resistance or emigration? response of alpine plants to the ice ages

INTRODUCTIONSince the end of the 19th century, there has been a

long-standing debate about the fate of the mountain floraof the European Alps during the Pleistocene ice ages(reviewed in Stehlik, 2000). Two contrasting scenarios ofglacial survival of alpine plant species have been dis-cussed, namely (1) total extinction within glaciatedmountain ranges, survival in peripheral refugia, and sub-sequent re-immigration into vacant areas after the retreatof glaciers (tabula rasa hypothesis), and (2) long-term insitu survival in glaciated regions in isolated ice-free areasabove the ice-shield (nunatak hypothesis).

The choice of tools for historical biogeographicalinvestigations was limited before the development ofmolecular methods. Geographic mapping of macrofos-

sils and pollen deposits documented large-scale post-glacial plant migrations (Bennett & al., 1991; Lang,1994; Burga & Perret, 1998). However, the fossil recordis limited in taxonomic resolution (Burga & Perret,1998). In addition, fossils of herbaceous and/or alpineplants are especially scarce. Earlier studies on alpineplants therefore mainly relied on their distribution pat-terns (Chodat & Pampanini, 1902; Briquet, 1906; Brock-mann-Jerosch & Brockmann-Jerosch, 1926; Merxmül-ler, 1952, 1953, 1954). In support of the tabula rasahypothesis, locations of peripheral refugia and post-glacial migration routes into the Central Alps had alreadybeen proposed more than a century ago (Chodat &Pampanini, 1902; Briquet, 1906). The Grajic, Cottic andLepontic Alps were postulated to have acted as importantperipheral refugia for the present-day alpine flora of the

499

Stehlik • Resistance or emigration?52 • August 2003: 499–510

Resistance or emigration? Response of alpine plants to the ice ages

Ivana Stehlik

Institute of Systematic Botany, University of Zurich, Zollikerstrasse 107, CH-8008 Zurich, Switzerland andDepartment of Botany, University of Toronto, 25 Willcocks St., Toronto M5S 3B2, Ontario, [email protected]; [email protected]

There is a long-standing debate about the fate of the mountain flora of the European Alps during the Pleisto-cene ice ages. Two main scenarios of glacial survival of alpine plant taxa have been discussed, namely (1) totalextinction within glaciated areas, survival in peripheral refugia, and postglacial re-immigration into vacantareas (tabula rasa hypothesis), and (2) long-term in situ survival within glaciated regions in isolated ice-freeareas above the ice-shield (nunatak hypothesis). Four alpine species with differing distributions and ecologi-cal demands were investigated to elucidate their glacial history using molecular methods (AFLPs, RFLPs ofcpDNA, RAPDs). Their glacial histories are very diverse. Whereas in situ survival in the most intenselyglaciated Central Alps played an important role in Eritrichium nanum, the low alpine Erinus alpinus survivedin situ on some mountains of the northern Swiss Prealps, and Rumex nivalis grows at intermediate alpine ele-vations in snow-beds in both the northern and the Central Swiss Alps. In the common arctic-alpine Saxifragaoppositifolia, the species with the widest distribution and ecological amplitude as compared to the other threespecies, it is not possible to reconstruct its glacial history. It is probable, therefore, that in the Alps, as in north-ern Europe, resident genotypes surviving glaciation in situ were integrated into the gene pool of postglaciallyimmigrating periglacial individuals. The size of refugia differed according to species and region. On the onehand, refugia were restricted to individual mountains (E. alpinus, R. nivalis). On the other hand, they spannedseveral mountain ranges in larger areas (E. nanum, E. alpinus). Postglacial migration over longer distances wasinferred for E. alpinus from southern France to northern Switzerland, and, over shorter distances, for R. nivalisfrom the northern Prealps into the Central Alps in Switzerland. Both postglacial immigration and in situ sur-vival shaped the phylogeography at least of E. alpinus and R. nivalis. It is likely, therefore, that the nunatakand the tabula rasa hypotheses are too simplistic to describe the rich diversity of glacial and postglacialprocesses in Alpine plant species. It rather appears that the glacial history of each species is to a certain degreeunique and influenced by its ecological demands or breeding systems. Moreover, stochasticity has to be regard-ed of essential importance, since factors such as preglacial distribution patterns or postglacial dispersal orextinction events should have had effects on the present genetic composition and the distribution of a species.

KEYWORDS: Alps, Erinus alpinus, Eritrichium nanum, phylogeography, Pleistocene glaciations, Rumex niva-lis, Saxifraga oppositifolia.

Central Alps. Additional peripheral refugia were pro-posed at the margin of the Northern Alps (Briquet, 1906).However, the present distribution of alpine plant speciesis often not in accordance with the tabula rasa hypothe-sis (Brockmann-Jerosch & Brockmann-Jerosch, 1926).For example, central alpine regions sometimes are “hot-spots” in terms of species diversity and even host endem-ic taxa (Brockmann-Jerosch & Brockmann-Jerosch,1926; Welten & Sutter, 1982), but, according to the tab-ula rasa hypothesis, they should have been colonisedlast, and, hence, show lower diversity. Therefore, it hasbeen proposed that at least high alpine plant species havesurvived ice ages in situ in regions such as the southernValais or the Upper Engadine on southerly exposed,windswept cliffs above the ice sheet (nunatak hypothe-sis; Brockmann-Jerosch & Brockmann-Jerosch, 1926;reviewed in Stehlik, 2000; Fig. 1; red areas).

Historical processes such as bottlenecks, migrationsand genetic drift are often reflected in the present-daygenetic composition of populations and permit recon-struction of the historical biogeography or phylogeogra-phy of extant species (Hewitt, 1996; Taberlet & al.,1998). Molecular methods open the possibility to test theproposed hypotheses on glacial survival of alpine species(Stehlik, 2000) because they should lead to alternativepresent-day distributions of neutral genetic markers(Stehlik & al., 2000; Widmer & Lexer, 2001). According

to the tabula rasa hypothesis, genetic similarity amongdescendants of populations of particular peripheral refu-gia along re-immigration routes into the Alps should becomparatively high. Therefore, populations of peripheralrefugia and of Central Alps would belong essentially tothe same gene pool. As recurrent founder events shouldhave taken place along re-immigration routes (geneticthinning), it is also expected that populations at greaterdistances from the periphery of the Alps should be genet-ically less variable than populations closer to peripheralrefugia. Expectations according to the nunatak hypothe-sis are different (Stehlik & al., 2000). High-alpinenunatak populations were probably small and isolated.Long-lasting isolation and inbreeding should have led torelatively low genetic variation within populations and tostrong differentiation among populations on differentnunataks. Therefore, populations from the Central Alpsare expected to be genetically clearly differentiated fromthose of peripheral refugia. Moreover, discontinuouslydistributed groups, each characterised by geneticallyclosely related populations, should be discernible.

Four alpine plant species, Eritrichium nanum (L.)Gaudin (Boraginaceae), Erinus alpinus L. (Scrophularia-ceae), Rumex nivalis Hegetschw. (Polygonaceae) andSaxifraga oppositifolia L. (Saxifragaceae; Fig. 2) withdistinctly different ecological demands and distributionpatterns were chosen to test explicit phylogeographichypotheses. They were investigated using several molec-ular techniques covering uniparental and biparentalgenetic markers. AFLPs and RAPDs represent biparentaland RFLPs of cpDNA uniparental and, supposedly,maternal markers. Additionally, one cpDNA region of E.nanum was further investigated by sequence analysis.For detailed descriptions of the molecular methods andreasons for applying them please refer to the originalpublications (Stehlik & al., 2001, 2002a, b; Holderegger& al., 2002; Stehlik, 2002). The aims of this investigationwere (1) to infer the mode of glacial survival for eachindividual species, (2) thereby to resolve the long-stand-ing controversy between the nunatak and the tabula rasahypotheses and (3) to deduce whether there are commonpatterns [location of glacial refugia and/or (postglacial)migration routes] among species or whether each speciesrepresents its own phylogeographic history influenced byan individual composition of ecological demands,preglacial distribution patterns, and/or stochastic events.

HIGH-ALPINE ERITRICHIUM NANUMThe European endemic Eritrichium nanum belongs

to a group of plant species reaching the highest altitudesin the Alps (2400–3600 m; Lechner-Pock, 1956; Fig.2A). It occurs mainly in exposed habitats, e.g., on ridges

Stehlik • Resistance or emigration? 52 • August 2003: 499–510

Fig. 1. Postulated nunatak areas and peripheral refugia inthe middle part of the Alps during the Pleistocene(Stehlik, 2000). Proposed peripheral refugia are given indark blue and the corresponding migration routestowards the Central Alps with a transition to light blue.The middle intensity blue shows the main area of re-immigration, but it does not illustrate a defined post-glacial point in time. Putative high-alpine nunatak areaswithin formerly most intensively glaciated regions of theCentral Alps are indicated in red (from West to East:southern Valais, Simplon, Val Avers, Upper Engadine).Figure redrawn from Stehlik (2000).

N

France

Switzerland

Germany

Italy

River

Po

River

Rhone

River

Rhine

50 km

500

or summits. Eritrichium nanum is a perennial cushionplant with short flowering shoots and showy blue flow-ers pollinated by a wide array of insects ranging fromsyrphids to bees (Zoller & al., 2001). The species ismainly outcrossing, although it is self-compatible andproduces viable seed upon self-pollination (Zoller & al.,2001). In the middle part of the Alps, its distribution cov-ers areas with many postulated high-alpine nunataks(Stehlik, 2000; Fig. 1, red areas). Based on all these facts,

it appears possible that the species could have survivedin situ in the most intensively glaciated regions duringthe ice ages.

In a first step, 20 mainly Central Alpine populationsof E. nanum were investigated using AFLPs (Stehlik &al., 2001). Two populations of proposed peripheral refu-gia in southern France and northern Italy were chosen forcomparison. Clear genetic differentiation among theselatter two populations and those of the Central Alps was


501

Fig. 2. A, Eritrichium nanum, a European endemic cushion plant occurring mainly in hostile habitats in high-alpinezones; B, Erinus alpinus, a perennial and runner-forming subalpine species restricted to calcareous bedrock; C, Rumexnivalis, a dioecious, wind-pollinated perennial species typical of snow beds at intermediate alpine elevations on cal-careous bedrock; D, arctic-alpine Saxifraga oppositifolia, which has the widest ecological and altitudinal amplitude ofthe four species investigated (photos A–C: I. Stehlik, D: F. Hadacek).

found. Within the Central Alps, three genetically distinctregions were identified. Two of them correspond toCentral Alpine nunatak areas in the southern Valais andin the Upper Engadine (Stehlik, 2000; Fig. 1). The thirdarea is located in northern Tessin. In summary, the resultsof the AFLP analysis reflect the expectations formulatedfor the nunatak hypothesis (Stehlik & al., 2000, 2001) asindividuals from peripheral refugia appeared to be genet-ically too distant from the Central Alpine individuals tohave acted as sources for postglacial re-colonisation.

The amount of genetic variation either within oramong populations as detected with AFLPs is very high(Stehlik & al., 2001). This is most probably caused byconsiderable levels of random gene flow among popula-tions within the three inferred Central Alpine nunatakregions. If much of this gene flow occurred after the lastglaciation, it could have masked originally more distinctpatterns of genetic variation. To circumvent this analyticdrawback of the highly resolving and recombinantAFLPs, which mostly cover nuclear DNA (Mueller &Wolfenbarger, 1999), chloroplast DNA variation wasinvestigated to search for phylogeographic patterns

indicative of glacial refugia in E. nanum and to compareit with the one detected with AFLPs (Stehlik & al., 2001,2002b). As cpDNA is non-recombinant, mostly mater-nally inherited and has a low mutation rate, it is morelikely to retain ancient genetic patterns and to reflect his-torical processes than biparentally inherited nuclearmarkers (Comes & Kadereit, 1998). At the same time,the geographic sample size was extended to the total dis-tribution area of E. nanum, i.e., the whole Alpine archwith 37 investigated populations (Stehlik & al., 2002b).

One cpDNA haplotype is most common and wide-spread (Stehlik & al., 2002b; black circles; Fig. 3A). Atthe same time, it is also genetically basal-most (Fig. 3B).All other haplotypes are inferred to have descended fromthis haplotype. Possibly, this common haplotype, pre-sumably being the ancestor of all extant individuals of E.nanum, colonised the Alps from the East, as the closestrelatives of E. nanum grow in the Carpathians, Caucasus,Asia, Arctic Europe and North America (Lechner-Pock,1956). Only after colonization of the Alps by this basalhaplotype, the accumulation of mutations in the chloro-plast genome and the limited dispersal of seeds resulted


Fig. 3. Thirty-seven populations of Eritrichium nanum in the Alps, eleven haplotypes identified, and a minimum span-ning tree connecting them (Stehlik & al., 2002b). A, Four individuals have been investigated per population by PCR-RFLPs and sequence analysis of non-coding regions of cpDNA (Stehlik & al., 2002b). Populations were always mono-morphic for one haplotype. Dots within the symbols represent the number of mutational steps from the common, black-circled haplotype. Grey-shaded are central Alpine areas (in the West: Southern Valais and southward bordering Aosta;in the East: Upper Engadine and southward bordering Bergamasc Alps) that harboured in situ surviving nunatak popu-lations during Pleistocene glaciations. B, The common, black-circled haplotype was inferred to be evolutionarily basalby the inclusion of Omphalodes verna and Myosotis sylvatica in genetic and statistic analyses (Stehlik & al., 2002b).Bars: indels; crosses: point mutations. Symbols correspond to those used in (A). Figure redrawn from Stehlik & al.(2002b).

terranean Sea

Venice

VillachBolzano

Zurich

Berne

Medi-

Turin

Milan

100 km

N

Geneva

Innsbruck

A

B

502

in a patchy distribution of the divergent haplotypes.These populations harbouring divergent haplotypes areat the present time embedded as “islands” in a “sea” ofthe common black-circled haplotype (Fig. 3A). Ten addi-tional haplotypes were identified, differentiated by one tosix mutations from the common haplotype (Fig. 3B).Three regions contain more derived haplotypes (Fig. 3A;coloured haplotypes). One of them lies in the AustrianCentral Alps of Carinthia and Styria with a haplotype ata maximum genetic distance from the central haplotype(Figs. 3A, B; green haplotype with six dots). From anevolutionary point of view, it can be hypothesised thatthis genetic subdivision occurred early in the Alpine his-tory of E. nanum (Stehlik & al., 2002b). The other tworegions rich in derived haplotypes are situated in theCentral Alps and cover two of the three nunatak regionsas Southern Valais and Aosta and the Upper Engadineincluding adjacent Bergamasc Alps (Fig. 3A; grey-shad-ed areas). The most probable explanation for the cpDNApattern in E. nanum (Stehlik & al., 2002b), in accordancewith AFLPs (Stehlik & al., 2001), implies in situ survivalin rare and isolated populations on nunataks within theCentral Alps (grey-shaded areas; Fig. 3) during Pleisto-cene glaciations.

This result on E. nanum is among the first to showthe capacity of alpine plants to resist the hostile climateon nunataks during Pleistocene glaciation (Stehlik & al.,2001, 2002b; see also Füchter & al., 2001).

LOW-ALPINE ERINUS ALPINUSThe perennial herb Erinus alpinus (Fig. 2B) has

many contrasting characteristics as compared toEritrichium nanum. It occurs at subalpine elevations, andits present distribution only rarely reaches above themaximum elevation of the ice sheet during thePleistocene (Hantke & Jäckli, 1978; Welten & Sutter,1982). The species’ main distribution area lies in south-western Europe (Meusel & al., 1978). As E. alpinus isrestricted to calcareous bedrock (Hartl, 1965), it did nothave the potential to survive on the mainly siliceousnunataks within the high Central Alps. There is a nearlycontinuous band of limestone from southern France tothe northern Prealps in Switzerland, also extending fur-ther to the East. This almost completely coincides withthe present distribution area of E. alpinus in CentralEurope, including its occurrence in southern France(Meusel & al., 1978). This distribution pattern corre-sponds to the theoretically expected postglacial immigra-tion route of E. alpinus from southern France north-wards. Brockmann-Jerosch & Brockmann-Jerosch(1926) therefore formulated the hypothesis of glacial sur-vival of E. alpinus in southern France with subsequent

(re-)immigration into its present-day Alpine distributionrange after the ice ages. Nevertheless, the species alsooccurs on some proposed northern Alpine peripheral re-fugia (Stehlik, 2000; Fig. 1), and in situ glacial survivalcannot be ruled out completely on these latter mountainpeaks.

The results of AFLPs gave a strong phylogeograph-ic signal, whereas there was no genetic variation detect-ed by PCR-RFLPs of cpDNA (Stehlik & al., 2002a).Three clearly differentiated genetic groups were foundamong the 22 investigated populations : (1) central Swisspopulations (Fig. 4; blue), (2) individuals from Mt. Rigi(Fig. 4; small arrow), and (3) all other populations south-west and northeast of the central Swiss populations(“west-eastern” populations; Fig. 4; red to yellow). Theisolation of the central Swiss populations of E. alpinus ishard to reconcile with a continuous postglacial immigra-tion from a southern refugium. Hence, in situ survival incentral Switzerland or the nearby Jura mountains is morelikely. Individuals from central Switzerland contain asignificantly lower number of fragments compared withindividuals from west-eastern populations (Stehlik & al.,2002a). This also supports in situ survival of the former,since long-term isolation with low numbers of individu-als, recurring bottlenecks, inbreeding and genetic driftcould have reduced genetic diversity. The geographiclocations with the highest probability for in situ survivalare the northern-most mountain tops of the Prealps, assome of them rose above the ice even during the glacialmaximum at 20 to 18 ky BP (Burga & Perret, 1998;Stehlik, 2000; Stehlik & al., 2002a). The second remark-able characteristic within the genetic patterns of E. alpi-nus is the genetic isolation of the population of Mt. Rigiand, to a lesser extent, of a population in the Säntis area(Fig. 4; small arrows). The persistence of the spatiallyrestricted populations in genetic isolation from the sur-rounding populations over a long period of time isremarkable. In contrast, Gabrielsen & al. (1997) andTollefsrud & al. (1998) concluded that potential traces ofnunatak survival were wiped out by massively immigrat-ing periglacial genotypes in Nordic Saxifraga oppositi-folia and S. cespitosa. The geographic position of Mt.Rigi as a separate and isolated mountain surrounded bylakes on three sides might have prevented a major genet-ic distortion of its population of E. alpinus.

Most of the genotypes of E. alpinus belong to thewest-eastern gene pool (Stehlik & al., 2002a; Fig. 4). Asmany individuals genetically do not consistently groupwithin their respective populations, the west-eastern pop-ulations have to be addressed as one more or less contin-uous gene pool. A probable postglacial scenario is animmigration of individuals from southern France via thelowlands north of the Alps, as these became ice-free ear-lier than the Prealps and Alps (Fig. 4). Subsequently,


503

southwards migration from the prealpine lowlands andrecolonization of the Prealps took place resulting in thepresent distribution of E. alpinus (Fig. 4). A slightly,although significantly lower level of AFLP fragments perindividual in the east than in the west of the west-easterngroup also fits this hypothesis (Stehlik & al., 2002a).

The investigation on E. alpinus demonstrates howtwo different processes acting over different time scalescan shape a species’ regional genetic composition. Thepopulations of E. alpinus surviving in northern peripher-al refugia have to be dated back at least to the last inter-glacial period (60,000–28,000 y BP), whereas the immi-gration from southern France took place not before theretreat of the glaciers at 14,000 y BP.

SNOW-BED SPECIALIST RUMEXNIVALIS

In the dioecious, wind-pollinated perennial Rumex

nivalis (Fig. 2C), two aspects of its distribution are con-spicuous (Stehlik, 2002). (1) There is a thinning in pop-ulation density towards the species’ western distributionlimit (Welten & Sutter, 1982), which could point toimmigration from a north-eastern peripheral refugium(tabula rasa hypothesis). (2) Its disjunct, alpine distribu-tion centered in the middle part of the Northern Alpsmight result from glacial in situ survival on high moun-tain peaks (nunatak hypothesis). The affinity to extremehabitats in snow-beds (Wagenitz, 1981) with short vege-tation periods might be a prerequisite for this type of gla-cial in situ survival. Therefore, no clear hypothesis onlocations of glacial survival can be formulated for R.nivalis (Stehlik, 2002).

The level of geographic structure in the AFLP data in23 populations of R. nivalis is very low (Stehlik, 2002).When compared with that found in other alpine speciesinvestigated with AFLPs at similar spatial scales, thislow geographic structure in R. nivalis becomes evenmore apparent. The most obvious difference between R.


504

Fig. 4. Hypothesised glacial and post-glacial history of Erinus alpinus in the Alps and the 22 populations as detectedwith AFLPs (Stehlik & al., 2002a). Red, orange and yellow areas represent the west-eastern gene pool of populationsprobably having immigrated postglacially from southern France. This gene pool can be subdivided into three geneti-cally slightly divergent subregions as indicated by the three colors. Postglacial main immigration probably took placevia the pre-alpine lowlands (big arrows), without (genetically detectable) colonization of central Switzerland. CentralSwiss populations (blue) are most probably descendants from populations that survived Pleistocene glaciation in situ.The population hatched in yellow and blue contains no genetically intermediate genotypes but individuals whichbelong genetically either to the west-eastern or the central Swiss gene pool. As populations of mount Rigi, and, to alesser degree of mount Säntis (open circles and small arrows), are genetically divergent from all other populations,they most likely represent local in situ surviving populations. Figure redrawn from Stehlik & al. (2002a).

Central Swisspopulations

West-easternpopulations

Immigration route

In situ survivingpopulations

Descendants ofglacial survivalpopulations

Rigi

Säntis

50 km

N

Po

Rhone

Rhine

River

River

River

France

Germany

Italy

Switzerland

nivalis and Eritrichium nanum, Erinus alpinus andPhyteuma globulariifolium is the breeding system(Stehlik & al., 2001, 2002a; Schönswetter & al., 2002),as R. nivalis is the only wind-pollinated species amongthem. Nearly all populations of R. nivalis are part of onegene pool interconnected via pollen flow (Stehlik, 2002).As the AFLP results do not help to infer glacial refugiaor migration routes of R. nivalis, the interpretation reliedon variation in cpDNA (Stehlik, 2002). The genetic vari-ation in cpDNA is surprisingly high and resulted in 24haplotypes (Fig. 5). As mentioned above, in Eritrichiumnanum, Stehlik & al. (2002b; Fig. 3A) distinguish 11haplotypes covering the entire Alps. Holderegger & al.(2002) found only four haplotypes in alpine S. oppositi-folia, while Abbott & al. (2000) identified 14 haplotypesin the same species at the circumpolar level. Levels sim-ilar to R. nivalis were detected in Dryas integrifolia (20haplotypes), but this sampling covered the whole ofNorth America (Tremblay & Schoen, 1999). In the pres-ent study, the majority of the investigated populationswere sampled within a radius of only 150 km, and thehigh haplotype variation is strongly geographically struc-tured (Stehlik, 2002). Populations at and near the north-ern border of the Alps are richer in both the total numberof haplotypes and the number of private haplotypes, i.e.,haplotypes which occur in just one population (Fig. 6A).This pattern is also reflected by a group of haplotypes (B,E, F, and K) occurring in the haplotypically rich popula-

tions (Figs. 6A, B). Otherwise these northern Alpine hap-lotypes only occur in neighbouring populations of thesehaplotypically rich populations (Figs. 6A, B). The patchydistribution of haplotypes, the high haplotype diversity,and the high number of private haplotypes at the north-ern periphery of the Alps support glacial survival of R.nivalis in northern peripheral refugia and a colonizationof populations towards the Central Alps (Fig. 6D). Onegroup of haplotypes (C, G, and D) is most distant from allother haplotypes (Fig. 5). These three haplotypes areconfined to a geographically restricted region in theCentral Alps (Fig. 6C). Haplotype A co-occurs with hap-lotypes C and D in certain populations (Fig. 6C).Haplotype A is also closely related to the C-G-D-group(Fig. 5). Judging from the regional distribution of haplo-types A, C, D, and G, nunatak survival of R. nivalis issuggested in the Central Alpine area containing thesederived haplotypes (Fig. 6D). This Central Alpine regionwas formerly not recognised as a nunatak area (Stehlik,2000). In concordance with this proposed glacial sce-nario based on more classical statistical methods(AMOVA, Mantel tests, cluster analyses; Stehlik, 2002),nested clade analysis (Templeton & al., 1987, 1995;Templeton, 1998) applied to the haplotypic data on R.nivalis indicated a high colonising capacity of theCentral Alpine haplotypes C, G, and D (as compared tothe northern Alpine haplotypes) or a long-lasting rangeexpansion after glaciation (Stehlik, 2002). On the otherhand, all haplotypes characterised by restricted gene floware situated at or near to the periphery of the Alps (north-ern Alpine haplotypes B, E, F, K; Fig. 6B).

The results of R. nivalis illustrate how stronglydependent the phylogeographic signal is on the type ofmolecular marker system used (no conclusion based onbiparental AFLPs, strong patterns based on uniparentalcpDNA). Furthermore, they show how diverse and com-plicated the Pleistocene history of an alpine species in acomparatively small area can be, and, thus, how impor-tant dense sampling is (Stehlik, 2002).

GENERALIST SAXIFRAGA OPPOSITI-FOLIA

Saxifraga oppositifolia (Fig. 2D) is a long-livedperennial forming loose cushions (Webb & Gornall,1989). Stems of S. oppositifolia usually bear single,mostly outbred protandrous flowers; selfing is, however,possible (Gugerli, 1997). Alpine populations are charac-terised by high RAPD variation (Gugerli & al., 1999).The species is of arctic-alpine distribution and occursalmost throughout the European mountain ranges as acommon, widely distributed plant species. It mainlygrows at alpine elevations between 1800 and 3800 m


505

Fig. 5. CpDNA haplotype minimum spanning tree basedon indel differences (bars) among 24 haplotypes in 23populations of Rumex nivalis in the Alps investigated byPCR-RFLP analysis of cpDNA (four individuals per popu-lation; Stehlik, 2002). No restriction site mutation wasdetected. A transgression from dark blue to white repre-sents the populational frequency of haplotypes. Figureredrawn from Stehlik (2002).

4 3 2 1

Occurrence in

population(s)

B

L

HF

A E

R

S

PK O

Q

T

X

V

N

W

YD

GC

U

M

I

(Kaplan, 1995). Saxifraga oppositifolia belongs to thosefew alpine plants which are able to grow on the highestalpine summits at more than 4000 m. The species’ habi-tats are diverse, ranging from outcrops, rocks, glacierforefields and moraines to gravel on siliceous and cal-careous bedrock (Webb & Gornall, 1989). Several phy-logeographic studies have been conducted on this species(Abbott & al., 1995, 2000; Gabrielsen & al., 1997). As inthe present compilation, an important question in thesestudies was whether this species survived glaciation onice-free nunataks within Pleistocene ice sheets inScandinavia or whether it re-immigrated from periglacialrefugia during the postglacial period. A circumpolar sur-vey of cpDNA haplotypes identified glacial refugia of S.oppositifolia in regions that have never been glaciatedduring Pleistocene ice ages in the Arctic and, potentially,at more southern latitudes (Abbott & al., 2000). The hap-

lotypic diversity in these regions was higher than that ofregions that have been covered by ice. However, theserelictual haplotypes hardly dispersed postglacially and,consequently, are regarded of little importance for thepostglacial colonisation of newly ice-free areas (Abbott& al., 2000). The same conclusion was made for S.oppositifolia in northern Europe (Gabrielsen & al.,1997). If any genotypes did survive in such refugia, theymust have been swamped by massive postglacial immi-gration of common periglacial genotypes (Gabrielsen &al., 1997).

Two widespread, common cpDNA RFLP haplotypeswere found in 15 populations of S. oppositifolia (blackand hatched haplotypes; Fig. 7; Holderegger & al.,2002). Two additional, rare haplotypes also occurred injust three populations (blue and red haplotypes; Fig. 7).In their circumpolar study on S. oppositifolia, Abbott &


506

Fig. 6. The 23 populations investigated of Rumex nivalis in the Alps (Stehlik, 2002). Twenty-two populations were sam-pled in the western part of the species' distribution, and one population from the eastern distribution limit in Austriawas included for comparison (dotted square in upper right corner of each figure). A, Haplotypic richness (four to onehaplotypes) per population is indicated by a transition from black to white. The number of private haplotypes per popu-lation is given by the number of circles around the populations; B, distribution of “northern Alpine” haplotypes occur-ring in more than one population (haplotypes B, E, F, and K); C, distribution of “central Alpine” haplotypes occurringin more than one population (A, C, D, and G); D, hypothesised glacial and post-glacial history of R. nivalis. Stars rep-resent populations that probably survived glaciations in peripheral refugia at the northern border of the Alps.Diamonds suggest in situ surviving populations in a central alpine nunatak area (surrounded by a broken line). Greypopulations were probably postglacially colonised. Bold arrows indicate the inferred migration routes out of peripheralrefugia or central alpine nunataks, whereas fine arrows imply mutational changes along with these migrations. No(detectable) dispersal events originated from the three encircled populations. Figure redrawn from Stehlik (2002).

Haplotypes A GC D

Haplotypes B E F K4 haplotypes

3 haplotypes

2 haplotypes

1 haplotype

A B

C D

al. (2000) found only the common black and hatchedhaplotypes among the individuals surveyed from theAlps. In both studies (Abbott & al., 2000; Holderegger &al., 2002), these haplotypes occurred throughout the Alpswithout any particular spatial pattern (Fig. 7). Since thesetwo haplotypes are common throughout Europe, theycannot be used to deduce the existence or locations ofpotential glacial refugia in the Alps or neighbouringregions (Holderegger & al., 2002).

RAPD analysis revealed a similarly high level ofvariation and a similar partitioning of genetic variance informer studies on S. oppositifolia (Gabrielsen & al.,1997; Gugerli & al., 1999). As in Rumex nivalis (Stehlik,2002), all sampled individuals of S. oppositifolia belongto one large nuclear gene pool without a conspicuousgrouping according to geographical origin or occurrencein nunatak or peripheral refugial areas (Stehlik, 2000).Nevertheless, some populations from the Central AlpineValais exhibit some genetic peculiarities, which may beinterpreted as weak evidence for in situ glacial survivalof the species within the Alps (Holderegger & al., 2002).

However, these genetically distinct populations asdetected with RAPDs are different ones as compared tothose harbouring the two rare RFLP-haplotypes (Fig. 7).

Taken together, both sets of molecular markers pro-vide at best weak indications for in situ survival of S.oppositifolia within the Alps (Holderegger & al., 2002).The best support for such relictual survival is given by anindividual occurrence of the blue haplotype otherwiseonly known from the Siberian Taymyr region (disregard-ing potential homoplasy; Fig. 7; Abbott & al., 2000).

It is thus not possible to reconstruct the glacial histo-ry of this common arctic-alpine species, as the hypothe-sis of in situ glacial survival can neither be proven norfalsified (Holderegger & al., 2002). It is therefore proba-ble that in the Alps, as in northern Europe (Gabrielsen &al., 1997; Abbott & al., 2000), resident genotypes surviv-ing glaciation in situ were integrated into the gene poolof postglacially immigrating periglacial individuals.


507

Fig. 7. Location of investigated populations of Saxifraga oppositifolia from the middle part of the Alps, distribution offour cpDNA haplotypes (black, hatched, blue, red) within populations and overall frequencies of haplotypes based onRFLP analysis using Southern hybridisation (Holderegger & al., 2002). Large symbols represent populations studied byHolderegger & al. (2002), whereas small symbols indicate populations formerly studied by Abbott & al. (2000). Figureredrawn from Holderegger & al. (2002).

Switzerland

Riv

erR

ho

ne

France

River

Rhine

Germany

N50 km

River

P o

Italy

Austria

CONCLUSIONSThe phylogeographic Pleistocene history of

Eritrichium nanum, Erinus alpinus, Rumex nivalis andSaxifraga oppositifolia is very diverse (Stehlik & al.2001, 2002a, b; Holderegger & al., 2002; Stehlik, 2002).The compiled studies are among the first to demonstratestrong evidence for in situ glacial survival in the CentralAlps for Alpine species during the Pleistocene (see alsoFüchter & al., 2001). In the case of Eritrichium nanum,the area of in situ nunatak survival coincides with aregion of intense Pleistocene glaciation (Stehlik, 2000;Stehlik & al., 2002b). The size of these in situ glacialrefugia differs according to species and region. Somerefugia were restricted to individual mountains, i.e., inthe case of Erinus alpinus and R. nivalis (Figs. 4, 6D),whereas others span larger mountain areas as the south-ern Valais for Eritrichium nanum (Fig. 3A), the centralpart of northern Alps for Erinus alpinus (Fig. 4), and aregion in the Central Alps for R. nivalis (Fig. 6D). Noevidence for (or no need to invoke) such Central Alpinenunatak survival was detected in studies with compara-ble sample sizes, molecular markers; and geographic dis-tributions, e.g., in Anthyllis montana, Saponaria pumilaor Phyteuma globulariifolium (Kropf & al., 2002;Schönswetter & al., 2002; Tribsch & al., 2002). Thegenetic patterns of these alpine species could be relatedto glacial survival in small peripheral refugia in southernand easternmost Alps and postglacial migration towardsthe species’ present distributions (see also Tribsch &Schönswetter, 2003). Such postglacial migration is alsoinferred for some of the four species of the present stud-ies reviewed. Over longer distances, Erinus alpinusmigrated postglacially from Southern France to NorthernSwitzerland (Fig. 4), or, over shorter distances, R. nivalisfrom the northern Prealps into the Central Alps (Fig. 6D).There is clear evidence that both postglacial immigrationand in situ survival shaped the genetic constitution andthe present distribution pattern of at least E. alpinus andR. nivalis. It is therefore probable that the nunatak andthe tabula rasa hypotheses are too simplistic to describethe rich diversity of glacial and postglacial processes. Itrather appears that the glacial history of each species isto a certain degree unique and influenced by its compo-sition of ecological demands or breeding systems.Moreover, stochasticity should be regarded of essentialimportance, since factors such as preglacial distributionpatterns or postglacial dispersals and extinctions haveclear effects on the present genetic composition and thedistribution of species.

While the dichotomy of the nunatak vs. tabula rasahypotheses might lead to effective results when dealingwith the massive Arctic ice cap (Gabrielsen & al., 1997;Tollefsrud & al., 1998; Abbott & al., 2000), for the much

smaller and more fragmented Alpine ice shield, a greaterdiversity of processes seems probable. The attempt toexplain this diversity of refugia and migration routesunder two hypotheses would ignore important phylogeo-graphic signals. A more productive way to infer generalpatterns in Alpine plants might be to compile the resultsof future biogeographical investigations with similarsampling designs and to search for repeated patterns,and, thereby refine or reformulate hypotheses for glacialsurvival in the Alps.

ACKNOWLEDGEMENTSI am particularly grateful to all co-authors of the studies of E.

nanum, E. alpinus and S. oppositifolia, i.e., to Richard J. Abbott,Konrad Bachmann, Frank R. Blattner, Rolf Holderegger and J.Jakob Schneller. Peter Linder and Peter Schönswetter read themanuscript thoroughly. I am obliged to Peter Schönswetter andAndreas Tribsch for the collection of many samples and for theinspiring time we spent together. This article is an enlarged versionof a talk presented at ICSEB 2002 in Patras, Greece. In this con-text, I would like to express my gratitude to Tod Stuessy andAndreas Tribsch for inviting me to their symposium (Evolutionand Phylogeography of Arctic and Alpine Plants in Europe) andgiving me the opportunity to write up this synthesis of my Ph.D.thesis. I have been supported by the Swiss National ScienceFoundation (grant numbers 31-55390.98 to JJS), by the SwissAlpine Club SAC, and the Georges and Antoine ClarazSchenkung.

LITERATURE CITEDAbbott, R. J., Chapman, H. M., Crawford, R. M. M. &

Forbes, D. G. 1995. Molecular diversity and derivationsof populations of Silene acaulis and Saxifraga oppositi-folia from the high Arctic and more southerly latitudes.Molec. Ecol. 4: 199–207.

Abbott, R. J., Smith, L. C., Milne, R. I., Crawford, R. M.M., Wolff, K. & Balfour, J. 2000. Molecular analysis ofplant migration and refugia in the Arctic. Science 289:1343–1346.

Bennett, K. D., Tzedakis, P. C. & Willis, K. J. 1991.Quaternary refugia of north European trees. J. Biogeogr.18: 103–115.

Briquet, J. 1906. Le développement des flores dans les Alpesoccidentales, avec aperçu sur les Alpes en général. Pp.130–173 in: von Wettstein, R., Wiesner, J. &Zahlbruckner, A. (eds.), Wissenschaftliche Ergebnisse desInternationalen Botanischen Kongresses Wien 1905.Gustav Fischer, Jena.

Brockmann-Jerosch, H. & Brockmann-Jerosch, M. 1926.Die Geschichte der Schweizerischen Alpenflora. Pp.1110–1215 in: Schröter, C. (ed.), Das Pflanzenleben derAlpen. Raustein, Zürich.

Burga, C. A. & Perret, R. 1998. Vegetation und Klima der


508

Schweiz seit dem jüngeren Eiszeitalter. Ott, Thun.Chodat, R. & Pampanini, R. 1902. Sur la distribution des

plantes dans les alpes austro-orientales et plus partic-ulièrement d’un choix de plantes des alpes cadoriques etvenitiennes. Le Globe 41: 1–70.

Comes, H. P. & Kadereit, J. W. 1998. The effect ofQuaternary climatic changes on plant distribution and evo-lution. Trends Plant Sci. 3: 432–438.

Füchter, R., Widmer, A. & Baltisberger, M. 2001. Nunataksas glacial refugia? A molecular phylogeographic casestudy with Draba aizoides. Bauhinia 15: 78. [Abstr.]

Gabrielsen, T. M., Bachmann, K., Jakobsen, K. S. &Brochmann, C. 1997. Glacial survival does not matter:RAPD phylogeography of Nordic Saxifraga oppositifolia.Molec. Ecol. 6: 831–842.

Gugerli, F. 1997. Sexual reproduction in Saxifraga oppositifo-lia L. and S. biflora All. (Saxifragaceae) in populations ofthe Alps. Int. J. Plant Sci. 158: 274–281.

Gugerli, F., Eichenberger, K. & Schneller, J. J. 1999.Promiscuity in populations of the cushion plant Saxifragaoppositifolia in the Swiss Alps as inferred from randomamplified polymorphic DNA (RAPD). Molec. Ecol. 8:453–461.

Hantke, R. & Jäckli, H. 1978. Die Schweiz zur letzten Eiszeit.Table 6 in: Imhof, E. (ed.), Atlas der Schweiz. Wabern,Bern.

Hartl, D. 1965. Familie Scrophulariaceae. Pp. 1–469 in: Hartl,D. & Wagenitz, G. (eds.), Gustav Hegi, Illustrierte Floravon Mitteleuropa, ed. 2, vol. VI (1). Hanser, München.

Hewitt, G. M. 1996. Some genetic consequences of the iceages, and their role in divergence and speciation. Biol. J.Linn. Soc. 58: 247–276.

Holderegger, R., Stehlik, I. & Abbott, R. J. 2002. Molecularanalysis of the Pleistocene history of Saxifraga oppositifo-lia in the Alps. Molec. Ecol. 11: 1409–1418.

Kaplan, K. 1995. Familie Saxifragaceae, Steinbrechgewächse.Pp. 130–229 in: Weber, H. E. (ed.), Gustav Hegi,Illustrierte Flora von Mitteleuropa, ed. 3, vol. IV2A.Blackwell, Berlin.

Kropf, M., Kadereit J. W. & Comes, H. P. 2002. LateQuaternary distributional stasis in the submediterraneanmountain plant Anthyllis montana L. (Fabaceae) inferredfrom ITS sequences and amplified fragment length poly-morphism markers. Molec. Ecol. 11: 447–463.

Lang, G. 1994. Quartäre Vegetationsgeschichte Europas. G.Fischer, Jena.

Lechner-Pock, L. 1956. Eritrichium nanum (Amann)Schrader und seine Verwandten. Phyton 6: 98–206.

Merxmüller, H. 1952. Untersuchungen zur Sippengliederungund Arealbildung in den Alpen, I. Jahrb. Ver. SchutzAlpenpfl. Alpentiere 17: 96–133.

Merxmüller, H. 1953. Untersuchungen zur Sippengliederungund Arealbildung in den Alpen, II. Jahrb. Ver. SchutzAlpenpfl. Alpentiere 18: 138–158.

Merxmüller, H. 1954. Untersuchungen zur Sippengliederungund Arealbildung in den Alpen, III. Jahrb. Ver. SchutzAlpenpfl. Alpentiere 19: 97–139.

Meusel, H., Jäger, E., Rauschert, S. & Weinert, E. 1978.Vergleichende Chorologie der zentraleuropäischen Flora.G. Fischer, Jena.

Mueller, U. G. & Wolfenbarger, L. L. 1999. AFLP genotyp-ing and fingerprinting. Trends Ecol. Evol. 14: 389–395.

Schönswetter, P., Tribsch, A., Barfuß, M. & Niklfeld, H.2002. Several Pleistocene refugia detected in the highalpine plant Phyteuma globulariifolium Sternb. & Hoppe(Campanulaceae) in the European Alps. Molec. Ecol. 11:2637–2647.

Stehlik, I. 2000. Nunataks and peripheral refugia for alpineplants during Quaternary glaciation in the middle parts ofthe Alps. Bot. Helv. 110: 25–30.

Stehlik, I. 2002. Glacial history of the alpine Rumex nivalis(Polygonaceae): a comparison of common phylogeo-graphic methods with nested clade analysis. Amer. J. Bot.89: 2007–2016.

Stehlik, I., Blattner, F. R., Holderegger, R. & Bachmann, K.2002. Multiple glacial nunatak survival of the high alpineEritrichium nanum (L.) Gaudin within the European Alps.Molec. Ecol. 11: 2027–2036.

Stehlik, I., Holderegger, R., Schneller J. J., Abbott R. J. &Bachmann, K. 2000. Molecular biogeography and popu-lation genetics of alpine plant species. Bull. Geobot. Inst.ETH 66: 47–59.

Stehlik, I., Schneller, J. J. & Bachmann, K. 2001. Resistanceor emigration: response of the high-alpine plantEritrichium nanum (L.) Gaudin to the ice age within theCentral Alps. Molec. Ecol. 10: 357–370.

Stehlik, I., Schneller J. J. & Bachmann, K. 2002. Immi-gration and in situ glacial survival of the low-alpine Erinusalpinus (Scrophulariaceae). Biol. J. Linn. Soc. 77: 87–103.

Taberlet, P., Fumagalli, L., Wust-Saucy, A.-G. & Cosson, J.-F. 1998. Comparative phylogeography and postglacial col-onization routes in Europe. Molec. Ecol. 7: 453–464.

Templeton, A. R. 1998. Nested clade analyses of phylogeo-graphic data: testing hypothesis about gene flow and pop-ulation history. Molec. Ecol. 7: 381–397.

Templeton, A. R., Boerwinkle, E. & Sing, C. F. 1987. Acladistic analysis of phenotypic associations with haplo-types inferred from restriction endonuclease mapping. I.Basic theory and an analysis of alcohol dehydrogenaseactivity in Drosophila. Genetics 117: 343–351.

Templeton, A. R., Routman, E. & Phillips, C. A. 1995.Separating population structure from population history: acladistic analysis of the geographical distribution of mito-chondrial DNA haplotypes in the tiger salamander,Ambystoma tigrinum. Genetics 140: 767–782.

Tollefsrud, M. M., Bachmann, K., Jakobsen, K. S. &Brochmann, C. 1998. Glacial survival does not matter—II: RAPD phylogeography of Nordic Saxifraga cespitosa.Molec. Ecol. 7: 1217–1232.

Tremblay, N. O. & Schoen, D. J. 1999. Molecular phylo-geography of Dryas integrifolia: glacial refugia and post-glacial recolonization. Molec. Ecol. 8: 1187–1198.

Tribsch, A. & Schönswetter, P. 2003. Patterns of endemismand comparative phylogeography confirm palaeo-environ-mental evidence for Pleistocene refugia in the EasternAlps. Taxon 52: 477–497.

Tribsch, A., Schönswetter, P. & Stuessy, T. F. 2002.Saponaria pumila (Caryophyllaceae) and the ice age in theEuropean Alps. Amer. J. Bot. 89: 2024–2033.

Wagenitz, G. 1981. Familie Polygonaceae. Pp. 352–436 in:Wagenitz, G. (ed.), Gustav Hegi, Illustrierte Flora vonMitteleuropa, ed. 3, vol. III (1). Parey, Berlin.

Webb, D. A. & Gornall, R. J. 1989. Saxifrages of Europe.Christopher Helm, London.


509

Welten, M. & Sutter, R. 1982. Verbreitungsatlas der Farn-und Blütenpflanzen der Schweiz. Birkhäuser, Basel.

Widmer, A. & Lexer, C. 2001. Glacial refugia: sanctuaries forallelic richness, but not for gene diversity. Trends Ecol.Evol. 16: 267–269.

Zoller, H., Lenzin, H. & Erhardt, A. 2001. Untersuchungenzum Lebenszyklus der hochalpinen PolsterpflanzeEritrichium nanum (L.) Schrad. ex Gaudin Teil I:Blütenbiologie. Bauhinia 15: 1–17.


510

resistance or emigration? response of alpine plants to the ice ages

Documents