Vicariance and dispersal in the alpine perennial Bupleurum ......(Apiaceae) Peter Schönswetter 1,3...

INTRODUCTIONStudies exploring the phylogeography of disjunctly

distributed alpine arctic or arctic-alpine plant specieshave shown that the resulting patterns are often complexand not easily explained by either vicariance or dispersalevents Whereas in some taxa disjunctions appear to haveoriginated from dispersal (Hagen amp al 2001) otherscoincide with vicariance events (eg Kropf amp al 2003)or are evidenced by deep genetic splits through contigu-ous distribution areas (Schoumlnswetter amp al 2004)

The identification of long-distance dispersal as amajor factor in the generation of broad distribution areas(eg Cain amp al 2000 Abbott amp Brochmann 2003Brochmann amp al 2003) has challenged the traditionalassumption that large eg circumarctic distributions areancient (Valentine 1972) Even taxa that lack adapta-tions for dispersal over long distances were found to behighly mobile On a regional scale relatively recent(late-glacial to postglacial) long-distance dispersal has

been detected as much between mountain ranges(Schoumlnswetter amp al 2002) as within them (Tribsch amp al2002)

Vicariance in isolated Pleistocene refugia has beenshown to have led to the evolution of genetically separa-ble lineages (eg Hewitt 1996 2000 Tribsch ampSchoumlnswetter 2003) in most of the taxa investigated upto now (reviewed eg in Taberlet amp al 1998 Comes ampKadereit 1998 Widmer amp Lexer 2001) When the cli-mate ameliorated at the end of the Pleistocene the refu-gial populations acted as sources for the recolonisation ofappropriate habitats Whereas some taxa managed toobtain a wide and contiguous distribution others stillshow disjunctions due to incomplete remigration In phy-logeographic investigations of Alpine plants both situa-tions have been detected even within the same studytaxa (eg Tribsch amp al 2002)

Our target species Bupleurum stellatum L (Apia-ceae Fig 1) is a perennial herb forming dense tussocksIt is diploid with 2n = 14 (Cauwet 1967 Favarger amp

725

Schoumlnswetter amp Tribsch bull Bupleurum stellatum54 (3) bull August 2005 725ndash732

PHYLOGEOGRAPHY

Vicariance and dispersal in the alpine perennial Bupleurum stellatum L(Apiaceae)

Peter Schoumlnswetter13 amp Andreas Tribsch23

1 Department of Biogeography Institute of Botany University of Vienna Rennweg 14 A-1030 Vienna Austriapeterschoenswetterunivieacat (author for correspondence)

2 Department of Systematic and Evolutionary Botany Institute of Botany University of Vienna Rennweg 14A-1030 Vienna Austria andreastribschnhmuiono

3 NCB ndash National Centre for Biosystematics The Natural History Museums and Botanical Garden Universityof Oslo PO Box 1172 Blindern N-0318 Oslo Norway

Using Amplified Fragment Length Polymorphism (AFLP) we explored the intraspecific phylogeography ofthe alpine vascular plant Bupleurum stellatum (Apiaceae) disjunctly distributed in the Alps and CorsicaWithin the Alps the speciesrsquo distribution area is also not contiguous spatially isolated groups of populationsoccuring in the Dolomites (Italy) and the Montafon (Austria) A main goal of our study was to explore thedegree of differentiation of the isolated populations in Corsica the Dolomites and the Montafon from the pop-ulations in the main distributional area and to test for hypotheses of vicariance or dispersal We also sought abetter general knowledge of glacial refugia of low alpine silicicolous plants Phenetic as well as cladistic analy-ses of the AFLP multilocus phenotypes reveal a substantial north-south split through the contiguous distribu-tion area of B stellatum in the Alps The resulting phylogeographic groups can be related to glacial refugia inperipheral areas of the Alps that were recognised in previous studies The disjunct populations in Corsica theDolomites and the Montafon however are less strongly differentiated For example the Corsican populationclustered with high bootstrap support with populations from the Eastern Alps suggesting immigration toCorsica from that region Our study shows that deep phylogeographic splits resulting from old vicarianceevents can be concealed by presently contiguous distribution areas In contrast disjunctions if they are due todispersal events need not be accompanied by genetic divergence

KEYWORDS AFLP Alps Bupleurum stellatum Corsica dispersal phylogeography Pleistocene refugia vic-ariance

Kuumlpfer 1968) Favargerrsquos (1954) count of 2n = 16 waserroneous (Favarger amp Kuumlpfer 1968) Bupleurum stella-tum is an insect-pollinated outcrosser the flowers arestrongly proterandric and the umbels shift from anexclusively male to an exclusively female stage (Hegi1925) Main habitats are open grassland communitiescovering dry sunny rocky outcrops or stabilised screefields from 1700 up to 2650 (2800) m (Hegi 1925Pignatti 1983) The species is confined to siliceousmountain ranges of the Alps mainly south of the maindivide and to Corsica Within the Alps B stellatum isrestricted to the Western Alps and to the southern part of

the middle Alps It is most frequent in the middle part ofits distribution area and becomes rare towards the south-west and the east (Pignatti 1983) There are disjunctoccurrences in the southwestern Dolomites (CimadrsquoAsta Lagorai) that provide an island of alpine vegeta-tion on silicates almost completely surrounded by lime-stone and in the Montafon (AustriaSwitzerland) wherethere is a group of populations that is isolated in spite ofa lack of obvious barriers The peculiar distribution pat-tern of Bupleurum stellatum L (Apiaceae) offers the pos-sibility for exploring patterns of vicariance and dispersalin a disjunctly distributed alpine plant (Fig 2) and allowsus to test the hypothesis that the phylogeographic patterncoincides with disjunctions

Thus a major goal of this study is to explore the his-tory of the disjunct populations of B stellatum in Cor-sica the Dolomites and the Montafon As in other phylo-geographic studies based on molecular fingerprintingdata old vicariance is expected to result in significantgenetic divergence whereas recent dispersal might leadto genetic depauperation due to founder effects but to lit-tle or no genetic divergence Apart from this major goalour study also contributes to a better understanding ofglacial refugia of silicicolous plants at the southern bor-der of the Alps (Tribsch amp Schoumlnswetter 2003)

MATERIALS AND METHODSSampling mdash We sampled 23 populations of B

stellatum from the Alps (populations 2ndash24) and one pop-ulation from Corsica (population 1 Fig 2) The gapbetween populations 8 and 9 (Fig 2) is a sampling arte-fact and does not reflect a break in the distribution Fresh

Schoumlnswetter amp Tribsch bull Bupleurum stellatum 54 (3) bull August 2005 725ndash732

726

Fig 1 Bupleurum stellatum in its natural habitat nearPasso Croce Domini (northern Italy population 19)

Fig 2 Distribution (shaded in lightgrey) and sampled populations ofBupleurum stellatum in the EuropeanAlps and in Corsica along with the max-imum extent of the Pleistocene iceshield during the last glacial period(broken line taken from Jaumlckli 1970Van Husen 1987 Voges 1995) Thearrow indicates the position of the gla-cier tongue in Val Camonica (see text)The symbols refer to population groupsidentified in the present study (+ W1Q W2 ∆∆ E+Co excluding population 1from Corsica) or to disjunct populationgroups ( population 1 from Corsica population 22 from Montafon gtpopulations 23 and 24 from theDolomites) Bottom right insert givesthe location of population 1 fromCorsica and indicates the position ofthe sampling area in Europe Popula-tion numbers refer to Table 1

1

2

6

9

10

12

14

16 17

18

19

22

23 2412

16

19

1

50 kmN

leaves were collected in the field and immediately storedin silica gel Details of the sampling locations as well asthe number of sampled individuals per population (fouror five) are given in Table 1 Voucher specimens for allpopulations are deposited in the herbarium of theInstitute of Botany of the University of Vienna (WU)

DNA isolation and AFLP fingerprinting mdashProtocols for DNA isolation and for AFLP fingerprintingare described in Schoumlnswetter amp al (2002 2004) For theselective amplification we used the fluorescence-labelled primer combinations EcoRI ACC (NED)-MseICAT EcoRI ACT (6-Fam)-MseI CTC and EcoRI AGG(HEX)-MseI CAA The selective amplification-productswere separated on a 5 polyacrylamide gel with aninternal size standard (GeneScan-500 [ROX] PEApplied Biosystems Foster City California USA) onan automated sequencer (ABI 377 PE AppliedBiosystems) Raw data were collected and aligned withthe internal size standard using the ABI Prism GeneScanAnalysis Software (PE Applied Biosystems) Sub-sequently the GeneScan sample files were imported intoGenographer (ver 160 copy Montana State University1999 httphordeummsumontanaedugenographer)for scoring of the fragments The results of the scoringwere exported as a presenceabsence matrix and used forfurther analysis

Data analysis mdash As a measure of within-popula-tion diversity the number of AFLP fragments (Fragtot)and the percentage of polymorphic fragments (poly)were calculated for all populations The number ofunique fragments (Fraguni) was calculated as a measureof divergence In order to detect divergent populationswith many rare markers we avoided applying a subjec-tive threshold defining the notion of ldquorarerdquo (eg markerspresent in lt 10 of the investigated individuals Stehlikamp al 2002 or in less than a certain number of individu-als Tribsch amp al 2002) We calculated an additionalmeasure of divergence ie ldquofrequency-down-weightedmarker valuesrdquo (DW) equivalent to range-down-weight-ed species values in historical biogeographical research(Crisp amp al 2001) For each population the number ofoccurrences of each AFLP marker in that population wasdivided by the number of occurrences of that particularmarker in the total dataset Finally these values weresummed up The value of DW is expected to be high inlong-term isolated populations where rare markersshould accumulate due to mutations whereas newlyestablished populations are expected to exhibit low val-ues thus helping in distinguishing old vicariance fromrecent dispersal To even out the unequal sample sizesthe genetic diversity measures and DW were calculatedwith just four randomly chosen individuals We per-


727

Table 1 Numbering of populations location country (A = Austria CH = Switzerland F = France I = Italy) co-ordinatesnumber of investigated individuals (N) number of AFLP fragments (Fragtot) per population percentage of fragmentswhich exhibit intrapopulational polymorphism (poly) number of fragments that only occur in one population (Fraguni)and frequency-down-weighted marker values (DW) of the 24 investigated populations of Bupleurum stellatum L Toeven out the unequal sample sizes the genetic diversity measures Fragtot poly and the divergence measures Fraguniand DW were calculated with only four randomly chosen individuals

Location Country Co-ordinates (EN) N Fragtot poly Fraguni DW1 Corsica Monte drsquoOro F 897 4213 4 133 3008 7 14872 Pian di Re I 711 4470 5 138 3623 2 13933 Punta Cialancia I 712 4487 5 148 4730 1 12574 Monte Palon I 713 4520 5 133 3459 2 11805 Val Savaranche I 720 4552 5 141 3617 3 12796 Le Brevent F 685 4593 5 152 4145 4 16657 Monte Nery I 773 4575 5 163 4847 5 18168 Vallone di Mos I 785 4585 5 149 4027 5 16129 Cima dellrsquoUomo CH 895 4622 4 153 4183 2 137510 Pizzo di Gino I 913 4612 5 152 4145 1 110311 Monte Legnone I 942 4608 5 141 3617 0 88512 Monte Spluga I 955 4618 5 149 3758 1 111613 Valle Vicime I 972 4610 5 146 4178 1 102914 Cima Cadelle I 973 4605 5 151 4040 0 105315 Bivacco Resnati I 1000 4607 5 145 4069 0 95516 Bochetta delle Forbici I 990 4632 5 138 3623 1 101217 Monte Verva I 1022 4642 4 152 3684 0 101618 Monte Colombine I 1036 4585 4 148 3716 1 100419 Passo Croce Domini I 1043 4593 5 144 3750 1 101520 Passo delle Toppette I 1060 4617 5 146 3767 0 99321 Passo Gavia I 1047 4633 5 147 3129 3 124322 Valisera A 993 4695 4 142 3732 3 126623 Monte Ziolera I 1145 4617 5 139 3525 3 122724 Cima drsquoAsta I 1160 4618 5 127 2992 0 719

formed a maximum parsimony analysis with PAUP 40b10 (Swofford 1998) Heuristic searches were doneusing a starting tree obtained by neighbour-joining andswapped to completion using TBR branch swapping andMULTREES on (keeping multiple shortest trees) Astrict consensus tree of the 160 equally most parsimo-nious trees was constructed Robustness of clades wasestimated using the bootstrap approach (Felsenstein1985) with 500 replicates with random sequence addition(10 replicates) saving no more than 500 trees per repli-cate Although the application of parsimony methods forfingerprinting data is not undisputed empirical evidencehas shown that with increasing amounts of AFLP data allcommonly used methods of reconstructing relationshipswith fragment data converge on the same tree and theseestimates are robust (Beardsley amp al 2003) To under-line this we also performed a neighbour-joining analysiswith TREECON 13b (Van de Peer amp De Wachter 1997)based on Nei amp Lirsquos (1979) genetic distance The robust-ness of the branches was estimated with 500 bootstrapreplicates

A principal coordinate analysis (PCoA) based oninter-individual Jaccard similarities (CJ = a [a + b + c]where a is the number of fragments shared between twoindividuals and b and c are the numbers of fragmentspresent in only one individual) was calculated and plot-ted with SPSS 1007 (Norusis 1999) Analyses ofmolecular variance (AMOVAs) were computed withARLEQUIN 11 (Schneider amp al 1997)

RESULTSWith the three primer combinations used 287 unam-

biguously scorable AFLP fragments were generated 46(160) of which were monomorphic The length of thefragments varied from 52 to 502 bp All individuals haddifferent multilocus genotypes The Fragtot variedbetween 127 in population 24 and 163 in population 7(mean 1449 SD = 78 Table 1) poly ranged from 299in population 24 to 485 in population 7 (mean 381 SD= 45 Table 1) and Fraguni varied between zero in pop-ulations 11 14 15 17 20 and 24 and seven in popula-tion 1 (mean 192 SD = 189 Table 1) The DW rangedfrom 719 in population 24 to 1816 in population 7

(mean 1196 SD = 262 Table 1)The maximum parsimony analysis (Fig 3) revealed

the following structure the population from Corsica(pop 1 = group Co) formed a highly supported group(bootstrap support 97) It clustered with populations 17to 21 23 and 24 from the Eastern Alps (group E) againwith very high support (bootstrap support 97) Thisgroup is henceforth referred to as E+Co

The neighbour-joining analysis (Fig 4) revealedessentially the same structure as the parsimony analysisIt differs mainly in that population 1 from Corsica clus-ters with high support (bootstrap support 100) with pop-ulations 17 to 21 from the Eastern Alps and populations23 and 24 from the Dolomites form a moderately (boot-strap support 70) supported sister group to that branchFurthermore populations 2 to 5 (except for one individ-ual of population 3) from the southwestern Alps formone unsupported branch instead of two in the parsimonyanalysis

The clustering of population 1 from Corsica is alsosupported by the direct comparison of the distribution ofAFLP markers all markers of that population except forits seven private fragments (Table 1) are shared with pop-ulations of E This is not the case for W in spite of thehigher number of populations in that group From the287 scored fragments 81 markers were private to W 19to E+Co and 187 were found in both population groups

The PCoA (Fig 5) largely confirmed the patterndetected by the maximum parsimony analysis Ad-ditionally the first factor (explaining 272 of the totalvariation) revealed the internal structure of W ie theseparation between a southwestern group W1 (popula-tions 2ndash5) and the remaining populations W2 (popula-tions 6ndash16) The individuals of population 22 fromMontafon fall into group W2 Along the second factorE+Co (240) is separated from W Whereas the indi-viduals from the Dolomites (populations 23 24) groupwith the other accessions from E the Corsican individu-als (population 1) are somewhat separated The third fac-tor (136) separates nearly all individuals of W1 fromthe rest of the dataset From the fourth factor onwardssingle populations are separated such as population 6along factor four (31)

Non-hierarchical AMOVAs (Table 2) assigned386 of the overall genetic variation to variation among


728

Table 2 Analysis of molecular variance (AMOVA) of the 24 investigated populations of Bupleurum stellatum LSource of variation df Sum of squares Variance components Total variance FSTAmong populations 1ndash24 23 140296 955 3858 039Within populations 91 138390 1521 6142

Among E and W 1 37194 648 2292 046Among populations 22 103006 660 2332Within populations 91 138390 1521 5376All P-values were lt 0001

the 24 populations Hierarchical AMOVAs of the Alpinepopulations attributed 229 of the overall variation tovariation among E and W 233 to variation among thepopulations within the groups and 538 to variationwithin populations

DISCUSSIONVicariance within the continuous distribu-

tion area of B stellatum in the Alps mdash The Alpinepopulations of Bupleurum stellatum fall into two clearlydivergent groups that are both characterised by a highnumber of private fragments and high bootstrap support(Figs 3ndash5) Group W comprises populations from thewestern part of the Alps eastwards to the Bernina moun-tain range (Italy Switzerland) and the Alpi Orobie (Italy)plus population 22 from the Montafon Group E+Coencompasses populations from the easterly adjacentmountain ranges along with population 1 from CorsicaThere appears to be no break in the distribution areabetween groups W and E+Co As shown by the strong

separation of individuals from east and west of the mainphylogeographic split (groups W2 and E in the PCoAFig 5) a possible hybrid zone was not sampled The splitbetween W and E+Co explains approximately a quarterof the overall genetic variation (Table 2) This compara-tively low value (see eg Schoumlnswetter amp al 2002 andTribsch amp al 2002 for comparison) relates to the highintrapopulational genetic variation accounting for gt 60of the overall variation (Table 2 see also the diversityindices in Table 1) Moreover this supports the assump-tion that the species is strongly outbreeding (Hegi 1925)In other studies on alpine plants however much lowervalues for variation among groups were detected (egHolderegger amp al 2002 Kropf amp al 2003)

The differentiation between W and E+Co that can-not be paralleled with any taxonomical intraspecificvariation (summarized in Hegi 1925) is most likely dueto a vicariance event ie the disruption of a formerlycontinuous distribution area and survival in disjunct gla-cial refugia This split appears to be older than the lastglaciation as group W is internally structured as revealedby the neighbour-joining analysis (Fig 4) and the PCoA(Fig 5) It falls into two subgroups W1 south of the ValledrsquoAosta and W2 north and east of it a phylogeographicpattern similar to those observed in Androsace alpinaLam (Schoumlnswetter amp al 2003) and less clearly inRanunculus glacialis L (Schoumlnswetter amp al 2004) Thesubgroups of W and the Alpine populations of groupE+Co overlap with presumed refugia along the southernmargin of the Alps (see Fig 2 in Schoumlnswetter amp al2002) areas that have been less affected by the glacia-tions than more interior parts (reviewed in Tribsch ampSchoumlnswetter 2003a) W1 overlaps with the hypotheticalrefugia in the eastern Cottic and the eastern Grajic AlpsW2 with the southern Penninic Alps and the AlpiBergamasche E+Co covers the southwestern Dolomitesand the northern Alpi Giudicarie Important barriers forthe populations in the refugia south of the ice sheet dur-ing the Last Glacial Maximum and probably also duringearlier glaciations might have been the large glaciertongue protruding from Val Camonica (to the west ofpopulations 18 and 19 Jaumlckli 1970 indicated by anarrow in Fig 1) and carbonate mountain ranges at theperiphery of the Alps that offered no potential growingsites for the strongly acidophilic B stellatum At presentthere are no obvious ecological factors that maintain theobserved genetic differentiation as the area between pop-ulations 16 and 17 (Fig 2) provides more or less contin-uous suitable habitats In conclusion due to the unam-biguous separation of only two large population groupsand the congruence of the split between W1 and W2 withprevious studies the results from the phylogeographicanalysis of B stellatum do not allow a finer separation ofrefugia in the Alps compared to previous studies but


729

Fig 3 Unrooted maximum parsimony network based onAFLP data of 115 individuals of Bupleurum stellatumfrom the Alps and Corsica The bold numbers above thebranches are bootstrap values above 50 (500 replicates)To enforce legibility they are given only for brancheswith at least 4 individuals The population numbers aregiven at the tips of the branches Symbols as in Fig 2

5 changes15

14

15

7

19 1818

3

7

8

8

88

8

10129

13

1113

10

10

10

11

11

9

9

9

14

16

1616

16

1216

15

1515

12

12

131413

13

1114

10

1112143

22

22

22

22

7

5

555

7

7

6

6

6

66

1

1

1

1

1919

19

20

20

20

20

21

21

21

2121

19

18

18

17

17 17

17

23

24

2424 24

24 2323

23

23

22

22

2 44

5

44

4

3

3

3

97

68

52

56

97

rather corroborate already known patternsThe populations in the Montafon and the

Dolomites are not clearly divergent mdash Ourresults indicate a close relationship of the population inMontafon to group W and that of the populations in theDolomites to E+Co (Figs 3 to 5) Although the diversityand divergence measures presented in this paper shouldbe treated with caution due to the low number of investi-gated individuals per population the level of genetic dif-ferentiation and diversity of the Montafon and Dolomitespopulations are not above average (DW poly Table 1)The lack of strong divergence resulting from long-termisolation fits well to the very strong glaciation of theMontafon during the Last Glacial Maximum (LGM vanHusen 1987) a priori excluding long-term survival ofthe low-alpine B stellatum on nunataks in that region

The southwestern Dolomites in contrast had alreadybeen identified as a glacial refugium in previous studies(Schoumlnswetter amp al 2002 2003b) Although glacial sur-vival of populations 23 and 24 in that region could hypo-thetically have been possible the two investigated popu-lations from that area do not appear to be strongly diver-gent (Table 1) and their moderate bootstrap support in theneighbour-joining analysis (Fig 4) but not in the maxi-mum parsimony analysis (Fig 3) could well be explainedby their low genetic diversity Thus for the isolatedgroups of populations in the Montafon and probably alsoin the Dolomites recent late or postglacial dispersal is apossible scenario

Dispersal from the Eastern Alps to Corsicamdash Population 1 from Corsica originated via dispersalfrom source populations in the Eastern Alps It groupswith Eastern Alpine populations with high bootstrap sup-port (Figs 3 4) and shares all of its markers except forseven private fragments with the other populations ofgroup E+Co In contrast to the maximum parsimonyanalysis (Fig 3) that provides no resolution in thisrespect the neighbour-joining analysis (Fig 4) suggestsa highly supported sister-relationship of the Corsicanpopulation to populations 17 to 21 and not to the popu-lations from the Dolomites The differentiation of theCorsican and the Alpine populations can hardly be relat-ed to the ldquoMessinian eventrdquo the desiccation of theMediterranean sea after the closure of the straight ofGibraltar in the late Miocene (c 59 mya eg Butler ampal 1999) for three reasons (1) although it is impossibleto date genetic fingerprinting data we expect separationsince the Tertiary to result in much stronger genetic dif-ferentiation (2) migration of an alpine plant through for-mer ocean-floor during a comparatively warm time peri-od appears highly unlikely (3) the Corsican populationsrelate to the Eastern Alpine populations instead of thegeographically more close populations of the WesternAlpine group W suggesting long-distance dispersalrather than continuous migration However above-aver-age level of divergence as expressed with DW (Table 1)and the presence of several private fragments some ofwhich are fixed (Table 1) indicate that the colonisationof Corsica from the Alps is not due to a Holocene dis-persal event as argued eg for the westernmost Alpinepopulations of Saponaria pumila Janchen (Tribsch amp al2002) but rather dates back to the Pleistocene

The alpine flora of Corsica is poor in otherwiseexclusively Alpine taxa Out of 2518 taxa native toCorsica of which 21 are regarded as oreophilous Centraland Southern European elements (Contandriopoulos1962) Gamisans (1991) lists only two species that areexclusively shared by Corsica and the Alps One of themis our study taxon B stellatum and the second is Violanummularifolia All (Violaceae) that in the Alps is


730

Fig 4 Neighbour-joining tree based on AFLP data of 115individuals of Bupleurum stellatum from the Alps andCorsica The bold numbers above the branches are boot-strap values above 50 (500 replicates) As in Fig 3 theyare given only for the major branches The populationnumbers are given at the tips of the branches Symbolsas in Fig 2

1616

1213

1113

1111

1110

1010

1313

1414

1414

1115

1212

1315

1515

22

55

55

33

33

2222

2222

36

6

7

666

77

78

88

88

7

9

1210

912

149

915

101616

16

2020

2020

1819

181918

1919

1918

171717

17 212121

2121

11

11

2323 23

2323

2424

2424

244

445

44

22

2

58

70

56

76

100

48

70

96

96

restricted to the southwesternmost part (AlpesMaritimes) geographically closest to Corsica Addition-ally Ranunculus kuepferi (Ranunculaceae) occurs bothin Corsica and in the Alps Its diploid sexual cytodemesubsp kuepferi is restricted to the Alpes Maritimeswhereas the tetraploid apomictic subsp orientalis ismore widespread from the Alpes Maritimes to HoheTauern (Huber 1988) and occurs locally in Corsica(Huber 1989) Since the polyploid apomict is most like-ly a derivative of subsp kuepferi dispersal has probablygone in parallel both in southerly and northerly directionA further taxon that illustrates Corsicarsquos floristic linkswith the Alps or Carpathians is Alnus alnobetula C Koch(syn Alnus viridis DC Betulaceae) with subspeciessuaveolens (Req) J Lambinon amp M Kergueacutelen endem-ic to Corsica and the type subspecies distributed through-out the Alps and major parts of the Carpathians A paral-lel case to B stellatum was found in a phylogenetic studyof the genus Phyteuma (G Schneeweiss P SchoumlnswetterA Tribsch unpubl) where sequences of nuclear andchloroplast markers as well as AFLP data suggest a high-ly supported sister relationship of the Corsican endemicP serratum Viv with the East Alpine Carpathian P con-fusum Kern Further studies covering alpine taxa sharedby Corsica and the Eastern Alps region should clarifywhether the unexpected biogeographic connectiondetected in our study taxon B stellatum is an idiosyn-crasy of the investigated plant species or indicates a moregeneral biogeographic pattern

ACKNOWLEDGEMENTSFunding by the Austrian Science Foundation (FWF P13874-

Bio) is gratefully acknowledged We thank Michael Barfuszlig forefficient technical support in the lab Two anonymous reviewersand Elvira Houmlrandl provided many helpful comments Alice Luckimproved the grammar in previous versions of the manuscriptPhilippe Kuumlpfer provided information on the chromosome countsof B stellatum Special thanks also go to Christoph Dobeš for col-lecting the samples from Corsica and to Gerald M Schneeweissand Magdalena Wiedermann who accompanied us during many ofour collection trips

LITERATURE CITEDAbbott R J amp Brochmann C 2003 History and evolution

of the arctic flora in the footsteps of Eric Hulteacuten MolecEcol 11 299ndash313

Beardsley P M Yen A amp Olmstead R G 2003 AFLP phy-logeny of Mimulus section Erythranthe and the evolutionof hummingbird pollination Evolution 57 1397ndash1410

Brochmann C Gabrielsen T M Nordal I Landvik JY amp Elven R 2003 Glacial survival or tabula rasa Thehistory of North Atlantik biota revisited Taxon 52417ndash450

Butler R W H McClelland E amp Jones R E 1999Calibrating the duration and timing of the MessinianSalinity Crisis in the Mediterranean linked tectono-cli-matic signals in thrust-top basin of Sicily J Geol SocLondon 156 827ndash835

Cain M L Milligan B G amp Strand A E 2000 Long-dis-


731

Fig 5 Principal Coordinate Analysis (first four factors cumulatively explaining 679 of the total variation) based onAFLP data of 115 individuals of Bupleurum stellatum from the Alps and Corsica Symbols as in Fig 2

8765432

8

7

6

5

4

3

2

Pop 22 Montafon

Pop 6-16 W2

Pop 2-5 W1

Pop 17-21 E

Pop 23 24 Dolomites

Pop 1 Corsica

Facto

r 2

Factor 187654321

7

6

5

4

3

2

1

00

Facto

r 4

Factor 3

tance seed dispersal in plant populations Amer J Bot 871217ndash1227

Cauwet A M 1967 Contribution agrave lrsquoeacutetude caryosysteacutematiquedu genre Bupleurum L I Bull Soc Bot France 114371ndash386

Comes H P amp Kadereit J W 1998 The effect ofQuaternary climatic changes on plant distribution and evo-lution Trends Plant Sci 3 432ndash438

Contandriopoulos J 1962 Recherches sur la floreendeacutemique de la Corse et sur ses origines Ann Fac SciMarseille 32 1ndash354

Crisp M D Laffan S Linder H P amp Monro A 2001Endemism in the Australian flora J Biogeogr 28183ndash198

Favarger C 1954 Sur la pourcentage des polyploides dans laflore de lrsquoeacutetage nival des Alpes Suisses Huitiegraveme CongrIntern Bot Paris C R des Seacuteances Sect 9 51ndash56

Favarger C amp Kuumlpfer P 1968 Contribution agrave la cytotax-onomie de la flore alpine des Pyreacuteneacutees Collect Bot 7325-357

Felsenstein J 1985 Confidence limits on phylogenies anapproach using the bootstrap Evolution 39 783ndash791

Gamisans J 1991 La veacutegeacutetation de la Corse Pp 1ndash391 inJeanmonod D amp Burdet H M (eds) Compleacutements auPodrome de la Flore Corse Editions des Conservatoire etJardin Botanique de la Ville de Genegraveve Genegraveve

Hagen A R Giese H amp Brochmann C 2001 Trans-Atlantic dispersal and phylogeography of Cerastiumarcticum (Caryophyllaceae) inferred from RAPD andSCAR markers Amer J Bot 88 103ndash112

Hegi G 1925 Illustrierte Flora von Mitteleuropa vol 2 J FLehmann Muumlnchen

Hewitt G M 1996 The genetic legacy of the Quaternary iceages Nature 405 907ndash913

Hewitt G M 2000 Some genetic consequences of ice agesand their role in divergence and speciation Biol JLinnean Soc 58 247ndash276

Holderegger R Stehlik I amp Abbott R J 2002 Molecularanalysis of the Pleistocene history of Saxifraga oppositifo-lia in the Alps Molec Ecol 11 1409ndash1418

Huber W 1988 Natuumlrliche Bastardierungen zwischenweiszligbluumlhenden Ranunculus-Arten in den Alpen (Naturalhybridizations between white-flowered species ofRanunculus in the Alps) Veroumlff Geobot Inst ETHStiftung Ruumlbel Zuumlrich 100 1ndash160 [German with Englishabstract]

Huber W 1989 Ranunculus kuepferi Greuter amp Burdet inKorsica (Gruppe R pyrenaeus L) Candollea 44630ndash637

Jaumlckli H 1970 Die Schweiz zur letzten Eiszeit (1550000)In Eidgenoumlssische Landestopographie (ed) Atlas derSchweiz Eidgenoumlssische Landestopographie Wabern-Bern

Kropf M Kadereit J W amp Comes H P 2003 Differentialcycles of range contraction and expansion in Europeanhigh mountain plants during the Late Quaternary insightsfrom Pritzelago alpina (L) O Kuntze (Brassicaceae)Molec Ecol 12 931ndash949

Nei M amp Li W H 1979 Mathematical model for studyinggenetic variation in terms of restriction endonucleasesProc Natl Acad Sci USA 76 5269ndash5273

Norusis M J 1999 SPSS 90 Guide to Data Analysis

Prentice Hall Englewood CliffsPignatti S 1983 Flora drsquoItalia Edagricole BolognaSchneider S Kueffer J M Roessli D amp Excoffier L

1997 Arlequin 11 A Software for Population GeneticAnalysis Genetics and Biometry Laboratory Universityof Geneva Geneva

Schoumlnswetter P Tribsch A Barfuss M amp Niklfeld H2002 Several Pleistocene refugia detected in the highalpine plant Phyteuma globulariifolium in the EuropeanAlps Molec Ecol 11 2637ndash2647

Schoumlnswetter P Tribsch A amp Niklfeld H 2003aPhylogeography of the high alpine cushion-plantAndrosace alpina (Primulaceae) in the European Alps PlBiol 5 623ndash630

Schoumlnswetter P Tribsch A Schneeweiss G M ampNiklfeld H 2003b Disjunctions in relict alpine plantsPhylogeography of Androsace brevis and A wulfeniana(Primulaceae) Bot J Linn Soc 141 437ndash446

Schoumlnswetter P Tribsch A Stehlik I amp Niklfeld H 2004Glacial history of high alpine Ranunculus glacialis(Ranunculaceae) in the European Alps in a comparativephylogeographical context Biol J Linn Soc 81183ndash195

Stehlik I Schneller J J amp Bachmann K 2002 Im-migration and in situ glacial survival of the low-alpineErinus alpinus (Scrophulariaceae) Biol J Linn Soc 7787ndash103

Swofford D L 1998 PAUP Phylogenetic Analysis UsingParsimony (and Other Methods) version 4 SinauerAssociates Sunderland Massachusetts

Taberlet P Fumagalli L amp Cosson-Anne G W 1998Comparative phylogeography and postglacial colonizationroutes in Europe Molec Ecol 7 453ndash464

Tribsch A amp Schoumlnswetter P 2003 In search for Pleistocenerefugia for mountain plants patterns of endemism andcomparative phylogeography confirm palaeo-environmen-tal evidence in the Eastern European Alps Taxon 52477ndash497

Tribsch A Schoumlnswetter P amp Stuessy T F 2002Saponaria pumila (Caryophyllaceae) and the ice-age inthe Eastern Alps Amer J Bot 89 2024ndash2033

Valentine D V 1972 Patterns of geographical distribution inspecies of the European flora Symp Biol Hung 1215ndash21

Van de Peer Y amp De Wachter R 1997 Construction of evo-lutionary distance trees with TREECON for Windowsaccounting for variation in nucleotide substitution rateamong sites Comp Appl Bioscience 13 227ndash230

Van Husen D 1987 Die Ostalpen in den EiszeitenGeologische Bundesanstalt Wien

Voges A 1995 International Quaternary Map of Europe B10Bern Bundesanstalt fuumlr Geowissenschaften undRohstoffeUNESCO Hannover

Widmer A amp Lexer C 2001 Glacial refugia sanctuaries forallelic richness but not for gene diversity Trends EcolEvol 16 267ndash269


732

Kuumlpfer 1968) Favargerrsquos (1954) count of 2n = 16 waserroneous (Favarger amp Kuumlpfer 1968) Bupleurum stella-tum is an insect-pollinated outcrosser the flowers arestrongly proterandric and the umbels shift from anexclusively male to an exclusively female stage (Hegi1925) Main habitats are open grassland communitiescovering dry sunny rocky outcrops or stabilised screefields from 1700 up to 2650 (2800) m (Hegi 1925Pignatti 1983) The species is confined to siliceousmountain ranges of the Alps mainly south of the maindivide and to Corsica Within the Alps B stellatum isrestricted to the Western Alps and to the southern part of

the middle Alps It is most frequent in the middle part ofits distribution area and becomes rare towards the south-west and the east (Pignatti 1983) There are disjunctoccurrences in the southwestern Dolomites (CimadrsquoAsta Lagorai) that provide an island of alpine vegeta-tion on silicates almost completely surrounded by lime-stone and in the Montafon (AustriaSwitzerland) wherethere is a group of populations that is isolated in spite ofa lack of obvious barriers The peculiar distribution pat-tern of Bupleurum stellatum L (Apiaceae) offers the pos-sibility for exploring patterns of vicariance and dispersalin a disjunctly distributed alpine plant (Fig 2) and allowsus to test the hypothesis that the phylogeographic patterncoincides with disjunctions

Thus a major goal of this study is to explore the his-tory of the disjunct populations of B stellatum in Cor-sica the Dolomites and the Montafon As in other phylo-geographic studies based on molecular fingerprintingdata old vicariance is expected to result in significantgenetic divergence whereas recent dispersal might leadto genetic depauperation due to founder effects but to lit-tle or no genetic divergence Apart from this major goalour study also contributes to a better understanding ofglacial refugia of silicicolous plants at the southern bor-der of the Alps (Tribsch amp Schoumlnswetter 2003)

MATERIALS AND METHODSSampling mdash We sampled 23 populations of B

stellatum from the Alps (populations 2ndash24) and one pop-ulation from Corsica (population 1 Fig 2) The gapbetween populations 8 and 9 (Fig 2) is a sampling arte-fact and does not reflect a break in the distribution Fresh


726

Fig 1 Bupleurum stellatum in its natural habitat nearPasso Croce Domini (northern Italy population 19)

Fig 2 Distribution (shaded in lightgrey) and sampled populations ofBupleurum stellatum in the EuropeanAlps and in Corsica along with the max-imum extent of the Pleistocene iceshield during the last glacial period(broken line taken from Jaumlckli 1970Van Husen 1987 Voges 1995) Thearrow indicates the position of the gla-cier tongue in Val Camonica (see text)The symbols refer to population groupsidentified in the present study (+ W1Q W2 ∆∆ E+Co excluding population 1from Corsica) or to disjunct populationgroups ( population 1 from Corsica population 22 from Montafon gtpopulations 23 and 24 from theDolomites) Bottom right insert givesthe location of population 1 fromCorsica and indicates the position ofthe sampling area in Europe Popula-tion numbers refer to Table 1

1

2

6

9

10

12

14

16 17

18

19

22

23 2412

16

19

1

50 kmN





727














728









729


5 changes15

14

15

7

19 1818

3

7

8

8

88

8

10129

13

1113

10

10

10

11

11

9

9

9

14

16

1616

16

1216

15

1515

12

12

131413

13

1114

10

1112143

22

22

22

22

7

5

555

7

7

6

6

6

66

1

1

1

1

1919

19

20

20

20

20

21

21

21

2121

19

18

18

17

17 17

17

23

24

2424 24

24 2323

23

23

22

22

2 44

5

44

4

3

3

3

97

68

52

56

97







730


1616

1213

1113

1111

1110

1010

1313

1414

1414

1115

1212

1315

1515

22

55

55

33

33

2222

2222

36

6

7

666

77

78

88

88

7

9

1210

912

149

915

101616

16

2020

2020

1819

181918

1919

1918

171717

17 212121

2121

11

11

2323 23

2323

2424

2424

244

445

44

22

2

58

70

56

76

100

48

70

96

96











731


8765432

8

7

6

5

4

3

2

Pop 22 Montafon

Pop 6-16 W2

Pop 2-5 W1

Pop 17-21 E

Pop 23 24 Dolomites

Pop 1 Corsica

Facto

r 2

Factor 187654321

7

6

5

4

3

2

1

00

Facto

r 4

Factor 3






































732





727














728









729


5 changes15

14

15

7

19 1818

3

7

8

8

88

8

10129

13

1113

10

10

10

11

11

9

9

9

14

16

1616

16

1216

15

1515

12

12

131413

13

1114

10

1112143

22

22

22

22

7

5

555

7

7

6

6

6

66

1

1

1

1

1919

19

20

20

20

20

21

21

21

2121

19

18

18

17

17 17

17

23

24

2424 24

24 2323

23

23

22

22

2 44

5

44

4

3

3

3

97

68

52

56

97







730


1616

1213

1113

1111

1110

1010

1313

1414

1414

1115

1212

1315

1515

22

55

55

33

33

2222

2222

36

6

7

666

77

78

88

88

7

9

1210

912

149

915

101616

16

2020

2020

1819

181918

1919

1918

171717

17 212121

2121

11

11

2323 23

2323

2424

2424

244

445

44

22

2

58

70

56

76

100

48

70

96

96











731


8765432

8

7

6

5

4

3

2

Pop 22 Montafon

Pop 6-16 W2

Pop 2-5 W1

Pop 17-21 E

Pop 23 24 Dolomites

Pop 1 Corsica

Facto

r 2

Factor 187654321

7

6

5

4

3

2

1

00

Facto

r 4

Factor 3






































732












728









729


5 changes15

14

15

7

19 1818

3

7

8

8

88

8

10129

13

1113

10

10

10

11

11

9

9

9

14

16

1616

16

1216

15

1515

12

12

131413

13

1114

10

1112143

22

22

22

22

7

5

555

7

7

6

6

6

66

1

1

1

1

1919

19

20

20

20

20

21

21

21

2121

19

18

18

17

17 17

17

23

24

2424 24

24 2323

23

23

22

22

2 44

5

44

4

3

3

3

97

68

52

56

97







730


1616

1213

1113

1111

1110

1010

1313

1414

1414

1115

1212

1315

1515

22

55

55

33

33

2222

2222

36

6

7

666

77

78

88

88

7

9

1210

912

149

915

101616

16

2020

2020

1819

181918

1919

1918

171717

17 212121

2121

11

11

2323 23

2323

2424

2424

244

445

44

22

2

58

70

56

76

100

48

70

96

96











731


8765432

8

7

6

5

4

3

2

Pop 22 Montafon

Pop 6-16 W2

Pop 2-5 W1

Pop 17-21 E

Pop 23 24 Dolomites

Pop 1 Corsica

Facto

r 2

Factor 187654321

7

6

5

4

3

2

1

00

Facto

r 4

Factor 3






































732







729


5 changes15

14

15

7

19 1818

3

7

8

8

88

8

10129

13

1113

10

10

10

11

11

9

9

9

14

16

1616

16

1216

15

1515

12

12

131413

13

1114

10

1112143

22

22

22

22

7

5

555

7

7

6

6

6

66

1

1

1

1

1919

19

20

20

20

20

21

21

21

2121

19

18

18

17

17 17

17

23

24

2424 24

24 2323

23

23

22

22

2 44

5

44

4

3

3

3

97

68

52

56

97







730


1616

1213

1113

1111

1110

1010

1313

1414

1414

1115

1212

1315

1515

22

55

55

33

33

2222

2222

36

6

7

666

77

78

88

88

7

9

1210

912

149

915

101616

16

2020

2020

1819

181918

1919

1918

171717

17 212121

2121

11

11

2323 23

2323

2424

2424

244

445

44

22

2

58

70

56

76

100

48

70

96

96











731


8765432

8

7

6

5

4

3

2

Pop 22 Montafon

Pop 6-16 W2

Pop 2-5 W1

Pop 17-21 E

Pop 23 24 Dolomites

Pop 1 Corsica

Facto

r 2

Factor 187654321

7

6

5

4

3

2

1

00

Facto

r 4

Factor 3






































732







730


1616

1213

1113

1111

1110

1010

1313

1414

1414

1115

1212

1315

1515

22

55

55

33

33

2222

2222

36

6

7

666

77

78

88

88

7

9

1210

912

149

915

101616

16

2020

2020

1819

181918

1919

1918

171717

17 212121

2121

11

11

2323 23

2323

2424

2424

244

445

44

22

2

58

70

56

76

100

48

70

96

96











731


8765432

8

7

6

5

4

3

2

Pop 22 Montafon

Pop 6-16 W2

Pop 2-5 W1

Pop 17-21 E

Pop 23 24 Dolomites

Pop 1 Corsica

Facto

r 2

Factor 187654321

7

6

5

4

3

2

1

00

Facto

r 4

Factor 3






































732











731


8765432

8

7

6

5

4

3

2

Pop 22 Montafon

Pop 6-16 W2

Pop 2-5 W1

Pop 17-21 E

Pop 23 24 Dolomites

Pop 1 Corsica

Facto

r 2

Factor 187654321

7

6

5

4

3

2

1

00

Facto

r 4

Factor 3






































732






































732

Vicariance and dispersal in the alpine perennial Bupleurum ......(Apiaceae) Peter Schönswetter 1,3...

Documents

Transcript of Vicariance and dispersal in the alpine perennial Bupleurum ......(Apiaceae) Peter Schönswetter 1,3...