Molecular Phylogenetics and Evolution 42 (2007) 92–103 www.elsevier.com/locate/ympev
Circumpolar phylogeography of Juncus biglumis (Juncaceae) inferred from AFLP Wngerprints, cpDNA sequences, nuclear DNA content and chromosome numbers
Peter Schönswetter a,¤, Jan Suda b,c, Magnus Popp a, Hanna Weiss-Schneeweiss d, Christian Brochmann a
a National Centre for Biosystematics, Natural History Museum, University of Oslo, P.O. Box 1172, Blindern, NO-0318 Oslo, Norway b Department of Botany, Charles University, Benatska 2, CZ-128 01 Prague, Czech Republic c Institute of Botany, Academy of Sciences of the Czech Republic, Pruhonice CZ-252 43, Czech Republic d Department of Systematic and Evolutionary Botany, Faculty Centre Botany, University of Vienna, A-1030 Vienna, Austria
Received 20 January 2006; revised 1 June 2006; accepted 14 June 2006 Available onilne 6 July 2006
Abstract
We explored the circumpolar phylogeographic history of the arctic-alpine Juncus biglumis using ampliWed fragment length polymor- phisms (AFLPs), sequences of cpDNA, relative nuclear DNA content and chromosome numbers. The analyses of the AFLP and cpDNA data gave congruent results and revealed three distinct clades. One of them, represented by a single population from the Taymyr penin- sula in northern Siberia, had approximately fourfold larger genome size than the other samples and produced an AFLP pattern that was too aberrant to be analysed together with the rest of the data set. The two other clades represented diVerent ploidy levels (2n D 60 and 120) as judged from chromosome counts of selected populations but diVered only in c. 6% relative DNA content. Based on the AFLP and partly also on the cpDNA data, each of the two main clades was further subdivided into two well-supported subgroups. Three of the sub- groups were widespread and exhibited largely overlapping distribution patterns. The fourth subgroup seems to be absent from the North Atlantic region and from western Siberia. We suggest that the four subgroups diverged during isolation in diVerent glacial refugia during the Quaternary. Interestingly, individuals of both main clades were encountered in geographically close populations in eastern Greenland and even within a single population from Svalbard, indicating that both areas were colonised at least twice. The diVerent genome sizes and ploidy levels strongly suggest that the three main clades represent distinct gene pools and act as cryptic species. © 2006 Elsevier Inc. All rights reserved.
Keywords: AFLP; Arctic-alpine; cpDNA; Flow cytometry; Immigration; Juncus; Ploidy levels
1. Introduction studies on circumpolar plants “should lead to a more com- plete understanding of arctic plant evolution since the late In contrast to the growing body of literature exploring Tertiary and make clear whether isolation in glacial refugia circumpolar phylogeographic patterns in arctic animals was as important in shaping patterns of biodiversity within (reviewed in Hewitt, 2004), few studies have searched for arctic plants as in temperate species” (Abbott et al., 2000). such patterns in arctic plant species. Five years have passed So far, only two such studies have been published in full, since it was stated that further range-wide phylogeographic namely a pioneering study on the purple saxifrage, Saxi- fraga oppositifolia (Abbott et al., 2000) and a recent study on the bog bilberry, Vaccinium uliginosum (Alsos et al., * Corresponding author. Present address: Department of Biogeography and Botanical Garden, Faculty Centre Botany, University of Vienna, 2005). Whereas phylogeographic breaks in arctic animals A-1030 Vienna, Austria. Fax: +43 1 4277 9541. exhibit some well-founded congruencies across the investi- E-mail address: [email protected] (P. Schönswetter). gated species (Hewitt, 2004), the low number of relevant
1055-7903/$ - see front matter © 2006 Elsevier Inc. All rights reserved. doi:10.1016/j.ympev.2006.06.016 P. Schönswetter et al. / Molecular Phylogenetics and Evolution 42 (2007) 92–103 93 studies has impeded a comparative phylogeographic approach towards understanding the driving forces that shaped the distribution patterns of arctic plants. In the arctic-alpine S. oppositifolia, there are two main clades of cpDNA haplotypes with mainly Eurasian and North American distributions, respectively, overlapping on the Taymyr Peninsula in northern Central Siberia and in northern Greenland (Abbott et al., 2000; Abbott and Brochmann, 2003; Abbott and Comes, 2004). One clade is distributed from eastern North America over Greenland and Europe to Taymyr, and the other ranges from Taymyr over North America to Greenland. The authors suggest that S. oppositifolia Wrst occurred in the Arctic in the Tay- myr area, from where it migrated east- and westwards to achieve a circumpolar distribution (Abbott et al., 2000; Abbott and Brochmann, 2003). Late-Tertiary fossils from Greenland suggest that, that area had already been colon- ised before the Pleistocene. The populations in the southern European mountain ranges (Alps, Pyrenees) have wide- spread Eurasian haplotypes. In the more boreal-montane Fig. 1. Total distribution (broken line) and sampled populations of Juncus V. uliginosum sensu lato three cpDNA clades were identi- biglumis. Symbols indicate diVerent nuclear DNA content and phylogeo- Wed, one amphi-Atlantic, one Beringian and one “arctic- graphical groups: Wlled circles, Alps-group, open circles, W. Canada-group, alpine” lineage (Alsos et al., 2005). The latter has a circum- open squares, Scandinavia-group, and Wlled square, E. Canada-group (see polar distribution and also includes one fully circumpolar the text for more details). A diamond symbolises the putatively highly haplotype. The divergence among the three clades was esti- polyploid Ary–Mas population with approximately fourfold nuclear DNA content. The sample from Cape Espenberg, Alaska (semi-Wlled mated to have taken place before the onset of the major square, not numbered) was not included in the AFLP study; its aYliation Quaternary glaciations >700,000 years ago. to the group with lower relative DNA content was determined with Xow Our study taxon, Juncus biglumis L., belongs to a group cytometry. Note that on Svalbard (populations 9–11), both the Alps- and of closely related species with its centre of diversity in the the Scandinavia-groups occur. See Table 1 for details of the sampled popu- Himalayas (Kirschner, 2002). It is one of the most charac- lations. teristic vascular plants of the Arctic (Polunin, 1940) and has a continuous circumpolar distribution reaching well over 2. Materials and methods 80° north on Greenland and Ellesmere Island (Canada; Hultén and Fries, 1986). The species also grows in more 2.1. The study species southern mountain ranges such as the Rocky Mountains of North America, the Central Asian mountain ranges from Juncus biglumis L. is a low-growing (2–30 cm) perennial the Altai north-eastwards, and— isolated from the arctic rush forming small dense tufts. It grows in a wide range of populations—in the European Alps (Fig. 1). In the North rather wet habitats in diVerent successional stages (Aiken Atlantic region, it is found on several islands and archipela- et al., 1999). The species is wind-pollinated, but apparently gos, e.g., in the British Isles (Scotland), Iceland, Greenland nothing is known about its breeding system. Seed dispersal and Svalbard. Two ploidy levels (2n D 60, 120) have been may be partly anemochorous since the seeds are small and documented in J. biglumis so far, both of them in Scandina- light. Hydrochory may also be important, as it has been via (Knaben and Engelskjøn, 1967; Engelskjøn and Kna- shown that seeds can Xoat for about a year (Goodson et al., ben, 1971). Chromosome counts from other areas have only 2001). revealed the higher ploidy level (2n D >100, Johnson and Packer, 1968; 2n D 130 § 10, Mosquin and Hayley, 1966), 2.2. Sampling but are less exact. The aim of the present study is to unravel the phylogeo- Twenty-Wve populations of J. biglumis were sampled in graphic history of J. biglumis mainly focussing on western 2003 and 2004, covering the entire range of the species but Eurasia and Greenland using ampliWed fragment length poly- focusing on the North Atlantic region (Table 1). Typically, morphisms (AFLPs), sequences of chloroplast DNA leaf material of 10 plants was collected per population. (cpDNA), Xow cytometry data and chromosome numbers. In Voucher specimens are deposited at the Faculty Centre particular, we ask (1) whether the Alpine populations are Botany, University of Vienna (WU), or in the Botanical most closely related to the Scandinavian populations as Museum, University of Oslo (O). Living plants from popu- hypothesised in the classical biogeographic literature (Vierh- lations 2 and 6 were collected in 2005 and cultivated in the apper, 1918; Noack, 1922); and (2) what are the source areas Botanical Garden of the University of Vienna (HBV), to for the colonisation of Greenland and Svalbard. conWrm Xow cytometry results conducted with silica 94 P. Schönswetter et al. / Molecular Phylogenetics and Evolution 42 (2007) 92–103
Table 1 Geographic origin, average gene diversity over loci, and average relative Xuorescence intensity of 4Ј,6-diamidino-2-phenylindole (DAPI)-stained nuclei (with Bellis perennis, 2C D 3.38 pg, as internal standard) in the investigated populations of J. biglumis
No. Country Sampling locality Long/lat NAFLP Average gene NFCM Average relative diversity § SD Xuorescence intensity § SD 1 Great Britain: Ben Lawers ¡4.78/58.33 8 0.0099 § 0.0073 4 0.453 § 0.004 E Scotland 2 Austria Rothenkarscharte 13.47/47.19 9 0.0069 § 0.0055 3 0.435 § 0.002 CD 3 Austria Elendkar 13.50/47.02 10 0.0061 § 0.0049 3 0.431 § 0.003 CD 4 Norway Jonfunut 7.54/60.61 9 0.0088 § 0.0066 2 0.414 § 0.001 AB 5 Norway Krossbu 8.04/61.59 10 0.0073 § 0.0056 3 0.415 § 0.001 AB 6 Norway South–eastern 8.58/61.38, 8 0.0284 § 0.0176 3 0.406 § 0.003 A Jotunheimen: 8.82/61.45 TorWnnsdalen, Brudskardsknappen 7 Sweden Latnjachorru 18.50/68.33 9 0.0092 § 0.0068 3 0.415 § 0.001 AB 8 Norway Lávkaslubbu 20.47/69.25 9 0.0360 § 0.0214 4 0.408 § 0.003 A 9Norway: Sørkapp Land: 15.88/76.94 10 0.0030 § 0.0031 3 0.435 § 0.004 CD Svalbard Gåshamna 10 Norway: Wedel Jarlsberg Land: 15.50/77.00 10 0.0525 § 0.0299 3 0.430 § 0.005 CD Svalbard Fuglebergsletta 11 Norway: Wedel Jarlsberg Land: 15.18/77.07 10 0.1135 § 0.0621 7 0.415 § 0.01 Svalbard Nottinghambukta 0.430 § 0.003 CD 0.409 § 0.004 A) 12 Russia Nenetskaya Gryada 53.24/68.36 8 0.0134 § 0.0093 3 0.424 § 0.005 BC 13 Russia Seyda 63.07/67.05 8 0.0260 § 0.0162 4 0.412 § 0.002 AB 14 Russia Polar Ural: Chernaya 65.50/66.84 10 0.0527 § 0.0299 3 0.434 § 0.004 CD 15 Russia Altai: Pass Teply Kluch 88.04/49.42 10 0.0011 § 0.0017 3 0.483 § 0.005 F 16 Russia Khatanga 102.57/71.97 10 0.0301 § 0.0179 3 0.435 § 0.004 CD 17 Russia Lena River: Tiksi 128.87/71.64 10 0.0731 § 0.0407 3 0.430 § 0.004 CD 18 Russia Lena River: Chekurovka 127.51/71.06 8 0.0400 § 0.0240 3 0.441 § 0.004 DE 19 Canada Wright Pass ¡136.25/67.05 6 0.0298 § 0.0184 3 0.430 § 0.004 CD 20 Canada Bellot Strait ¡94.23/72.02 8 0.0499 § 0.0294 3 0.405 § 0.002 A 21 Greenland Qeqertat ¡66.70/77.50 8 0.0166 § 0.0111 4 0.425 § 0.002 BC 22 Greenland Antarctic Havn ¡23.11/71.99 8 0.0465 § 0.0275 5 0.415 § 0.002 AB 23 Greenland Myggbukta ¡21.58/73.50 7 0.0611 § 0.0364 3 0.437 § 0.003 CD Ary-Mas Russia Taymyr: Ary-Mas 101.86/72.46 — — 4 1.695 § 0.015 G Espenberg USA Cape Espenberg ¡163.98/66.59 — — 6 0.411 § 0.003 AB W X NAFLP, number of individuals investigated with AFLP ngerprinting. NFCM, number of individuals investigated with ow cytometry. The same letter indi- cates groups of populations with nuclei Xuorescence not signiWcantly diVerent at P D 0.05. For the heterogeneous population 11 from Svalbard, average relative Xuorescence intensity was determined separately for individuals belonging to main groups A and B (see text). Populations Ary–Mas and Espen- berg were only used for Xow cytometry. gel-dried samples (see below), and to determine their chro- individual, 2l 6-FAM-, 2 l VIC-, and 3 l NED-labelled mosome numbers. selective PCR products were combined with 0.3 l Gene- Scan ROX 500 (Applied Biosystems, Foster City, CA, 2.3. DNA extraction and AFLP Wngerprinting USA) and 11.7 l formamide. Blind samples and replicates were routinely included to test for contamination and Total genomic DNA was extracted from silica gel- reproducibility. Raw data were collected and aligned with dried leaf material following the 2 £ Cetyl trimethyl the internal size standard using ABI Prism GeneScan ver. ammonium bromide (CTAB) method (Doyle and Doyle, 3.7. analysis software (Applied Biosystems, Foster City, 1987) with minor modiWcations (Schönswetter et al., CA, USA). The GeneScan Wles were imported into Genog- 2002). The AFLP procedure followed Gaudeul et al. rapher (version 1.6; http://hordeum.oscs.montana.edu/ (2000), but the PCR reaction volume was reduced by 50%. genographer) for scoring. Fragments in the size range 60– Since initial tests using MseI primers with three selective 500 bp were scored and the results were exported as a nucleotides yielded few fragments, the following three presence/absence matrix. In the blind samples some frag- selective primer combinations were chosen (Xuorescent ments were detected which, however, were not present in dye in brackets): EcoRI ACT (6-FAM)–MseI CA; EcoRI the regular samples. The few fragments that were not AAC (NED)–MseI CA; EcoRI AAG (VIC)–MseI CT. reproducible in the c. 20 replicated samples were not For electrophoresis on a capillary sequencer ABI 3100 scored for the whole dataset. By mistake, population (Applied Biosystems, Foster City, CA, USA) and for each Espenberg was not included in the AFLP analysis. P. Schönswetter et al. / Molecular Phylogenetics and Evolution 42 (2007) 92–103 95
2.4. cpDNA sequencing pattern inferred from silica-dried samples, fresh material from two populations (populations 2 and 6) was Two chloroplast regions were ampliWed and sequenced evaluated using both DAPI and propidium iodide Xow in one individual from each population included in the cytometry. Five repetitions on diVerent days were AFLP analyses except population 11 from Svalbard (Table performed for each sample when estimating absolute 1), from which two individuals were sequenced. The rps16 genome size. intron was ampliWed and sequenced using the primers rpsF For fresh material from two populations absolute and rpsR2R (Oxelman et al., 1997), and the trnL–trnF genome size was estimated in propidium iodide-stained region was ampliWed and sequenced using primers tabC nuclei (propidium iodide + RNase IIA, both at concentra- and tabF (Taberlet et al., 1991). In addition, tabD and tabE tions of 50 l/ml) using two-step Otto procedure with cen- (Taberlet et al., 1991) were used as internal sequencing trifugation as described in Suda et al. (2003). primers. PCR conditions were 0.4 U Taq (ABgene, Epsom, Lycopersicum esculentum cv. Stupické polní tybkové rané UK) per 10 l reactions, buVer supplied with the enzyme, (2C D 1.96 pg; Doleqel et al., 1992) was selected as an 2.5 mM Mg2+, 0.4 M of each primer, 1 mM of each dNTP appropriate internal standard due to its genome size close (Applied Biosystems, Foster City, CA, USA), 0.04% BSA to that of Juncus samples (to minimise potential instru- and 3 l template DNA of unknown concentration. PCR ment non-linearity). Measurements were performed on cycling was performed with a GeneAmp 3700 (Applied Partec CyFlow cytometer with a green solid state laser Biosystems), PTC100 or PTC200 (MJ Research, Water- (Cobolt Samba 532 nm, 100 mW, Cobolt, Stockholm, town, MA) thermocycler with the following parameters: Sweden) as excitation source. initial denaturation for 5 min at 95 °C followed by 35 cycles of 30 sec at 95 °C, 30 sec at 52 °C (trnL–trnF) or 56 °C 2.6. Chromosome counts (rps16), and 2 min at 72 °C, ending with 10 min extension at 72 °C. Sequencing was performed using BigDye V 1.1 Two populations representing the two main clades (Applied Biosystems) according to the manufacturer’s revealed by the molecular analyses (see below) were ana- manual except for using 10 l reaction volumes, and visual- lysed karyologically (populations 2 and 6). Actively grow- ised with an ABI 3100 capillary sequencer (Applied Biosys- ing root tips were pre-treated with 0.1% colchicine tems). (Harriman and Redmond, 1976) in darkness at room tem- perature for 1 h, Wxed in 3:1 absolute ethanol:glacial acetic 2.5. Flow cytometry acid at room temperature for 24 h and stored at ¡20 °C until use. Chromosomes were stained with SchiVs’s reagent The relative nuclear DNA content of 2–7 individuals (Sigma–Aldrich, Vienna, Austria; according to a standard per population was determined by Xow cytometry. Nuclei Feulgen method; Weiss et al., 2003) or with acetocarmine isolation and staining generally followed the protocol (Sigma–Aldrich). The time of hydrolysis for the Feulgen used for ploidy estimation in dehydrated plant vouchers method (5 N HCl treatment at room temperature) ranged (Suda and Trávníbek, 2006). Silica gel-dried leaves and/or from 5 to 45 min, followed by staining with SchiV’s reagent. stems of the analysed Juncus sample were chopped Alternatively, chromosomes were prepared by enzymatic together with an appropriate volume of fresh leaf tissue of digestion/squashing as described by Schwarzacher and the internal standard (Bellis perennis, 2C D 3.38 pg; J. Heslop-Harrison (2000) and Weiss-Schneeweiss et al. Suda, unpublished) in a Petri dish containing 0.5 ml fresh (2003), with the following modiWcations. Material was Otto I buVer (Otto, 1990). The crude suspension was digested with 0.4% (w/v) cytohelicase (Sigma–Aldrich), 1% Wltered through a 42 m nylon mesh and incubated at (w/v) cellulase Onozuka (Serva, Heidelberg, Germany), and room temperature for 30 min. The staining solution con- 0.4% (w/v) pectolyase (Sigma–Aldrich) for 10–30 min at sisted of 1 ml of Otto II buVer supplemented with 4Ј,6- 37 °C. Squash preparations were made in a drop of 45% diamidino-2-phenylindole (DAPI; Wnal concentration acetic acid. Slides were frozen and, after cover slip removal, 4 l/ml) and 2-mercaptoethanol (2 l/ml). The relative stained with 2g/ml DAPI in a Vectashield mounting Xuorescence of isolated nuclei was recorded using a Partec medium (Vector Laboratories, Burlingame, CA, USA). PA II Xow cytometer (Partec GmbH, Münster, Germany) Preparations with at least 10 good-quality chromosome equipped with a mercury arc lamp (HBO-100 long life, spreads were analysed with an epiXuorescence microscope Osram, München, Germany). The cytometer was adjusted (Axioplan2, Zeiss, Oberkochen, Germany). Images were so that the G0/G1 peak of the internal standard was acquired with a CCD camera and AXIOVISION software located on channel 300 (using 1000 linear scale). The Xow (Zeiss). rate was 30–50 events per second and Xuorescence of 5000 particles was scored. 2.7. Analysis of AFLP data To minimise potential between-day instrument Xuctuation, more than 40% of DAPI-stained samples Average gene diversity over loci was calculated using were re-analysed two or three times on diVerent days. To ARLEQUIN 2.0 (Schneider et al., 1997). We performed further conWrm the observed nuclear DNA content neighbour joining (NJ) analyses of Nei and Li (1979) 96 P. Schönswetter et al. / Molecular Phylogenetics and Evolution 42 (2007) 92–103 genetic distance matrices with TREECON 1.3b (Van de rical peaks of approximately the same height were used Peer and De Wachter, 1997). The trees were midpoint for subsequent calculations. Between-population diVer- rooted. Branch support was estimated with 500 bootstrap ences in relative DNA content were tested using SAS 8.1 replicates. (SAS Institute, NC, USA), procedure General Linear Bayesian analysis of population structure (BAPS v.3.1, Model (GLM; because of unbalanced data design), and Corander et al., 2003, available at http://www.rni.helsinki.W/ Tukey’s algorithm was applied to compare the mean ~jic/bapspage.html) was used to estimate hidden popula- values. The Wnal population grouping was performed tion substructure by clustering individuals into panmictic manually. groups. The program treats both the allele frequencies of the molecular markers and the number of genetically 3. Results diverged groups as random variables. Stochastic optimisa- tion is used to infer the posterior mode of the genetic struc- 3.1. AFLP ture. BAPS was run with the maximal number of groups (K) set to 1–25. Each run was replicated three times. The We scored 176 fragments, of which 153 (86.9%) were partition with the highest log marginal likelihood was plot- polymorphic. Average gene diversity over loci varied from ted onto the neighbour joining tree. 0.0011 in population 15 (Russia, Altai) to 0.1135 in popula- Analyses of molecular variance (AMOVAs) were com- tion 11 (Svalbard, Nottinghambukta) (Table 1). Population puted with ARLEQUIN 2.0. To explore the origin of the Ary–Mas (Russia, Taymyr) produced AFLP banding pat- Alpine populations, we performed assignment tests based terns that were poorly reproducible and too strongly devi- on the multilocus genetic data following Duchesne and ating from all other accessions to be included in a joint Bernatchez (2002), using their AFLPOP v1.0 program with analysis. the following settings: marker frequencies of zero were In the neighbour joining analysis (Fig. 2), two major replaced by (1/sample size + 1); the minimal log likelihood groups were separated with high bootstrap support. diVerence to allocate an individual was set to 2, i.e. it was Within major group A, two well-supported subgroups only allocated if the allocation to a certain population was were diVerentiated. The Wrst subgroup comprised popula- 100 times more probable than to another population; and tions from Scotland (population 1), Svalbard (popula- the number of artiWcial (simulated) genotypes to compute tions 9, 10, and three individuals from population 11), the P-values was set to 500. Alps (populations 2 and 3), Greenland (populations 21 and 23), the Urals (populations 12 and 14), and central 2.8. Analysis of cpDNA sequence data (population 16) and eastern (population 18) Siberia. In the following, this subgroup is referred to as the The sequences were edited using STADEN ver. 1.5.3 Alps-group. The second subgroup, the W. Canada-group, (https://sourceforge.net/projects/staden) and manually consisted of populations from the Altai mountains (popu- aligned using Se–Al v.2.0a11 (Rambaut, 1996). Eight gaps lation 15), eastern Siberia (population 17) and W. Canada of varying length had to be introduced in the alignment. (population 19). Major group B was also divided into two Because of sequencing problems it was diYcult to deter- well-supported subgroups. The Wrst comprised the single mine the number of nucleotides in several poly-A/T regions, population analysed from E. Canada (population 20) and and we chose not to use the putative phylogenetic informa- the second, referred to as the Scandinavia-group, con- tion contained in the gaps in the analyses. tained populations from Scandinavia (populations 4–8), Phylogenetic analyses were made using PAUP¤ (Ver. Svalbard (seven individuals from population 11), the 4.0b10) for Macintosh (SwoVord, 2002). Maximum parsi- Urals (population 13) and Greenland (population 22). mony analyses were performed using a heuristic search The relationships within the four subgroups were mostly strategy starting from 100 random trees, the MULTREES unresolved. Average gene diversity over loci did not option on, and DELTRAN optimisation. Maximum parsi- diVer signiWcantly among the four subgroups (t-tests, all mony bootstrap analyses were carried out with full heuris- P-values À 0.05). tics, 1000 replicates, TBR branch swapping, the The partition with the highest log marginal likelihood MULTREES option oV, and random addition of sequences (¡5538.8) produced by BAPS consisted of eight clusters with four replicates. trnL–trnF sequences from J. castaneus, that corresponded to the four subgroups in the NJ analysis J. stygius, and J. triglumis (Accession Nos. AY437954–56) with further subdivisions. Groups that were characterised were used for outgroup rooting in accordance with Dráb- by long, well-supported branches in the NJ analysis were ková et al. (2004). recognised as separate BAPS clusters, i.e. each of the three populations of the W. Canada group and the two Alpine 2.9. Flow cytometry populations. Not so obvious from NJ analysis was the Wnd- ing that the population from Scotland and most individuals The histograms were evaluated by Partec FloMax from populations 9 and 10 from Svalbard were separated software (both Find peak algorithm and manual peak from the other members of the Alps-group. If BAPS was gating were applied). Only acquisitions yielding symmet- run with K (maximal number of groups) D 4, the grouping P. Schönswetter et al. / Molecular Phylogenetics and Evolution 42 (2007) 92–103 97
Fig. 2. Neighbour-joining analysis of AFLP phenotypes of Juncus biglumis based on Nei and Li’s (1979) genetic distances. Numbers above major branches are bootstrap values higher than 50% (500 replicates). Numbers at the tips of branches are population numbers (Table 1). The four main branches are indi- cated with the same symbols as in Fig. 1. The optimal grouping achieved by Bayesian model clustering with BAPS is plotted to the right of the geographi- cal area names; the groups are labelled from 1 to 8. Populations with diVerent relative nuclear DNA contents (corresponding to the main group A with three variants and the group B) are symbolised with vertical bars (using Bellis perennis, 2C D 3.38 pg, as unit value). Population Ary–Mas was excluded because of its highly deviating AFLP banding pattern that precluded a common analysis with the other accessions. was identical to the four subgroups identiWed by the NJ (“Wxed”). In the Scandinavia-group and in population 20 analysis. from eastern Canada, we detected four and Wve Wxed pri- We detected 12 private fragments in the Alps-group; vate fragments, respectively. In the W. Canada-group, we none of which were found in all investigated individuals found four private fragments, one of which was Wxed. 98 P. Schönswetter et al. / Molecular Phylogenetics and Evolution 42 (2007) 92–103
Table 2 Analyses of molecular variance (AMOVA) for AFLP phenotypes in Juncus biglumis a Source of variation df Sum of squares Variance components % Total variance FST Within populations 182 504.48 2.77 14.95 Among populations 22 3152.13 15.77 85.05 0.85 Within populations 181 425.91 2.35 9.09 Among populations within groups 22 1778.68 9.21 35.59 Among groups (1, 2) and (3, 4) 1 1452.02 14.32 55.32 0.91 Within populations 181 425.91 2.35 10.00 Among populations within groups 20 1186.04 6.70 28.44 Among groups (1, 2, 3 and 4) 3 2044.66 14.49 61.57 0.90 The grouping in the hierarchical AMOVA was derived from the neighbour joining analysis presented in Fig. 2 (1, Alps-group; 2, W.Canada-group; 3, E.Canada-group; 4, Scandinavia-group). The heterogeneous population 11 was separated into the Alps-group (three individuals) and the Scandinavia-group (seven individuals). a All P-values were <0.001.
AMOVAs (Table 2) attributed 85% of the overall genetic variation to the among-population component. In nested AMOVA analyses, the variation between the two main groups A and B accounted for 55% and the variation among the four subgroups for 62% of the overall variation. In the assignment test, the Alpine individuals were allo- cated to the populations from Svalbard. If those were excluded from the analysis, the Alpine individuals were allocated to the Scottish population, and if the latter was removed, they were allocated to the populations from the Urals (all P-values < 0.01).
3.2. cpDNA sequence data
The concatenated trnL–trnF and rps16 matrix (EMBL/ GenBank Accession Nos. AM085712–AM085761) con- sisted of 1851 aligned positions for 25 ingroup terminals and three outgroup taxa. Fifteen of the 55 variable charac- ters were parsimony informative (seven of 18 characters excluding the outgroup). Maximum parsimony analysis resulted in a single most parsimonious tree of 59 steps with CI/RI D 0.882/0.974 excluding uninformative characters (Fig. 3). No homoplasy was detected within the ingroup (CI D RI D 1 excluding the outgroup). The topology was congruent with the AFLP analysis but less resolved. Maxi- mum parsimony bootstrapping (MPB) supported the two major groups A and B, and the Alps-group (62, 99, and 62%, respectively; Fig. 3). Population Ary–Mas (Taymyr, Rus- Fig. 3. The single most parsimonious tree (20 steps, CI D 0.882 and sia) was sister to the rest of the J. biglumis-clade (MPB RI D 0.974, CI D RI D 1.000 excluding the outgroup) recovered in the 78%). analysis of the concatenated trnL–trnF and rps16 cpDNA regions. Num- bers connected to nodes indicate bootstrap values. Branch lengths are proportional to number of changes. Symbols indicate diVerent nuclear 3.3. Nuclear DNA content DNA content and phylogeographical groups: Wlled circles, Alps-group, open circles, W. Canada-group, open squares, Scandinavia-group, and Relative nuclear DNA content was successfully esti- Wlled square, E. Canada-group (see the text for more details). A diamond mated in 88 out of 91 samples (33 of them were re-ana- symbolises the Ary–Mas population. lysed twice, and three of them three times). DAPI-stained nuclei isolated from silica gel-dried Juncus tissues mostly intra-population variation in relative nuclear DNA con- formed very distinct peaks with unimodal Xuorescence tent (except for population 11, see below) nor variation distribution. Their coeYcient of variation (CV) varied between diVerent runs of the same sample exceeded 3%, from 3.1% to 7.4% (mean 5.4%), showing hardly any cor- thereby demonstrating a stability of nuclei Xuorescence respondence with the age of the material. The CV of the properties. SigniWcant variation amounting to nearly 8% internal standard was always less than 3.4%. Neither was observed within population 11. P. Schönswetter et al. / Molecular Phylogenetics and Evolution 42 (2007) 92–103 99
The relative nuclear DNA values for individual popula- samples (0.424–0.441, mean 0.432) (Table 1). There was no tions are summarised in Table 1. The overall diVerence distinct DNA content diVerentiation at the subgroup level. among the analysed samples was considerable and spanned Absolute genome sizes (mean 2C-values § SD) for popu- a c. 4.2-fold range. However, population Ary–Mas from the lations 2 (Rothenkarscharte, Austria) and 6 (Bru- Taymyr area in northern Siberia had a mean relative Xuo- dskardsknappen, Norway) were estimated to 1.22 § 0.01 pg rescence value 1.695 (as against the internal standard B. and 1.13 § 0.01 pg, respectively, thus corroborating the perennis) and was thus probably highly polyploid; if this genome size pattern revealed in silica gel-dried material. population was excluded, the diVerence shrank to less than 20%. 3.4. Chromosome counts Good congruence was found between the relative DNA amount data and the groupings based on the AFLP and Determination of chromosome numbers using acetocar- cpDNA data (Figs. 2 and 3). Samples belonging to major mine and Feulgen stained material was diYcult due to group A possessed signiWcantly higher amounts of DNA weak stainability, small size of the chromosomes and low (mean Xuorescence 0.437 of the Bellis value, range 0.424– number of dividing cells. Therefore, chromosome numbers 0.483) than major group B (mean 0.411, range 0.405–0.415), for the two analysed populations were established by analy- corresponding to a diVerence of approx. 6.3% (Fig. 4). sis of enzymatic preparations stained with DAPI which Whereas the variation within group B was negligible proved to be more sensitive. The analysis of the images (2.5%), Tukey’s grouping revealed two outlier populations revealed 2n D 60 § 1 (Fig. 5A) in population 6 and in group A, namely population 15 from Altai and popula- 2n D 120 § 1 in population 2 (Fig. 5B). The chromosomes of tion 1 from Scotland, with higher relative Xuorescence val- population 6 were relatively larger than the chromosomes ues (0.483 and 0.453, respectively) than the remaining of population 2 (c. 60%). In both analysed populations, chromosomes of diVerent length were observed, rendering the exact establishment of chromosome numbers diYcult due to possibility of the presence of supernumerary chro- mosomes or counting “small DNA fragments” (similar to NOR-carrying satellites) as separate chromosomes. It was impossible to determine whether the chromosomes were holocentric or monocentric given their size and appearance. The chromosomes of both cytotypes harboured a rather large amount of heterochromatin as judged from DAPI staining of interphase nuclei (Fig. 5A).
4. Discussion
4.1. Three lineages with overlapping geographical distribution Fig. 4. Simultaneous Xow cytometric analysis of DAPI-stained nuclei iso- lated from silica gel-dried Juncus tissues from populations 20 (Canada, Bellot Strait) and 18 (Russia, Chekurovka), demonstrating divergence in AFLPs, sequences of cpDNA and relative nuclear DNA their relative nuclear DNA content. content yielded three congruent main groups in J. biglumis;
Fig. 5. Mitotic chromosomes of Juncus biglumis; (A), population 6 (2n D 60 § 1); (B), population 2 (2n D 120 § 1). Scale bar 5 m. 100 P. Schönswetter et al. / Molecular Phylogenetics and Evolution 42 (2007) 92–103
(i) population Ary–Mas; (ii) an arctic-alpine lineage termed numbers (not only in J. biglumis but also in the few other main group A; and (iii) a strictly arctic lineage, termed main counted species of the same section; Kirschner, 2002) rather group B. indicates polyploidisation. The small diVerence in genome Population Ary–Mas from the Taymyr area in northern size may in this case be explained by massive DNA elimina- Central Siberia had an approximately four-fold relative tion (Bennetzen, 2002), processes that can occur rapidly in nuclear DNA content as compared with the average of the newly synthesised polyploids (Song et al., 1995; Grover other samples (Table 1), indicating a higher ploidy level et al., 2004; Levy and Feldman, 2004). From our data it is than in the other populations (higher than 4£). The pres- impossible to speculate about the number of polyploidisa- ence of several copies of each AFLP fragment resulted in tion events. However, populations 1 from Scotland and 15 poorly reproducible AFLP banding patterns that deviated from the Altai had signiWcantly higher relative DNA con- strongly from all other accessions in having a much higher tents than all other investigated individuals of main group number of fragments. As a consequence, this population A (Fig. 2 and Table 1). This may suggest that these two could not be included in the AFLP analysis. Although the populations originated from independent polyploidisation cpDNA analyses resolved the Ary–Mas population as sis- events or that the genome evolution after polyploidisation ter to the rest of the J. biglumis clade (Fig. 3), it did not followed independent pathways. deviate morphologically, and further did not Wt the descrip- The geographic distributions of the two main groups tion of J. longirostris Kuvaev (Kirschner, 2002), a probably were strongly overlapping (Fig. 1). The main group A was doubtful taxon described from Taymyr. However, the pres- sampled in Scotland, the Alps, the Urals, the Altai, north- ence of a narrowly distributed, strongly divergent intraspe- ern Siberia, W. Canada, Greenland and Svalbard. The main ciWc lineage of J. biglumis on Taymyr suggests that Taymyr group B is spread from the western USA over eastern arctic may have played a key role in the early intraspeciWc diversi- Canada, eastern Greenland and Scandinavia to the Urals. Wcation of circumarctic plant species. This was previously Due to sampling gaps in the central part of arctic North suggested for phylogeographic data from S. oppositifolia America and in easternmost Siberia, the distribution of the (Abbott et al., 2000), Ranunculus pygmaeus and Minuartia two main groups in these areas cannot be determined, but, biXora (Schönswetter et al., 2006). as exempliWed by population 19 and the Espenberg popula- The two main groups A and B formed sister clades in the tion, both main groups occur in western North America MP analysis of cpDNA sequences (Fig. 3). The diVerentia- (Fig. 1). Based on the present sampling, the two groups co- tion between the two groups accounted for as much as occur in eastern Greenland, Svalbard and the Urals (Fig. 1), 55.3% of the overall AFLP variation (Table 2). The Xow partly in mixed populations. cytometry investigation identiWed the two main groups as Individuals of both main groups A and B were sampled DNA variants diVering by c. 6% in their relative DNA con- within a few square metres in southern Svalbard (popula- tent (Fig. 2 and Table 1). tion 11). It is plausible that the diVerent ploidy levels and Changes in DNA content are often connected to various especially the strong reduction of the genome size after types of changes in the karyotype, including eu- and hetero- polyploidization created strong reproductive barriers that chromatin ampliWcation/elimination, structural rearrange- prevent the two gene pools from homogenising. We thus ments and change of chromosome number (Bennett and suggest that the main groups A and B and potentially also Leitch, 2005). The two diVerent chromosome numbers the lineage represented by the Ary–Mas population should (2n D 60 and 120) determined for representatives of the two be regarded as cryptic species (e.g., Whittall et al., 2004). main clades A and B of J. biglumis suggest that change of Based on AFLP evidence, similar heterogeneous ‘popula- chromosome number was involved in the initial divergence tions’ of Comastoma tenellum (Schönswetter et al., 2004) of these groups. The two cytotypes diVer also in the size of and Veronica alpina (D. Albach, Johannes Gutenberg Uni- the chromosomes, as the 2n D 120 cytotype has substan- versity, Mainz, Germany, unpublished) have been reported tially smaller chromosomes (roughly about 60%) than the in the Alps. 2n D 60 cytotype. Morphologically, J. biglumis is fairly variable in charac- Two mechanisms can explain the observed increase in ters such as plant size, shape of the capsule, and presence chromosome number, polyploidisation or agmatoploidy, and width of a hyaline margin on the bracts subtending the the latter being chromosomal Wssion of holocentric chro- Xowers (R. Elven, Oslo, personal communication). We mosomes (Kuta et al., 2004). While the closely related genus could not, however, Wnd any morphological characters Luzula has been shown to possess holocentric chromo- diVerentiating between the main genetic groups A and B. somes (e.g., Haizel et al., 2005; Nagaki et al., 2005), this has Even the highly polyploid Ary–Mas population, which also never been proven for Juncus but often hypothesised (God- has a divergent cpDNA haplotype, does not deviate mor- ward, 1985). If Juncus chromosomes are holocentric, agma- phologically from the other investigated populations. toploidy might have given rise to the 2n D 120 cytotype (from 2n D 60), and the 6% increase in the genome size may 4.2. Further phylogeographic subdivision have been caused by processes such as heterochromatin ampliWcation. However, the presence of only two cytotypes According to the neighbour joining analysis of the both based on x D 60 without intermediate chromosome AFLP multilocus phenotypes, but not the nuclear DNA P. Schönswetter et al. / Molecular Phylogenetics and Evolution 42 (2007) 92–103 101 content data, the two main groups A and B were further (Abbott et al., 2000), and recent dispersal between Svalbard subdivided into four well-supported subgroups (Fig. 2). and Scandinavia was suggested for this species based on The four AFLP subgroups accounted for a high proportion RAPD evidence (Gabrielsen et al., 1997). In S. cernua, four of the overall genetic variation (Table 2). Within main widespread cpDNA haplotypes were sampled on both group A, the Alps-group comprised samples from Green- Svalbard and Greenland (Brochmann et al., 2003). Only in land over the Alps to northern Siberia, whereas the W. V. uliginosum, one of the rare “thermophilic” elements on Canada-group consisted of populations from the Altai, Svalbard, was only one cpDNA haplotype present in Sval- northern Siberia and W. Canada. Group B was divided into bard (Alsos et al., 2005). These Wndings clearly demonstrate a single population from eastern arctic Canada (population that the North Atlantic hardly is a dispersal barrier for 20) and the Scandinavia-group with samples from eastern many plant species, which probably dispersed across the Greenland, Svalbard, Scandinavia and the Urals. The frozen ocean (reviewed in Brochmann et al., 2003). (2) It Bayesian cluster analysis (Fig. 2) of the AFLP data showed has been suggested that immigration from “Nordic” popu- good congruence with the neighbour joining analysis, but lations was the source of the few populations of J. biglumis introduced a Wner splitting in the Alps- and the W. Canada- in the Alps (Vierhapper, 1918). Based on our data, Scandi- groups. Due to the little resolution in the parsimony analy- navia is unlikely a source area for the colonisation of the sis of the cpDNA sequence data, only the Alps-group was Alps (Figs. 2 and 3). Within the main group A, the Alpine recovered (Fig. 3). The resolution within the groups in the populations were Wrst assigned to previously glaciated neighbour joining analysis was limited. This again is cor- areas (Svalbard, Scotland), which are very unlikely source roborated by the Bayesian cluster analysis (Fig. 2), which areas, but rather were colonised from the same glacial refu- combined, for example, individuals from western Green- gia. The next likely, and potentially meaningful, allocation land eastwards to River Lena in eastern Siberia into one was to the Urals, a mountain range that remained largely (“panmictic”) gene pool. Altogether, there is strong evi- unglaciated during the Pleistocene. However, even if not dence in J. biglumis for extensive dispersal/gene Xow over supported by our data, immigration from Scandinavia can- enormous distances in relatively recent times. not be excluded as the presumably scarce presence of main The hierarchical organisation of AFLP (and partly also group A in Scandinavia is evidenced by a count of the cpDNA) divergence into three levels, i.e. (i) the split 2n D 120 cytotype from northern Scandinavia (Knaben and between population Ary-Mas and all other investigated Engelskjøn, 1967). individuals; (ii) between two main groups A and B; and (iii) Our study has revealed that circumpolar phylogeo- between their subgroups, suggests that gene pools have graphic patterns of arctic-alpine plants can be fairly com- been disrupted at three time horizons. Whereas the diver- plex. In our study species J. biglumis, only a multi-method gence within each main group may result from isolation in approach enabled a satisfactory interpretation of the Pleistocene refugia during the last glaciation(s), the diVer- observed AFLP and cpDNA sequence data that would entiation between the main groups A and B and the Ary– have remained dubious without the knowledge of the Mas lineage is probably much older. This result is in line genome size and chromosome number variants. Further with previous observations that the initial divergence chromosome counts from the investigated populations among the main lineages of arctic species often occurred at would be highly desirable to test if individuals of main the time when the present-day arctic ecosystem developed, groups A and B always have the same chromosome number i.e. c. 3 million years ago (Matthews and Ovenden, 1990). and to conWrm the polyploidy hypothesis. The time of divergence among the three main lineages of V. uliginosum has been estimated to 3.0–0.7 million years ago Acknowledgments (Alsos et al., 2005), and that of the two main clades of S. oppositifolia to 5.4–3.8 million years ago (Abbott and The authors thank I. G. Alsos, T. A. Carlsen, R. Elven, Comes, 2004). Other taxa, however, such as the arctic spe- O. Gilg, A. Granberg, K. Hamli, H. Jacobsen, P. Larson, M. cies of Cerastium, have been suggested to be much younger H. Jørgensen, M. Kapralov, A. Ross, I. Skrede, H. Solstad, (Scheen et al., 2004). A. Tribsch and K. Westergaard for collecting many of the Due to the lack of a supported pattern within the four plant samples used in the present study. J. Kirschner and R. subgroups and the largely overlapping distributions, the Elven kindly revised Juncus specimens, J. Kirschner also location of glacial refugia could not be determined. Genetic commented on earlier versions of the manuscript and pro- variation does not diVer signiWcantly among the four sub- vided information on Juncus cytology. F. Tod took care of groups and thus gives no clue to elucidating their glacial the living plants of J. biglumis in the Botanical Garden of history. Some phylogeographic inferences, however, can be the University of Vienna (HBV) in an excellent way. Per- made. (1) The arctic archipelago of Svalbard, separated mission to collect in NNR Ben Lawers was granted to P.B. from the Scandinavian mainland by c. 800 km, was obvi- Eidesen by the park authorities. C. Dixon improved the ously colonised independently by individuals representing grammar of the manuscript. The Austrian Science Fund both main clades A and B. The same situation holds true (Erwin–Schrödinger-Stipend to P. Schönswetter, J2311- for eastern Greenland. In S. oppositifolia, two widespread B03) Wnanced the stay of P. Schönswetter as a postdoctoral cpDNA haplotypes, A and B, were sampled on Svalbard fellow at the National Centre for Biosystematics at the 102 P. Schönswetter et al. / Molecular Phylogenetics and Evolution 42 (2007) 92–103
University of Oslo. Laboratory running costs were covered Haizel, T., Lim, Y.K., Leitch, A.R., Moore, G., 2005. Molecular analysis of by the Strategic University Programme Grant 146515/420 holocentric chromosomes of Luzula species. Cytogenet. Genome Res. from the Research Council of Norway to C. Brochmann 109, 134–143. Harriman, N.A., Redmond, D., 1976. Somatic chromosome numbers for (under the botany subprogram Migration and evolution of some North American species of Juncus L. Rhodora 78, 727–739. arctic plants in response to Quaternary climate changes). Hewitt, G.M., 2004. Genetic consequences of climatic oscillations in the Flow cytometric analyses were covered by projects sup- Quaternary. Phil. Trans. Roy. Soc. London B 359, 183–196. ported by the Ministry of Education of the Czech Republic Hultén, E., Fries, M., 1986. Atlas of North European vascular plants. Koe- W (MSM 0021620828) and the Academy of Sciences of the ltz Scienti c Books, Königstein, Germany. Johnson, A.W., Packer, J.G., 1968. Chromosome numbers in the Xora of Czech Republic (AV0Z6005908 and KSK6005114). Ogotoruk Creek. N.W. Alaska. Bot. Not. 121, 403–456. Kirschner, J., 2002. Juncaceae 2: Juncus subg. Juncus, species plantarum: References Flora of the world, part 7. Australian Biological Resource Study, Can- berra, Australia. Abbott, R.J., Brochmann, C., 2003. History and evolution of the arctic Knaben, G., Engelskjøn, T., 1967. Chromosome numbers of Scandinavian Xora: in the footsteps of Eric Hultén. Mol. Ecol. 11, 299–313. Arctic-Alpine plant species. II. Acta Bor. A. Sci. 21, 1–58. Abbott, R.J., Comes, H.P., 2004. Evolution in the Arctic: a phylogeo- Kuta, E., Bohanec, B., Dubas, E., Vizintin, L., Przywara, L., 2004. Chromo- graphic analysis of the circumarctic plant, Saxifraga oppositifolia (Pur- some and nuclear DNA study on Luzula–a genus with holokinetic ple saxifrage). New Phytol. 161, 211–224. chromosomes. Genome 47, 246–256. Abbott, R.J., Smith, L.C., Milne, R.I., Crawford, R.M.M., WolV, K., Bal- Levy, A.A., Feldman, M., 2004. Genetic and epigenetic reprogramming of four, J., 2000. Molecular analysis of plant migration and refugia in the the wheat genome upon allopolyploidization. Biol. J. Linnean Soc. 82, Arctic. Science 289, 1343–1346. 607–613. Aiken, S.G., Dallwitz, M.J., Consaul, L.L., McJannet, C.L., Gillespie, L.J., Matthews, J.V., Ovenden, L.E., 1990. Late Tertiary plant macrofossils Boles, R.L., Argus, G.W., Gillett, J.M., Scott, P.J., Elven, R., LeBlanc, from localities in arctic, sub-arctic North America–a review of the data. M.C., Brysting, A.K., Solstad, H., 1999 onwards. Flora of the Canadian Arctic 43, 364–392. Arctic Archipelago: Descriptions, Illustrations, IdentiWcation, and Mosquin, T., Hayley, D.E., 1966. Chromosome numbers and taxonomy of Information Retrieval. Version: 29th April 2003. Available at
Song, K., Lu, P., Tang, K., Osborn, T.C., 1995. Rapid genome change in Van de Peer, Y., De Wachter, R., 1997. Construction of evolutionary distance synthetic polyploids of Brassica and its implications for polyploid evo- trees with TREECON for Windows: accounting for variation in nucleo- lution. Proc. Natl. Acad. Sci. USA 92, 7719–7723. tide substitution rate among sites. Comp. Appl. Bioscience 13, 227–230. Suda, J., Kyncl, T., Wildová, R., 2003. Nuclear DNA amount in Macaro- Vierhapper, F., 1918. Juncus biglumis L. in den Alpen. Österr. Bot. Zeitschr. nesian angiosperms. Ann. Bot. 92, 153–164. 67, 49–51. Suda, J., Trávníbek, P., 2006. Reliable DNA ploidy determination in dehy- Weiss, H., Stuessy, T.F., Grau, J., Baeza, C.M., 2003. Chromosome reports drated tissues of vascular plants by DAPI Xow cytometry–new pros- from South American Hypochaeris (Asteraceae). Ann. Miss. Bot. Gard. pects for plant research. Cytometry, part A 69, 273–280. 90, 56–63. SwoVord, D. L. 2002. PAUP¤ Phylogenetic Analysis Using Parsimony Weiss-Schneeweiss, H., Stuessy, T.F., Siljak-Yakovlev, S., Baeza, C.M., (¤and Other Methods), Version 4.10. Sinauer Associates, Sunderland, Parker, J., 2003. Karyotype evolution in South American species of Massachusetts. Hypochaeris (Asteraceae, Lactuceae). Plant Syst. Evol. 241, 171–184. Taberlet, P., Gielly, L., Pautou, G., Bouvet, J., 1991. Universal primers for Whittall, J.B., Hellquis, C.B., Schneider, E.L., Hodges, S.A., 2004. Cryptic ampliWcation of three non-coding regions of chloroplast DNA. Plant species in an endangered pondweed community (Potamogeton, Pota- Mol. Biol. 17, 1105–1109. mogetonaceae) revealed by AFLP markers. Amer. J. Bot. 91, 2022–2029.