, S. R. MANCHESTER,
The evolutionary success of the genus Arabis as a whole may AND S. MATHEWS. 2010. Dated molecular phylogenies indicate a
be attributed to a combination of adaptive characters: (1) High- Miocene origin for Arabidopsis thaliana. Proceedings of the National
elevation plants that are outcrossing, long-lived perennials and Academy of Sciences, USA 107: 18724–18728.
produce large flowers to attract insects and winged seeds suit- BIRKS, H. H. 2008. The late-quaternary history of arctic and alpine plants.
able for long-distance dispersal. (2) Plants, mostly from mon- Plant Ecology & Diversity 1: 135–146.
tane regions, produce medium-sized flowers and narrowly BUSCH, N. A. 1906. Sistematika i botaniceskaja geografija kavkazskich
vidov roda Arabis L. osobjenno sekzii Alliariopsis m. Moniteur du
winged seeds and tend to shift from perennial to biennial growth
Jardin Botanique de Tiflis 6: 3–23.
forms. (3) Low-elevation plants are primarily selfing annuals CLEMENT, M., D. POSADA, AND K. A. CRANDALL. 2000. tcs: A computer program
with small flowers and small seeds (Koch et al., 2010). Conse- to estimate gene genealogies. Molecular Ecology 9: 1657–1659.
quently, we find annuals primarily in lowland regions and with COMES, H. P., AND J. W. KADEREIT. 1998. The effect of Quaternary cli-
more southern distribution areas than the present-day distribu- matic changes on plant distribution and evolution. Trends in Plant
tion areas of the perennials. The generally higher number of Science 3: 432–438.
species in the perennial groups reflect their alpine and montane COMES, H. P., AND J. W. KADEREIT. 2003. Spatial and temporal patterns in
habitats, creating opportunities for lineage diversification during the evolution of the flora of the European alpine system. Taxon 52:
the Pleistocene environmental fluctuations. These features have 451–462.
recently been shown to have accelerated speciation and poly- COUVREUR, T. L. P., A. FRANZKE, A. I. AL-SHEHBAZ, F. T. BAKKER, M. A.
ploidization rates in Draba and core Arabis (Jordon-Thaden KOCH, AND K. MUMMENHOFF. 2009. Molecular phylogenetics, tempo-
and Koch, 2008; Koch et al., 2010). ral diversification and principles of evolution in the mustard family
(Brassicaceae). Molecular Biology and Evolution 27: 55–71.
Conclusion and taxonomic remarks— The phylogenetic DAVIS, P. H. 1965. Flora of Turkey and the East Aegean islands, vol. 1,
422–429. Edinburg University Press, Edinburgh, UK.
results presented revealed that Arabis alpina in its current DAVIS, P. H. 1971. Distribution patterns in Anatolia with particular refer-
limits is a paraphyletic species and the origin of several evo- ence to endemism. In P. H. Davis, P. C. Harper, and I. C. Hedge [eds.],
lutionary lineages located in Anatolian mountain regions and Plant life of south-west Asia, 15–28. Botanical Society of Edinburgh,
the eastern Mediterranean. Whereas the four mainland spe- Edinburgh, UK.
cies (A. aubrietioides, A. deflexa, A. ionocalyx, and A. tians- DOBEŠ, C., T. MITCHELL-OLDS, AND M. KOCH. 2004. Extensive chloroplast
chanica) are completely nesting within A. alpina, the two haplotype variation indicates Pleistocene hybridization and radiation
Cypriote endemics (A. cypria and A. purpurea) are both geo- of North American Arabis drummondii, A. × divaricarpa, and A. hol-
graphically and geneologically, much more distinct. Pres- boellii (Brassicaceae). Molecular Ecology 13: 349–370.
ently, there is no evidence of a common evolutionary history DOWSETT, H. J., M. A. CHANDLER, T. M. CRONIN, AND G. S. DWYER. 2005.
for these two Cypriote taxa, suggesting that the shift from Middle Pliocene sea surface temperature variability. Paleoceanography
white to red flowers has evolved independently more than 20: PA2014.
once. Our results imply the need for a number of taxonomic DOYLE, J. J., AND J. L. DOYLE. 1987. A rapid DNA isolation procedure for
small amounts of fresh leaf tissue. Phytochemical Bulletin 19: 11–15.
changes within Arabis alpina and its closest relatives. The
DRUMMOND, A. J., AND A. RAMBAUT. 2007. BEAST: Bayesian evolution-
recognition of the perennial relatives as subspecies of A. alpina ary analysis by sampling trees. BMC Evolutionary Biology 7: 214.
would fit the observed pattern of genetic distribution and re- doi:10.1186/1471-2148-7-214.
store monophyly for A. alpina. Furthermore, as A. alpina is EHRICH, D., M. GAUDEUL, A. ASSEFA, M. KOCH, K. MUMMENHOFF, S.
the type species of the genus Arabis, major adjustments within NEMOMISSA, INTRABIODIV CONSORTIUM, AND C. BROCHMANN. 2007.
the whole tribe Arabideae seem to be necessary as well. How- Genetic consequences of Pleistocene range shifts: Contrast between the
ever, to make any profound suggestions in this respect requires Arctic, the Alps and the East African mountains. Molecular Ecology
additional data gathering and experimental work focusing on 16: 2542–2559.
the tribal phylogeny. EKIM, T., AND A. GÜNER. 1986. The Anatolian diagonal: Fact or fiction?
Proceedings of the Royal Society of Edinburgh 89B: 69–77.
LITERATURE CITED ERINÇ, S. 1978. Changes in the physical environment in Turkey since the
end of the last glacial. In W. C. Brice [ed.], The environmental history
AL-SHEHBAZ, I. A., D. A. GERMAN, R. KARL, I. JORDON-THADEN, AND M. of the Near and Middle East since the last Ice Age, 87–110. Academic
A. KOCH. 2011. Nomenclatural adjustments in the tribe Arabideae Press, London, UK.
(Brassicaceae). Plant Diversity and Evolution 129: 71–76. FELSENSTEIN, J. 1985. Confidence limits on phylogenies: An approach
ANSELL, S. W., M. GRUNDMANN, S. J. RUSSELL, H. SCHNEIDER, AND J. C. using the bootstrap. Evolution; International Journal of Organic
VOGEL. 2008. Genetic discontinuity, breeding system change and Evolution 39: 783–791.
population history of Arabis alpina in the Italian Peninsula and adja- FONT, M., N. GARCIA-JACAS, R. VILATERSANA, C. ROQUET, AND A. SUSANNA.
cent Alps. Molecular Ecology 17: 2245–2257. 2009. Evolution and biogeography of Centaurea section Acrocentron
ANSELL, S. W., H. K. STENOIEN, M. GRUNDMANN, S. J. RUSSEL, M. A. KOCH, inferred from nuclear and plastid DNA sequence analyses. Annals of
H. SCHNEIDER, AND J. C. VOGEL. 2011. The importance of Anatolian Botany 103: 985–997.
792 AMERICAN JOURNAL OF BOTANY [Vol. 99

FRANZKE, A., M. A. LYSAK, I. A. AL-SHEHBAZ, M. A. KOCH, AND K. transcribed spacer of nuclear ribosomal DNA (ITS) in North American
MUMMENHOFF. 2011. Cabbage family affairs: The evolutionary his- Arabis divaricarpa (Brassicaceae). Molecular Biology and Evolution
tory of Brassicaceae. Trends in Plant Science 16: 108–116. 20: 338–350.
GERMAN, D. A., AND I. A. AL-SHEHBAZ. 2008. Dendroarabis, a new Asian KOCH, M. A., C. DOBEŠ, C. KIEFER, R. SCHMICKL, L. KLIMES, AND M. A.
genus of Brassicaceae. Harvard Papers in Botany 13: 289–291. LYSAK. 2007. SuperNetwork identifies multiple events of plastid trnF
GERMAN, D. A., N. FRIESEN, B. NEUFFER, I. A. AL-SHEHBAZ, AND H. HURKA. (GAA) pseudogene evolution in the Brassicaceae. Molecular Biology
2009. Contribution to ITS phylogeny of the Brassicaceae, with spe- and Evolution 24: 63–73.
cial reference to some Asian taxa. Plant Systematics and Evolution KOCH, M., B. HAUBOLD, AND T. MITCHELL-OLDS. 2000. Comparative evo-
283: 33–36. lutionary analysis of chalcone synthase and alcohol dehydrogenase
GERNHARD, T. 2008. New analytic results for speciation times in neutral loci in Arabidopsis, Arabis and related genera. Molecular Biology and
models. Bulletin of Mathematical Biology 70: 1082–1097. Evolution 17: 1483–1498.
GÜLDALI, N. 1979. Geomorphologie der Türkei. Beihefte Tübinger Atlas KOCH, M., B. HAUBOLD, AND T. MITCHELL-OLDS. 2001. Molecular system-
des vorderen Orients A 4: 1–226. atics of the Cruciferae: Evidence from coding plastome matK and
HADJISTERKOTIS, E. 2007. Review of biodiversity research results from nuclear CHS sequences. American Journal of Botany 88: 534–544.
Cyprus that directly contribute to the sustainable use of biodiversity in KOCH, M. A., R. KARL, C. KIEFER, AND I. A. AL-SHEHBAZ. 2010. Colonizing the
Europe. University of Nicosia, Unit of Environmental Studies, Nicosia, American continent: systematics of the genus Arabis in North America
Cyprus. (Brassicaceae). American Journal of Botany 97: 1040–1057.
HADJISTERKOTIS, E., AND J. R. BIDER. 1997. Cyprus. In D. M. Shackleton KOCH, M., C. KIEFER, J. VOGEL, D. EHRICH, C. BROCHMANN, AND K.
[ed.] and the IUCN/SSC Caprinae Specialist Group, Wild sheep and MUMMENHOFF. 2006. Three times out of Asia Minor—The phylo-
goats and their relatives. Status Survey and Conservation Action Plan geography of Arabis alpina L. (Brassicaceae). Molecular Ecology 15:
for Caprinae, 89-92. International Union for Conservation of Nature, 825–839.
Gland, Switzerland and Cambridge, UK. KOCH, M. A., AND M. MATSCHINGER. 2007. Evolution and genetic differ-
HAND, R. 2006. Supplementary notes to the flora of Cyprus V. Willdenowia entiation among relatives of Arabidopsis thaliana. Proceedings of the
36: 781–809. National Academy of Sciences, USA 104: 6272–6277.
HARRIS, S. A., AND R. INGRAM. 1991. Chloroplast DNA and biosystematics: KOCH, M., K. MUMMENHOFF, AND H. HURKA. 1998. Molecular biogeogra-
The effects of intraspecific diversity and plastid transmission. Taxon phy and evolution of the Microthlaspi perfoliatum s.l. polyploid com-
40: 393–412. plex (Brassicaceae): Chloroplast DNA and nuclear ribosomal DNA
HEDBERG, O. 1962. Intercontinental crosses in Arabis alpina L. Caryologia restriction site variation. Canadian Journal of Botany 76: 382–396.
15: 253–260. KOCH, M., K. MUMMENHOFF, AND H. HURKA. 1999b. Molecular phyloge-
HEWITT, G. M. 1999. Post-glacial re-colonization of European biota. netics of Cochlearia L. and allied genera based on chloroplast trnL
Biological Journal of the Linnean Society. Linnean Society of London intron and nuclear ribosomal ITS DNA sequence analysis contra-
68: 87–112. dict traditional classification. Plant Systematics and Evolution 216:
HEWITT, G. M. 2000. The genetic legacy of Quaternary ice ages. Nature 207–230.
405: 907–913. KROPF, M. 2002. Vergleichende Biogeographie europäischer Gebirgspflanzen:
HEWITT, G. M. 2004. Genetic consequences of climatic oscillations in the Molekulare und morphometrische Untersuchungen an der montan-
Quaternary. Proceedings of the Royal Society of London. Series B. subalpinen Anthyllis montana L. (Fabaceae) und der hochalpinen
Biological Sciences 359: 183–195. Pritzelago alpina (L.) O.Kuntze (Brassicaceae). Cuvillier, Göttingen,
HICKERSON, M. J., B. C. CARSTENS, J. CAVENDER-BARES, K. A. CRANDALL, C. Germany.
H. GRAHAM, J. B. JOHNSON, L. RISSLER, ET AL. 2010. Phylogeography’s MANCHESTER, S. R., AND E. L. O’LEARY. 2010. Phylogenetic distribution
past, present, and future. Molecular Phylogenetics and Evolution 54: and identification of fin-winged fruits. Botanical Review 76: 1–82.
291–301. MANEL, S., B. N. PONCET, P. LEGENDRE, F. GUGERLI, AND R. HOLDEREGGER.
HOFFMANN, M. H., H. SCHMUTHS, C. KOCH, A. MEISTER, AND R. M. FRITSCH. 2010. Common factors drive adaptive genetic variation at different
2010. Comparative analysis of growth, genome size, chromosome scales in Arabis alpina. Molecular Ecology 19: 3824–3835.
numbers and phylogeny of Arabidopsis thaliana and three cooccur- MÉDAIL, F., AND K. DIADEMA. 2009. Glacial refugia influence plant di-
ring species of the Brassicaceae from Uzbekistan. Journal of Botany: versity in the Mediterranean basin. Journal of Biogeography 36:
Article ID 504613. doi:10.1155/2010/504613. 1333–1345.
HOLMGREN, P. K., N. H. HOLMGREN, AND L. C. BARNETT. 1990. Index her- MEUSEL, H., E. J. JÄGER, AND E. WEINERT. 1965. Vergleichende Chorologie
bariorum, part I, The herbaria of the world, 8th ed. Regnum Vegetabile der Zentraleuropäischen Flora. Fischer Verlag, Jena, Germany.
120: 1–693. MUMMENHOFF, K., A. FRANZKE, AND M. KOCH. 1997a. Molecular phyloge-
HUCK, S., B. BÜDEL, J. W. KADEREIT, AND C. PRINTZEN. 2009. Range-wide netics of Thlaspi s.l. (Brassicaceae) based on chloroplast DNA restric-
phylogeography of the European temperate-montane herbaceous tion site variation and sequences of the internal transcribed spacer of
plant Meum athamanticum Jacq.: Evidence for periglacial persis- nuclear ribosomal DNA. Canadian Journal of Botany 75: 469–482.
tence. Journal of Biogeography 36: 1588–1599. MUMMENHOFF, K., A. FRANZKE, AND M. KOCH. 1997b. Molecular data
JORDON-THADEN, I., I. HASE, I. A. AL-SHEHBAZ, AND M. A. KOCH. 2010. reveal convergence in fruit characters, traditionally used in the clas-
Molecular phylogeny and systematics of the genus Draba (Brassicaceae) sification of Thlaspi s.l. (Brassicaceae) - Evidence from ITS-DNA se-
and identification of its closest related genera. Molecular Phylogenetics quences. Botanical Journal of the Linnean Society 125: 183–199.
and Evolution 55: 524–540. MUMMENHOFF, K., P. LINDER, N. FRIESEN, J. L. BOWMAN, J. Y. LEE, AND A.
JORDON-THADEN, I., AND M. A. KOCH. 2008. Diversity patterns in the ge- FRANZKE. 2004. Molecular evidence for bicontinental hybridogenous ge-
nus Draba: A first global perspective. Plant Ecology and Diversity 1: nome constitution in Lepidium sensu stricto (Brassicaceae) species from
255–263. Australia and New Zealand. American Journal of Botany 91: 254–261.
KAY, K. M., J. B. WHITTALL, AND S. A. HODGES. 2006. A survey of nuclear NICHOLAS, K. B. 1997. GeneDoc: A tool for editing and annotating mul-
ribosomal internal transcribed spacer substitution rates across angio- tiple sequence alignments. Computer program and documentation
sperms: An approximate molecular clock with life history effects. distributed by the author, website http://www.nrbsc.org/gfx/genedoc/
BMC Evolutionary Biology 6: 36. doi:10.1186/1471-2148-6-36. index.html [accessed May 2009].
KOCH, M., J. BISHOP, AND T. MITCHELL-OLDS. 1999a. Molecular sys- NYLANDER, J. A. A., F. RONQUIST, J. P. HUELSENBECK, AND J. L. NIEVES-
tematics and evolution of Arabidopsis and Arabis. Plant Biology 1: ALDREY. 2004. Bayesian phylogenetic analysis of combined data.
529–537. Systematic Biologists 53: 47–67.
KOCH, M., C. DOBEŠ, AND T. MITCHELL-OLDS. 2003. Multiple hybrid for- NYLANDER, J. A. A., J. C. WILGENBUSCH, D. L. WARREN, AND D. L. SWOFFORD.
mation in natural populations: Concerted evolution of the internal 2007. AWTY (Are we there yet?): A system for graphical exploration
April 2012] KARL ET AL.—EVOLUTIONARY HISTORY OF ARABIS ALPINA 793

of MCMC convergence in Bayesian phylogenetics. Bioinformatics TABERLET, P., L. GIELLY, G. PAUTOU, AND J. BOUVET. 1991. Universal
(Oxford, England) 24: 581–583. primers for amplification of three non-coding regions of chloroplast
PAROLLY, G., B. NORDT, W. BLEEKER, AND K. MUMMENHOFF. 2010. DNA. Plant Molecular Biology 17: 1105–1109.
Heldreichia Boiss. (Brassicaceae) revisited: A morphological and TEDDER, A., S. W. ANSELL, X. LAO, J. C. VOGEL, AND B. K. MABLE. 2011.
molecular study. Taxon 59: 187–202. Sporophytic self-incompatibility genes and mating system variation
PETIT, R. J., I. AGUINAGALDE, J.-L. DE BEAULIEU, C. BITTKAU, S. BREWER, R. in Arabis alpina. Annals of Botany 108: 699–713.
CHEDDADI, R. ENNOS, ET AL. 2003. Glacial refugia: Hotspots but not TEMPLETON, A. R., K. A. CRANDALL, AND CH. F. SING. 1992. A cladis-
melting pots of genetic diversity. Science 300: 1563–1565. tic analysis of phenotypic associations with haplotypes inferred
PLANTHOLT, U. 1995. Molekulare Untersuchungen zur Arealgeschichte from restriction endonuclease mapping and DNA sequence data. III.
von Arabis alpina L. (Brassicaceae). Ph.D. dissertation, University of Cladogram estimation. Genetics 132: 619–663.
Osnabrück, Osnabrück, Germany. WANG, R., S. FARRONA, C. VINCENT, A. JOECKER, H. SCHOOF, F. TURCK, C.
PONCET, B. N., D. HERRMANN, F. GUGERLI, P. TABERLET, R. HOLDEREGGER, ALONSO-BLANCO, ET AL. 2009. PEP1 regulates perennial flowering in
L. GIELLY, D. RIOUX, ET AL. 2010. Tracking genes of ecological rel- Arabis alpina. Nature 459: 423–427.
evance using a genome scan in two independent regional population WARWICK, S. I., AND I. A. AL-SHEHBAZ. 2006. Brassicaceae: Chromosome
samples of Arabis alpina. Molecular Ecology 19: 2896–2907. number index and database on CD-rom. Plant Systematics and
POSADA, D., AND K. A. CRANDALL. 1998. Modeltest: Testing the Evolution 259: 237–248.
model of DNA substitution. Bioinformatics (Oxford, England) 14: WARWICK, S. I., K. MUMMENHOFF, C. A. SAUDER, M. A. KOCH, AND I. A.
817–818. AL-SHEHBAZ. 2010. Closing the gaps: Phylogenetic relationships in
RAMBAUT, A., AND A. J. DRUMMOND. 2007. Tracer v1.4. Computer pro- the Brassicaceae based on DNA sequence data of nuclear ribosomal
gram available from website http://beast.bio.ed.ac.uk/Tracer. ITS region. Plant Systematics and Evolution 285: 209–232.
ROBERTSON, A. H. F. 1990. Tectonic evolution of Cyprus. In E. Moores, J. WEBB, T., AND P. J. BARTLEIN. 1992. Global changes during the last 3 mil-
Malpas, and C. Xenophontas [eds.], Ophiolites and oceanic lithosphere, lion years: Climatic controls and biotic responses. Annual Review of
Proceedings of the International Symposium, Nicosia, Cyprus, 1987. Ecology and Systematics 23: 141–173.
Ministry of Agriculture and Natural Resources, Nicosia, Cyprus. WHITE, T. J., T. BURNS, S. LEE, AND J. TAYLOR. 1990. Amplification and
RONQUIST, F., AND J. P. HUELSENBECK. 2003. MrBayes 3: Bayesian phy- direct sequencing of fungal ribosomal RNA genes for phylogenetics.
logenetic inference under mixed models. Bioinformatics (Oxford, In M. A. Innis, D. H. Gelfand, J. J. Sninsky, and T. J. White [eds.],
England) 19: 1572–1574. PCR protocols: A guide to methods and applications. Academic Press,
SALZMANN, U., A. M. HAYWOOD, AND D. J. LUNT. 2009. The past is a San Diego, California, USA.
guide to the future? Comparing Middle Pliocene vegetation with pre- WILLIS, K. J., AND R. J. WHITTAKER. 2000. The refugial debate. Science
dicted biome distributions for the twenty-first century. Philosophical 287: 1406–1407.
Transactions of The Royal Society A 367: 189–204. WING, S. L. 1987. Eocene and Oligocene floras and vegetation of the
SCHMITT, T. 2009. Biogeographical and evolutionary importance of the Rocky Mountains. Annals of the Missouri Botanical Garden 74:
European high mountain systems. Frontiers in Zoology 6: 9. doi:10. 748–784.
1186/1742-9994-6-9. YANG, Z., AND B. RANNALA. 1997. Bayesian phylogenetic inference using
SCHULZ, O. E. 1936. Cruciferae. In A. Engler and K. Prantl [eds.], DNA sequences: A Markov chain Monte Carlo method. Molecular
Die natürlichen Pflanzenfamilien, vol. 17B. Verlag von Wilhelm Biology and Evolution 14: 717–724.
Engelmann, Leipzig, Germany. YILDIZ, K., AND S. GÜCEL. 2006. Chromosome numbers of 16 endemic
SHIMODAIRA, H., AND M. HASEGAWA. 1999. Multiple comparisons of log- plant taxa from northern Cyprus. Turkish Journal of Botany 30:
likelihoods with applications to phylogenetic inference. Molecular 181–192.
Biology and Evolution 16: 1114–1116. YULE, G. U. 1924. A mathematical theory of evolution based on the con-
SWOFFORD, D. L. 2002. PAUP*: Phylogenetic analysis using parsi- clusions of Dr J. C. Willis. Philosophical Transactions of the Royal
mony (*and other methods), version 4.0b10. Sinauer, Sunderland, Society of London. Series B, Biological Sciences 213: 21–87.
Massachusetts, USA. ZWICKL, D. J. 2006. Genetic algorithm approaches for the phylogenetic
TABERLET, P., L. FUMAGALLI, A.-G. WUST-SAUCY, AND J.-F. COSSON. 1998. analysis of large biological sequence data sets under the maximum
Comparative phylogeography and postglacial colonization routes in likelihood criterion. Ph.D. dissertation, University of Texas, Austin,
Europe. Molecular Ecology 7: 453–464. Texas, USA.
794 AMERICAN JOURNAL OF BOTANY

APPENDIX 1. Species sampled in this study, GenBank accession numbers for ITS, trnL, trnL-F sequences, and voucher information. Herbarium abbreviations follow
Holmgren et al. (1990) and are defined in the chapter “Plant material” in the “Materials and methods” section.

Taxon, [internal code; GenBank accessions: ITS; trnL; trnL-F; Herbarium; Voucher number if provided or collector details; origin; ploidy level if determination
was possible].

Arabis alpina L., [MA001; GU182038; GU181905; GU181973; MA; 321468; [JR048; FJ187938; FJ188235; FJ188084; HEID; 400762; Croatia];
Spain], [MA004; HQ200931; GU181906; GU181974; MA; 224236; Arabis ionocalyx Boiss., [RK117; HQ200936; HQ200950; HQ200943;
Canary Islands], [RK001; GU182066; GU181934; GU181999; HEID; B; 10 0201502; Turkey], [RK118; HQ200937; HQ200951; HQ200944;
809695; Slovakia], [RK001b; GU182067; GU181935; GU182000; HEID; B; 10 0201504; Turkey], [RK252; JN189752; JN189731; JN189772; E;
502914; Turkey], [RK002; GU182068; GU181936; GU182001; NSK; 00381016; Turkey], [RK253; -; JN189732; JN189773; E; 00381020;
A. Telpukhovskaya, 1964, s.n.; Siberia/Russia], [A01JO2; DQ060109; Turkey], [RK254; JN189753; JN189733; JN189774; E; 00381021;
DQ060115 (trnLF); BM; T.R.I. Woods, 09.05.1975, no. Y/75/176; Syria], [RK255; JN189754; JN189734; JN189775; E; 00381025; Turkey];
Yemen], [A19JO2; DQ060102; DQ060113 (trnLF); BM; J.R. Press & M.J. Arabis kennedyae Meikle, [RK215; HQ646643; HQ646764; HQ646703;
Short, 29.03.1984, no. 398; Madeira], [A22JO; DQ060101; DQ060116 HEID; 809841; M. Andreou & C. Constantinou, 28.04.2010, AR-23-4;
(trnLF); BM; A. Hemp, s.n.; Tanzania], [A36JO2; DQ060100; DQ060112 Cyprus]; Arabis montbretiana Boiss., [A32MO; GU182046; GU181912;
(trnLF); BM; N.K.B. Robson, 07.04.1973, no. 3004; Turkey], [A40JO2; GU181980; MO; 1262438; Pakistan], [A59M; FJ187865; FJ188162;
DQ060105; DQ060128 (trnLF); BM; A. Hemp, no. 3676; Ethiopia], FJ188011; M; Sh. Zarre, 06.04.1999, no. 258; Iran], [RK043; HM046193;
[AW-30; DQ060101; DQ060117 (trnLF); WU; K.H. Rechinger, 28.07. - HM046245; HM046218; HUB; B. Mutlu, 30.05.2002, no. BM-7968;
01.08.1957, no. 1960-3968; Iraq]; Arabis aubrietioides Boiss., [A01M; Turkey], [RK121; HQ200938; HQ200952; HQ200945; B; 10 0201515;
FJ187845; FJ188142; FJ187991; M; ex. cult., 1976; Turkey], [MG003; Iran]; Arabis nordmanniana Rupr., [A54MO; FJ187924; FJ188221;
HQ646647; HQ646768; HQ646707; C; Th. Kotschy, 25.06.1862, no. FJ188070; MO; 05060970; Georgia]; Arabis nova subsp. iberica Rivas
88; Turkey], [RK240; -; JN189720; JN189761; E; 00381050; Turkey], Mart. ex Talavera, [MA022; GU182039; GU181907; GU181975; MA;
[RK241; JN189742; JN189721; JN189762; E; 00381053; Turkey], 732814; Spain]; Arabis nova subsp. nova Vill., [A34M; GU182063;
[RK243; JN189743; JN189722; JN189763; E; 00381056; Turkey], GU181929; GU181995; M; Doppelbaur, 06.08.1963, no. 11160; Italy],
[RK244; JN189744; JN189723; JN189764; E; 00381057; Turkey], [MA024; GU182041; GU181909; GU181977; MA; 48509; France];
[RK261; JN189759; JN189739; JN189780; MA; 765874; Turkey];
Arabis parvula Dufour, [A38M; FJ407224; GU181931; FJ407244; M;
Arabis aucheri Boiss., [A06MO2; FJ187881; FJ188178; FJ188027;
D. Podlech, 11.04.1993, no. 51465; Morocco]; Arabis purpurea Sm.,
MO; 2363411; Syria]; Arabis auriculata Lam., [A56M; FJ187863;
[A53M; FJ187861; FJ188158; FJ188007; M; W. Lang, 04.04.1999, s.n.;
FJ188160; FJ188009; M; F.A. Tscherning, 09.06.1895, s.n.; Austria],
Cyprus], [RK004; GU182070; GU181938; GU182003; HEID; 809699;
[A58M; FJ187864; FJ188161; FJ188010; M; D. Podlech, 28.04.1987, no.
Cyprus; diploid], [RK006; GU182072; GU181940; GU182005; HEID;
43319; Morocco], [RK027; GU182091; GU181959; GU182024; HEID;
809703; Cyprus; diploid], [RK247; JN189747; JN189726; JN189767; E;
809774; Uzbekistan]; Arabis blepharophylla Hook. & Arn., [Arab1795;
FJ187986; FJ188287; FJ188136; GH; R.C. Rollins, 30.04.1942, no. 00381038; Cyprus; diploid], [RK257; JN189755; JN189735; JN189776;
3006; California/USA]; Arabis brachycarpa Rupr., [A35MO; FJ187913; E; 00124777; Cyprus], [RK258; JN189756; JN189736; JN189777; E;
FJ188210; FJ188059; MO; 04958380; Armenia]; Arabis christani 00381041; Cyprus], [RK259; JN189757; JN189737; JN189778; HEID;
N. Busch, [JR099; GU182044; FJ188271; FJ188120; HEID; 502919; 00381043; Cyprus], [RK260; JN189758; JN189738; JN189779; HEID;
Armenia]; Arabis cypria Holmboe, [RK086; HQ200932; HQ200946; 00381048; Cyprus]; Arabis sachokiana N. Busch, [A69MO; FJ187929;
HQ200939; B; 10 0206185; Cyprus], [RK087; HQ200933; HQ200947; FJ188226; FJ188075; MO; 05060963; Georgia]; Arabis tianschanica
HQ200940; B; 10 0206362; Cyprus], [RK167; JN189741; JN189719; Pavl., [RK007; GU182073; GU181941; GU182006; OSBU; 16518;
JN189760; HEID; 809784; Cyprus; diploid], [RK245; JN189745; Kyrgyzstan]; Arabis verna (L.) R. Br., [A20MO2; FJ187890; FJ188187;
JN189724; JN189765; E; 00381031; Cyprus], [RK246; JN189746; FJ188036; MO; 05040147; Greece]; Aubrieta canescens subsp.
JN189725; JN189766; E; 00381036; Cyprus]; Arabis deflexa Boiss., macrostyla Cullen & Hub., [RK214; HQ646642; HQ646763; HQ646702;
[RK088; HQ200934; HQ200948; HQ200941; B; 10 0212852; Turkey], HEID; ex cult., 2010; Turkey]; Aubrieta deltoidea (L.) DC., [AJ232909;
[RK089; HQ200935; HQ200949; HQ200942; B; 10 0212853; Turkey], DQ180257; DQ180303; sequences from GenBank]; Aubrieta parviflora
[RK248; JN189748; JN189727; JN189768; E; 00381005; Turkey], Boiss., [DQ357518; —; —; sequence from GenBank]; Draba hederifolia
[RK249; JN189749; JN189728; JN189769; E; 00381008; Turkey], (Coss.) Hyam & Jury, [L241; DQ467438; DQ467072 (trnLF); BM;
[RK250; JN189750; JN189729; JN189770; E; 00381009; Turkey], Ait-Lafkikh et al., 01.10.1991, no. 4920; Morocco]; Draba hispanica
[RK251; JN189751; JN189730; JN189771; E; 00381014; Turkey]; Boiss., [B76; DQ467289; DQ466988 (trnLF); B; 10 0127430; Spain];
Arabis erikii Mutlu, [RK042; HM046192; HM046244; HM046217; Draba reptans (Lam.) Fernald, [72; AF146495; AF146942; AF146979;
HUB; B. Mutlu, 10.06.1999, BM-4900; Turkey]; Arabis graellsiiformis Muehlenbach, 20.04.1971, no. 3498; Missouri/USA]; Pseudoturritis
Hedge, [MG056; HQ646648; HQ646769; HQ646708; C; P.H. Davis & turrita (L.) Al-Shehbaz, [A52JO; GU592179; GU592180; GU592181;
O.V. Polunin, 08.07.1954, no. 22574; Turkey]; Arabis hirsuta (L.) Scop., BM; J. Vogel, s.n.; Italy].

Vous aimerez peut-être aussi