Abstract

Amplified Fragment Length Polymorphism (AFLP) was used to clarify the glacial history of the rare, disjunctly distributed, alpine cushion plant Androsace wulfeniana, which is endemic to the Eastern Alps (Austria and Italy). Disjunct populations in the Dolomites are genetically very distinct from those in the main distributional area. It is hypothesized that they are descendants of long‐term isolated glacial survivors and are not a result of recent long‐­distance dispersal. Within the main distributional area of the species in the central Eastern Alps, two groups of ­populations can be distinguished, which are congruent with hotspots of rare relictual vascular plant taxa. In the ­taxonomically closely related A. brevis growing in the Southern Alps (Italy, Switzerland), no genetic‐geographical structure was found. Genetic variation is extremely low in disjunct populations of A. wulfeniana in the Dolomites and in A. brevis. In contrast, in the main distributional area of A. wulfeniana, genetic variation is similar to that of the colonizing widespread congener A. alpina. © 2003 The Linnean Society of London, Botanical Journal of the ­Linnean Society, 2003, 141, 437–446.

INTRODUCTION

Disjunctions are a striking biogeographical aspect exhibited by taxa of different ranks and on different geographical scales. Well documented examples on the smaller scale are provided by mountain ranges like the European Alps (Merxmüller, 1952, 1953, 1954). Disjunctions often result in allopatric speciation. An excellent example in the Alps is provided by members of Primula sect. Auricula, where in several subsections para‐ and allopatric species occur in parallel (maps in Merxmüller, 1952; Meusel et al. 1978). Disjunctions, however, need not be accompanied by a speciation event, leading instead to intraspecific disjunctions.

Intraspecific disjunctions can be caused by ecological or historical factors. A well‐known example of the former is substrate specificity, e.g. calciphilous species distributed in the northern and the southern Calcareous Alps (examples in Merxmüller, 1952, 1953, 1954). As a historical factor, Pleistocene glaciations have been regarded as a driving force in generating disjunctions in the Alps (e.g. Merxmüller, 1952). In these cases the resulting distributional patterns are not explainable through ecological correlates but rather through striking geographical concordance with formerly unglaciated or only weakly glaciated regions (e.g. Merxmüller, 1952, 1953, 1954; Schneeweiss & Schönswetter, 1999). In the Alps these regions are in good congruence with centres of endemism (Pawlowski, 1970), e.g. the north‐eastern (Niklfeld, 1972) and southern Calcareous (Pitschmann & Reisigl, 1959) Alps.

Since the central Eastern Alps are flanked by peripheral limestone ranges to the north and south, only a few regions with siliceous bedrock were unglaciated or only locally glaciated during the maximum extent of the Pleistocene ice shield (Van Husen, 1987). By far the largest ice‐free area was the easternmost Central Alps. Furthermore, smaller ice‐free areas were situated in the Southern Alps in the south‐western Dolomites, southern Adamello, western Alpi Bergamasche and adjacent Alpi Lepontine (Jäckli, 1970). These regions provided, if not entirely unglaciated territory, at least nunataks in a very peripheral situation. Thus, judging from geological evidence these regions are likely to have acted as Pleistocene refugia for calcifuge alpine plants.

Modern DNA techniques are excellent tools to reconstruct population histories. The level of genetic differentiation gives insights, e.g. on the relative timing of splitting events and the relationships between populations. This can be used to determine the origin of disjunctions (recent long‐distance dispersal vs. old splitting event) as well as to reconstruct migration patterns and to recognize refugial populations. The genetic variability of populations can be used for the detection of bottlenecks and the estimation of ­viability.

A good model to investigate such questions is provided by the species pair Androsace wulfeniana Sieber ex Koch and A. brevis (Hegetschw.) Cesati (Primulaceae), which are sister species according to preliminary phylogenetic data (DNA‐sequencing and AFLP‐fingerprinting, G. M. Schneeweiss et al. unpubl.). ­Androsace wulfeniana is distributed in the easternmost Central Alps of Austria and disjunct (260 km) in the siliceous parts of the southern Dolomites in Italy (Fig. 1). In contrast, A. brevis has a compact distributional area in the southern Alps around Lake Como in Italy and nearby Switzerland (Fig. 1). All three regions were either weakly glaciated or situated at the periphery of the ice shield during the last glaciation (Fig. 1). The habitat requirements of the two species are very similar. Both are restricted to acidic to subneutral siliceous bedrock and grow on wind‐exposed ridges with low vegetation cover or on rocky outcrops, between 2000 and 2600 m (Franz, 1988; Käsermann & Moser, 1999). The high‐alpine, colonizing A. alpina (L.) Lam., which is distributed throughout the high parts of the Alps, was recently investigated applying the same methodology (P. Schönswetter, A. Tribsch & H. Niklfeld, unpubl.). This allows a comparison of the genetic variability of the two rare relict taxa A. brevis and A. wulfeniana with their widespread congener A. alpina.

Figure 1.

Distribution of Androsace brevis (squares and lined area around squares) and A. wulfeniana (circles and squared areas). Filled symbols are sampled populations, the numbers refer to the population numbers given in Table 1. The maximum extent of the ice‐shield during the last glaciation period (Wuerm) in the Alps is given as a black line.

Figure 1.

Distribution of Androsace brevis (squares and lined area around squares) and A. wulfeniana (circles and squared areas). Filled symbols are sampled populations, the numbers refer to the population numbers given in Table 1. The maximum extent of the ice‐shield during the last glaciation period (Wuerm) in the Alps is given as a black line.

The following questions are addressed in this study: are the populations of A. wulfeniana in the Dolomites glacial survivors or did they originate from rather recent long‐distance dispersal? Is there structure within the two disjunct distributional areas of A. wulfeniana or the compact one of A. brevis? Are there differences regarding genetic diversity between the two species or the three regions, respectively? Are there such differences between the two investigated relict taxa and the more widespread high‐alpine congener A. alpina, a colonizing species?

MATERIAL AND METHODS

SAMPLING

Eight populations of A. wulfeniana and eight populations of A. brevis covering the entire distributional ranges of the two species were sampled (Table 1, Fig. 1). The population size was roughly estimated in two categories (Table 1). Usually, leaf‐material of five individuals per population was collected. Due to extremely small population sizes, in four populations of A. brevis only material of one or two individuals was collected. Young shoots (flowers and buds were removed) were immediately stored in silica gel. Herbarium specimens of all sampled populations are deposited in the herbarium of the Institute of Botany of the University of Vienna (WU).

Table 1.

Populations, location name (A: Austria, I: Italy), coordinates, altitude in m a.s.l. and estimated population size [1: small (<100); 2: large (>100)] of the 16 investigated populations of Androsace wulfeniana (1–8) and A. brevis (9–16)

Population number Location Coordinates (E/N) Altitude Population size 
A: Dreistecken 14°23′ 50″ / 47°27′ 27″ 2360 
A: Gstoder 14°13′ 00″ / 47°17′ 50″ 2240 
A: Gumma 13°46′ 48″ / 47°12′ 42″ 2310 
A: Zirbitzkogel 14°34′ 07″ / 47°04′ 45″ 2225 
A: Bretthöhe 13°56′ 18″ / 46°54′ 50″ 2300 
A: Falkert 13°49′ 16″ / 46°51′ 30″ 2020 
I: Cavallazza Piccola 11°47′ 20″ / 46°17′ 45″ 2170 
I: Forcola di Coldose 11°37′ 34″ / 46°15′ 31″ 2180 
I: Valle del Muretto 09°44′ 35″ / 46°20′ 25″ 2280 
10 I: Rifugio Gianetti 09°35′ 35″ / 46°17′ 00″ 2540 
11 I: Passo d'Oro 09°34′ 00″ / 46°15′ 40″ 2570 
12 I: Cima Pianchette 09°08′ 50″ / 46°07′ 30″ 2140 
13 I: Monte Azzarini 09°38′ 40″ / 46°03′ 30″ 2400 
14 I: Monte Colombana 09°30′ 20″ / 46°03′ 45″ 2355 
15 I: Monte Rotondo 09°29′ 30″ / 46°04′ 10″ 2495 
16 I: Monte Legnone 09°25′ 00″ / 46°05′ 45″ 2600 
Population number Location Coordinates (E/N) Altitude Population size 
A: Dreistecken 14°23′ 50″ / 47°27′ 27″ 2360 
A: Gstoder 14°13′ 00″ / 47°17′ 50″ 2240 
A: Gumma 13°46′ 48″ / 47°12′ 42″ 2310 
A: Zirbitzkogel 14°34′ 07″ / 47°04′ 45″ 2225 
A: Bretthöhe 13°56′ 18″ / 46°54′ 50″ 2300 
A: Falkert 13°49′ 16″ / 46°51′ 30″ 2020 
I: Cavallazza Piccola 11°47′ 20″ / 46°17′ 45″ 2170 
I: Forcola di Coldose 11°37′ 34″ / 46°15′ 31″ 2180 
I: Valle del Muretto 09°44′ 35″ / 46°20′ 25″ 2280 
10 I: Rifugio Gianetti 09°35′ 35″ / 46°17′ 00″ 2540 
11 I: Passo d'Oro 09°34′ 00″ / 46°15′ 40″ 2570 
12 I: Cima Pianchette 09°08′ 50″ / 46°07′ 30″ 2140 
13 I: Monte Azzarini 09°38′ 40″ / 46°03′ 30″ 2400 
14 I: Monte Colombana 09°30′ 20″ / 46°03′ 45″ 2355 
15 I: Monte Rotondo 09°29′ 30″ / 46°04′ 10″ 2495 
16 I: Monte Legnone 09°25′ 00″ / 46°05′ 45″ 2600 

DNA ISOLATION AND AFLP FINGERPRINTING

In contrast to other congeners, e.g. A. alpina (Schönswetter et al. unpubl.), DNA extraction was problematic – apparently due to a high amount of polysaccharides in the plants. The best quality of extracts was obtained by a combination of different protocols. Plant material was ground to a fine powder with a shaking‐mill (Retsch MM 200). Then 800 µL of AP1 extraction buffer (Dneasy Plant Mini Kit, Qiagen) were added and shaken on a thermoblock at 60 °C. After 10 min 260 µL of AP2 precipitation buffer (Dneasy Plant Mini Kit, Qiagen) were added. After 5 min on ice and subsequent centrifuging at 14 000 r.p.m. for 5 min the supernatant was moved to a new Eppendorf tube and 500 µL chloroform/isoamyl‐alcohol (24 : 1) were added. After mixing and standing for 5 min it was centrifuged again and the aqueous phase was moved to a new Eppendorf tube. 500 µL isopropanol were added and after standing for 5 min followed by centrifuging the supernatant was discarded. 1 ml 96% ethanol was added. After vigorous manual shaking the extractions were put on a thermoblock shaking at maximum speed at 37 °C. After c. 0.5 h the pellet became isolated from a diffuse slime. If necessary, this step was repeated. The quality of the extracted DNA was checked on 1% TAE‐agarose gels. The amount of DNA was estimated photometrically (UV‐160 A, Shimadzu).

Genomic DNA (c. 500 ng) was digested with MseI (New England BioLabs) and EcoRI (Promega) and ligated (T4 DNA‐Ligase; Promega) to double‐stranded adapters and preamplified using the AFLP Ligation and Preselective Amplification Module for regular genomes following the manufacturer's instructions (PE Applied Biosystems, 1996). The incubation of the restriction‐ligation reactions (2 h at 37 °C) as well as the PCRs were performed on a GeneAmp PCR System 9700 thermal cycler. Deviating from the manufacturer's instructions, the PCRs were run in a reaction volume of 5 µL. Three primer combinations, which were used for the closely related A. alpina (Schönswetter et al. unpubl.) gave clear, easily and unambiguously scoreable results (EcoRI FAM‐ACA‐MseI CAT; EcoRI JOE‐AAG‐MseI CTG; EcoRI NED‐AAC‐MseI CTT). On some samples independent AFLP‐­reactions were performed for internal control. The fluorescence‐labelled selective amplification‐products were separated on a 5% polyacrylamide gel with an internal size standard (GeneScan‐500 [ROX], PE Applied Biosystems) on an automated sequencer (ABI 377). Raw data were collected and aligned with the internal size standard using the ABI Prism GeneScan Analysis Software (PE Applied Biosystems). Subsequently, the GeneScan‐files were imported into Genographer (version 1.1.0, © Montana State University, 1998; http://hordeum.msu.montana.edu/genographer/) for scoring of the fragments. Each AFLP‐fragment was scored using the thumbnail option of the program which allows comparison of the signal per locus over all samples. The few AFLP‐­fragments that exhibited ambiguous peaks were excluded from the analysis. Peaks of low intensity were included in the analysis when an unambiguous scoring was possible. The results of the scoring were exported as a presence/absence matrix and used for further manipulation.

DATA ANALYSIS

Shannon Diversity index HSh = ‐Σ(pi ln pi), where pi is the relative frequency of the ith fragment (Legendre & Legendre, 1998) and the mean Jaccard similarities between and within populations were calculated. The number of variable fragments and the numbers of private (i.e. confined to a single population) and fixed private (i.e. found in all investigated individuals of a single population) fragments were estimated. All these parameters were determined only for populations with four or five investigated individuals. Hence, four populations (pops. 9–12) of A. brevis were omitted. Analyses of molecular variance (AMOVAs) were calculated with Arlequin 1.1 (Schneider et al., 1997). A neighbour‐­joining tree of all individuals based on the distance measure by Nei & Li (1979) was constructed with ­Treecon 1.3b (Van de Peer & De Wachter, 1994). A Principal Coordinate Analysis (PCoA) based on a matrix of between‐individual Jaccard similarities and boxplots of HSh were calculated and plotted with SPSS 8.0.

RESULTS

AFLP‐PATTERNS AND POLYMORPHISM

With the three primer combinations used, 205 un‐ambiguously scoreable fragments were generated. Thirty‐six fragments (17.6%) are monomorphic in the entire data set. In A. brevis 33 (28.4%) of 116 fragments were polymorphic, in A. wulfeniana 119 (69.6%) of 171. The number of AFLP‐fragments per individual varied from 96 to 104 in A. brevis and from 97 to 112 in A. wulfeniana.

WITHIN‐POPULATION VARIATION (TABLE 2)

Table 2.

Genetic variation in the 16 investigated populations of Androsace wulfeniana and A. brevis. N: number of individuals which were included in the analyses; HSH: Shannon Diversity Index; NPF (NFPF): number of private fragments (number of fixed private fragments); PPOLY: percentage of variable fragments

Population number Location N HSH NPF (NFPFPPOLY 
Dreistecken 14.22 4(1) 38.9 
Gstoder 9.39 27.4 
Gumma 10.10 29.2 
Zirbitzkogel 9.52 6(2) 28.1 
Bretthöhe 13.68 33.1 
Falkert 10.82 27.7 
Cavallazza Piccola 1.40 2(2) 5.9 
Forcoladi Coldose 2.73 5(4) 8.3 
Valledel Muretto – – – 
10 Rifugio Gianetti – – – 
11 Passod’ Oro – – – 
12 Cima Pianchette – – – 
13 Monte Azzarini 4.98 15.6 
14 Monte Colombana 2.57 7.8 
15 Monte Rotondo 3.39 11.5 
16 Monte Legnone 2.62 8.6 
Population number Location N HSH NPF (NFPFPPOLY 
Dreistecken 14.22 4(1) 38.9 
Gstoder 9.39 27.4 
Gumma 10.10 29.2 
Zirbitzkogel 9.52 6(2) 28.1 
Bretthöhe 13.68 33.1 
Falkert 10.82 27.7 
Cavallazza Piccola 1.40 2(2) 5.9 
Forcoladi Coldose 2.73 5(4) 8.3 
Valledel Muretto – – – 
10 Rifugio Gianetti – – – 
11 Passod’ Oro – – – 
12 Cima Pianchette – – – 
13 Monte Azzarini 4.98 15.6 
14 Monte Colombana 2.57 7.8 
15 Monte Rotondo 3.39 11.5 
16 Monte Legnone 2.62 8.6 

Six pairs of individuals with identical AFLP profiles were detected. Four were found in A. brevis (within pop. 11 and pop. 14, between pop. 12 and pop. 9, between pop. 14 and pop. 15), and two in A. wulfeniana from the Dolomites (within pop. 7 and pop. 8). HSh ranged from 2.57 to 4.98 with an average of 3.39 in A. brevis, and from 1.4 to 14.2 with an average of 8.99 in A. wulfeniana. Excluding pops. 7 and 8 from the Dolomites (average 2.01), the average for the main distribution area is 11.29. Boxplots of HSh of A. brevis, A. wulfeniana and A. alpina (data from Schönswetter et al. unpubl.) are given in Figure 2. HSh of A. alpina and A. wulfeniana excluding pops. 7 and 8 are not significantly different (t‐test, P < 0.01). Significant differences were found between A. brevis and A. alpina and A. wulfeniana (excluding pops. 7 and 8). The percentage of variable loci ranged from 7.8 to 15.6 in populations of A. brevis and from 5.9 to 38.9 in A. wulfeniana. Private fragments (1–4 per population) were found in one population of A. brevis and five of A. wulfeniana. Fixed private fragments were only detected in A. wulfeniana in pop. 4 (2 fixed private fragments), pop. 1 (1), pop. 7 (2) and pop. 8 (4).

Figure 2.

Comparison of Shannon Diversity Indices of Androsace alpina, A. brevis, A. wulfeniana in the Dolomites and A. wulfeniana in the eastern Central Alps. Boxplot: the box represents the interquartile range which contains 50% of the values and the median (horizontal line across the box); the whiskers are lines that extend from the box to the highest and lowest values, excluding outliers (○).

Figure 2.

Comparison of Shannon Diversity Indices of Androsace alpina, A. brevis, A. wulfeniana in the Dolomites and A. wulfeniana in the eastern Central Alps. Boxplot: the box represents the interquartile range which contains 50% of the values and the median (horizontal line across the box); the whiskers are lines that extend from the box to the highest and lowest values, excluding outliers (○).

BETWEEN‐POPULATION VARIATION

The neighbour‐joining tree (Fig. 3) of the individuals shows no geographical structuring at all in A. brevis, but a very high one in A. wulfeniana, where three geographical regions are clearly separated: (a) Zirbitzkogel (pop. 4) and eastern Niedere Tauern (pops. 1, 2); (b) Gurktaler Alpen (pops. 5, 6) and western Niedere Tauern (pop. 3); and (c) Dolomites (pops. 7, 8). This difference between the two species is also reflected by the analyses of molecular variance (AMOVA, Table 3). The proportion of the total genetic variation assigned to variation between populations is four times higher in A. wulfeniana than in A. brevis. If in A. wulfeniana a geographical structure (i.e. the three groups listed above) is added as a third level of variation, 35.6% is assigned to variation among those groups. If only the main distributional area in the easternmost Central Alps is considered, the value decreases to 18.1%. Because of the vicinity of A. brevis populations represented with a sufficient number of individuals in the data set, and due to the obvious lack of geographical structure in the neighbour‐joining tree, the introduction of a third level of variance is not meaningful in this species. However, to enable a comparison with A. alpina and A. wulfeniana AMOVAs for all possible groupings of populations of A. brevis were calculated. The highest values for variation between groups were obtained for a grouping of pops. 13 and 16 vs. pops. 14 and 15 with 6.4% for variation between groups.

Figure 3.

Neighbour‐joining tree of Androsace brevis (left) and A. wulfeniana (right); the numbers refer to populations listed in Table 1. Scale bar = 10% genetic distance (Nei & Li, 1979).

Figure 3.

Neighbour‐joining tree of Androsace brevis (left) and A. wulfeniana (right); the numbers refer to populations listed in Table 1. Scale bar = 10% genetic distance (Nei & Li, 1979).

Table 3.

Results of the AMOVAs (Analyses of Molecular Variance). (a), (b) Partitioning of the overall genetic variation on two levels (within and between populations). (c)–(e) Level of variation between groups added. (c) Grouping with highest value for variation between groups (pop. 13, pop. 16) (pop. 14, pop. 15). (d) Grouping according to PCoA and neighbour‐joining (pop. 1, pop. 2, pop. 4) (pop. 3, pop. 5, pop. 6) (pop. 7, pop. 8). (e) Populations from the Dolomites (pop. 7, pop. 8) excluded. d.f., degrees of freedom; MS, mean sum of squares; *P < 0.001. Significance levels are based on 1023 permutations. (f) For comparison, values of the widespread congener A. alpina taken from Schönswetter et al. (unpubl.)

 Source of variation d.f. MS Variance components Percentage of variation accounted for Fst 
(a) A. brevis among populations 16.00 0.49 14.37 0.144 
 within populations 16 46.40 2.90 85.63  
(b) A. wulfeniana among populations 438.53 11.90 60.72 0.607* 
 within populations 29 223.25 7.70 39.28  
(c) A. brevis among groups 6.80 0.22 6.36 0.162* 
 among populations 9.20 0.34 9.83  
 within populations 16 46.40 2.90 83.82  
(d) A. wulfeniana among groups 259.69 7.63 35.57 0.641* 
 among populations 178.84 6.13 28.56  
 within populations 29 223.25 7.70 35.87  
(e) A. wulfeniana (excl. Dolomites) among groups 80.45 3.40 18.13 0.469* 
 among populations 136.74 5.41 28.82  
 within populations 21 209.25 9.96 53.06  
(f) A. alpina among populations 52 1843.75 5.55 39.36 0.394* 
 within populations 204 1744.15 8.55 60.64  
 Source of variation d.f. MS Variance components Percentage of variation accounted for Fst 
(a) A. brevis among populations 16.00 0.49 14.37 0.144 
 within populations 16 46.40 2.90 85.63  
(b) A. wulfeniana among populations 438.53 11.90 60.72 0.607* 
 within populations 29 223.25 7.70 39.28  
(c) A. brevis among groups 6.80 0.22 6.36 0.162* 
 among populations 9.20 0.34 9.83  
 within populations 16 46.40 2.90 83.82  
(d) A. wulfeniana among groups 259.69 7.63 35.57 0.641* 
 among populations 178.84 6.13 28.56  
 within populations 29 223.25 7.70 35.87  
(e) A. wulfeniana (excl. Dolomites) among groups 80.45 3.40 18.13 0.469* 
 among populations 136.74 5.41 28.82  
 within populations 21 209.25 9.96 53.06  
(f) A. alpina among populations 52 1843.75 5.55 39.36 0.394* 
 within populations 204 1744.15 8.55 60.64  

A clear difference between the two species is found in mean Jaccard distances between pairs of populations (Table 4). The mean is 0.07 for A. brevis and 0.32 for A. wulfeniana. If the disjunct populations from the Dolomites are excluded, the value decreases to 0.29.

Table 4.

Mean Jaccard distances within and between pairs of populations of Androsace wulfeniana and A. brevis

wulfeniana pop. 1 pop. 2 pop. 3 pop. 4 pop. 5 pop. 6 pop. 7 pop. 8 
pop. 1 0.213 0.278 0.321 0.299 0.319 0.311 0.400 0.417 
pop. 2  0.163 0.313 0.312 0.300 0.276 0.390 0.372 
pop. 3   0.151 0.312 0.224 0.218 0.303 0.345 
pop. 4    0.150 0.318 0.299 0.407 0.419 
pop. 5     0.217 0.219 0.328 0.355 
pop. 6      0.173 0.322 0.322 
pop. 7       0.026 0.177 
pop. 8        0.042 
brevis pop. 13 pop. 14 pop. 15 pop. 16     
pop. 13 0.08 0.09 0.09 0.08     
pop. 14  0.05 0.05 0.04     
pop. 15   0.06 0.05     
pop. 16    0.04     
wulfeniana pop. 1 pop. 2 pop. 3 pop. 4 pop. 5 pop. 6 pop. 7 pop. 8 
pop. 1 0.213 0.278 0.321 0.299 0.319 0.311 0.400 0.417 
pop. 2  0.163 0.313 0.312 0.300 0.276 0.390 0.372 
pop. 3   0.151 0.312 0.224 0.218 0.303 0.345 
pop. 4    0.150 0.318 0.299 0.407 0.419 
pop. 5     0.217 0.219 0.328 0.355 
pop. 6      0.173 0.322 0.322 
pop. 7       0.026 0.177 
pop. 8        0.042 
brevis pop. 13 pop. 14 pop. 15 pop. 16     
pop. 13 0.08 0.09 0.09 0.08     
pop. 14  0.05 0.05 0.04     
pop. 15   0.06 0.05     
pop. 16    0.04     

A Principal Coordinate Analysis (PCoA) calculated for A. wulfeniana visualizes the drastic differences between the three geographical groups (explanation of the total variance: 1st factor 31.4%, 2nd factor 27.6%, 3rd factor 22.5%; Fig. 4). The populations from the Dolomites (pop. 7, pop. 8) are clearly separated from all other populations, but within the main distribution area in the easternmost Alps there is a pronounced differentiation in a western (pop. 3, pop. 5, pop. 6) and an eastern (pop. 1, pop. 2, pop. 4) group.

Figure 4.

Principal Coordinate Analysis (PCoA) of all investigated individuals of Androsace wulfeniana. The two axes explain 70.8% (factor 1) and 7.6% (factor 2) of the overall variation. Squares, Dolomites; triangles, Gurktaler Alpen and western Niedere Tauern; circles, Zirbitzkogel and eastern Niedere Tauern; numbers refer to populations listed in Table 1.

Figure 4.

Principal Coordinate Analysis (PCoA) of all investigated individuals of Androsace wulfeniana. The two axes explain 70.8% (factor 1) and 7.6% (factor 2) of the overall variation. Squares, Dolomites; triangles, Gurktaler Alpen and western Niedere Tauern; circles, Zirbitzkogel and eastern Niedere Tauern; numbers refer to populations listed in Table 1.

DISCUSSION

As outlined above, the three geographical entities of Androsace brevis, A. wulfeniana in its main distributional area in the eastern Central Alps, and A. wulfeniana in the south‐western Dolomites, differ remarkably regarding the level of genetic variation and its structure.

The current ecological conditions allow no explanation for these differences. Population sizes as well as autecological aspects are very similar in A. brevis and A. wulfeniana: both species are nearly identical in their habitat requirements. Furthermore, in both species very large populations with hundreds of individuals (A. brevis: pop. 16; A. wulfeniana: pop. 5) as well as very small ones (A. brevis: pops. 4 and 7; A. wulfeniana: pops. 13 and 14) were investigated. According to the very sparse statements given in literature, both species are outbreeders (Lüdi, 1927). In the related high‐alpine A. alpina a pollen/ovule ratio ranging between facultative xenogamy and xenogamy (Cruden, 1977) combined with high self‐compatibility was found (Schönswetter et al. unpubl.). Hence, explanations need to be sought in the histories of the ­populations.

Androsace brevis exhibits low genetic variation and lacks any geographical structuring. The largest population known to the authors (pop. 16) is less genetically variable than the very small population of A. wulfeniana on Zirbitzkogel in the eastern Central Alps (pop. 4). Individuals with identical AFLP profiles were not only found within populations (as in A. wulfeniana in the Dolomites), but also c. 70 km apart. These data suggest that this species survived the Pleistocene glaciation in one (or perhaps more) strongly bottlenecked population in a single refugium. From there, it subsequently (re)colonized the currently occupied distributional area.

Androsace wulfeniana in the Dolomites is genetically differentiated from A. wulfeniana in the eastern Central Alps on a high level and possesses several fixed private alleles. Recent phylogeographical AFLP studies on alpine plants (Schönswetter et al., 2002; Tribsch, Schönswetter & Stuessy, 2002) demonstrate that descendants of relatively recent long‐distance dispersals are characterized by extremely low genetic variation in combination with a lack of genetic differentiation (e.g. no private fragments). Thus we suggest that the populations in the Dolomites are not a result of recent long‐distance dispersal but descendants of glacial survivors. Although the genetic variation of A. wulfeniana in the Dolomites is even lower than that of A. brevis, the two populations investigated are clearly separated from each other and each possesses fixed private alleles. This points to in situ survival of (at least) two spatially isolated populations trapped between ice to the north and below 1400 m above sea level (a.s.l.) (Van Husen, 1987) and a limestone barrier to the south. A postglacial range expansion, if it occurred, was obviously very limited.

The highest genetic variation is exhibited by A. wulfeniana in the eastern Central Alps. Although some of the populations investigated are very small (e.g. pop. 4), no individuals with identical AFLP patterns were found. Apart from pops. 5 and 6 from the Gurktaler Alpen, which is the current centre of frequency of A. wulfeniana (Fig. 5), all populations are clearly separated from each other. The populations of the western Niedere Tauern (pop. 3) are more similar to those of the Gurktaler Alpen (pops. 5 and 6) to the south, than to those in the eastern Niedere Tauern (pops. 1 and 2). We propose therefore a postglacial recolonization of the western Niedere Tauern from a refugium in the unglaciated Gurktaler Alpen (Fig. 5). The easternmost population of A. wulfeniana in the eastern Niedere Tauern (pop. 1) and on Zirbitzkogel (pop. 4) are possibly remnants of refugial populations as indicated by the high number of private and fixed private fragments (Table 2). They have probably been reduced to their present very small population sizes in rather recent times, and therefore still exhibit relatively high genetic variability.

Figure 5.

Distributional area of Androsace wulfeniana (circles) in the eastern Central Alps and the maximum extent of the ice‐shield during the last Pleistocene glaciation period (Wuerm; redrawn from Fink & Nagl, 1979). White, glaciated area; grey, unglaciated areas and nunataks. Insert: position of mapped area in Austria. Arrow indicates the direction of recolonization of the western Niedere Tauern from a refugial area in the Gurktaler Alpen.

Figure 5.

Distributional area of Androsace wulfeniana (circles) in the eastern Central Alps and the maximum extent of the ice‐shield during the last Pleistocene glaciation period (Wuerm; redrawn from Fink & Nagl, 1979). White, glaciated area; grey, unglaciated areas and nunataks. Insert: position of mapped area in Austria. Arrow indicates the direction of recolonization of the western Niedere Tauern from a refugial area in the Gurktaler Alpen.

The two refugia in the Southern Alps, which supported A. wulfeniana in the Dolomites and A. brevis, and the refugium in the eastern Central Alps, differ regarding the intensity of Pleistocene glaciation (Jäckli, 1970; Van Husen, 1987). The eastern Central Alps provided, due to their position on the eastern border of the continuous ice‐shield (Fig. 1), vast unglaciated areas of siliceous bedrock. The siliceous part of Alpi Bergamasche and the southern Dolomites, however, were situated within the ice‐shield (Fig. 1), albeit close to its southern margin. Even though the upper surface of the glaciers was relatively low in this area (c. 1100–1400 m a.s.l.; Jäckli, 1970; Van Husen, 1987), conditions were certainly harsh in these comparatively small areas. Although the distinctness of A. wulfeniana in the Dolomites can be regarded as a strong hint for nunatak‐survival in the south‐western Dolomites, it should be interpreted with caution. Considering the notion of ‘nunataks’ as ice‐free mountain tops protruding from an ice shield, this term could certainly be applied to the mode of glacial survival A. brevis and A. wulfeniana in the Dolomites most possibly have experienced. Misleadingly, the term ‘nunatak survivors’ is usually only used for populations surviving on the highest, most extensively glaciated mountains in the very central parts of the Alps (e.g. Stehlik, Schneller & Bachmann, 2001). In order not to be confused with this controversial phenomenon (reviewed in Stehlik, 2000), we propose this type of less spectacular, but probably much more important, survival more precisely as ‘survival on peripheral nunataks’.

The heterogeneity within the formerly unglaciated eastern Central Alps is reflected by irregularities in distributional patterns of many relict vascular plant species (Schneeweiss & Schönswetter, 1999). An interesting example already analysed with molecular techniques is provided by Cochlearia excelsa (Koch,2002). It is an endemic species of the easternmost Central Alps and only known from two populations in the eastern Niedere Tauern and the Gurktaler Alpen. Enzyme analysis revealed that the populations are differentiated by some private alleles detected in the one from Niedere Tauern, indicating long‐term isolation. Thus, this taxon is a well‐investigated parallel case to Androsace wulfeniana, indicating an ancient split in at least two separate refugia within the easternmost siliceous parts of the Alps.

COMPARISON OF THE TWO RELICTUAL SPECIESANDROSACE BREVISANDA. WULFENIANAWITH THE PIONEER SPECIESA. ALPINA

Androsace alpina is a high alpine to subnival pioneer species endemic to the Alps, where it is widespread and often frequent at higher elevations. A phylogeographical analysis of this taxon has recently been undertaken by Schönswetter et al. (unpubl.). In contrast to A. wulfeniana and A. brevis, which occur in stable habitats, A. alpina often occurs in pioneer assemblages. Accordingly, different strategies and population dynamics and corresponding differences in HSh can be expected. Androsace brevis exhibits generally less intrapopulational genetic variation than A. alpina (Fig. 2). Androsace wulfeniana in the eastern Central Alps, however, exhibits levels of genetic variation similar to A. alpina (Fig. 2). In contrast, the populations of A. wulfeniana in the Dolomites are the least polymorphic in the data set (Fig. 2). The partitioning of the total variation to within population and among population variation of A. alpina is intermediate between A. brevis and A. wulfeniana (Table 3).

These results contradict the hypothesis that species with small ranges generally exhibit lower genetic variation than widespread ones (Hamrick, Linhart & Mitton, 1979; Hamrick & Godt, 1989). Loveless & Hamrick (1984), however, argue, that due to the influence of historical factors and habitat heterogeneity, geographical range is not a good predictor of genetic structure. In a recent review, Gitzendanner & Soltis (2001) found significant, but small, differences in most diversity indices between rare plants and widespread species. Furthermore, the high level of genetic variation in A. wulfeniana allows rejection of the assumption that relict species were not able to extend their distributional area due to genetic depauperation (e.g. Stebbins, 1942). However, in some studies it has been demonstrated that rare species can be as polymorphic, or even more polymorphic, than their widespread congeners (Vogelmann & Gastony, 1987; Ranker, 1994; Young & Brown, 1996).

The high level of genetic variation in the main distributional area of A. wulfeniana indicates the need to search for other causes of the rarity of the species. Autecological requirements, however, do not give any hints as there are no apparent differences either in the floral syndrome or in seed shape or size in A. alpina, which has colonized vast areas of the Alps after the ice age. The types of bedrock to which A. wulfeniana is restricted, as well as its preferred habitats, are widespread and common in the Alps. Furthermore, the species, similar to A. alpina, is apparently a weak competitor and always restricted to microsites with very low vegetation cover. As a consequence, changes in patterns of interspecific competition due to climatic changes or the arrival of new immigrating taxa in the course of floristic exchange during the Pleistocene is likely to have had only minor effect on A. wulfeniana.

The situation in the three species of Androsace discussed here shows clearly that the interpretation of the phenomenon ‘rarity’ must take into account the history of the investigated taxa and the ecological factors leading to their present‐day restriction.

ACKNOWLEDGEMENTS

Funding by the Austrian Science Foundation (FWF, P13874‐Bio) is gratefully acknowledged. Thanks go to all the people who accompanied us in the field or collected samples for us (Karl Hülber, Sonja Latzin, Luise Schratt‐Ehrendorfer, Corinna Schmiderer, Erich Sinn, Magdalena Wiedermann, Manuela Wink‐ler). We are indebted to Tod F. Stuessy and an anon‐ymous reviewer for very helpful comments on the manuscript and correcting our English. Thanks also to Michael Barfuß for technical assistance in the laboratory.

REFERENCES

Cruden
RW
.
1977
.
Pollen‐ovule ratios: a conservative indicator of breeding systems in flowering plants
.
Evolution
 
31
:
32
46
.
Fink
J
,
Nagl
H
.
1979
.
Quartäre Sedimente und Formen (Karte II/6)
. In:
Kommision für Raumforschung der österreichischen Akademie der Wissenschaften
, ed.
Atlas der Republik Österreich
 .
Wien
:
Freytag‐Berndt & Ataria.
Franz
WR
.
1988
.
Das Androsacetum wulfenianae Franz 82 ass. nov., eine endemische Pflanzengesellschaft in den Ostalpen
.
Sauteria
 
4
:
71
110
.
Gitzendanner
MA
,
Soltis
PS
.
2001
.
Patterns of genetic variation in rare and widespread plant congeners
.
American Journal of Botany
 
87
:
783
792
.
Hamrick
JL
,
Godt
MJW
.
1989
.
Allozyme diversity in plant species
. In.
Brown
AHD
,
Clegg
MT
,
Kahler
AL
,
Weir
BS
, eds.
Plant population genetics, breeding and genetic resources
 .
Sunderland, MA
:
Sinauer
43
63
.
Hamrick
JL
,
Linhart
YB
,
Mitton
JB
.
1979
.
Relationships between life history characteristics and electrophoretically detectable genetic variation in plants
.
Annual Review of Ecology and Systematics
 
10
:
173
200
.
Jäckli
H
.
1970
.
Die Schweiz zur letzten Eiszeit
. In:
Redaktion Atlas der Schweiz, Eidgenössische Technische Hochschule Zürich
ed.
Atlas der Schweiz.
 
Wabern‐Bern
:
Eidgenössiche Landestopographie.
Käsermann
C
,
Moser
DM
.
1999.
Merkblätter Artenschutz – Blütenpflanzen und Farne. Stand: Oktober 1999
 .
Bern
:
Bundesamt für Umwelt, Wald und Landschaft.
Koch
M
.
2002
.
Genetic differentiation and speciation in prealpine Cochlearia (Brassicaceae): Allohexaploid Cochlearia bavarica Vogt compared to its diploid ancestor Cochlearia pyrenaica L. in Germany and Austria
 .
Plant Systematics and Evolution
 
232
:
35
49
.
Legendre
P
,
Legendre
L
.
1998.
Numerical ecology
 .
Amsterdam
:
Elsevier
.
Loveless
MD
,
Hamrick
JL
.
1984
.
Ecological determinants of genetic structures in plant populations
.
Annual Review of Ecology and Systematics
 
15
:
65
95
.
Lüdi
W
.
1927.
Primulaceae
. In.
Hegi
,
G
, eds.
Illustrierte Flora von Mitteleuropa
 , Vol.
3
.
Wien
:
A. Pichler's Witwe & Sohn
,
1715
1877
.
Merxmüller
H
.
1952
.
Untersuchungen zur Sippengliederung und Arealbildung in den Alpen. Teil 1
.
Jahrbuch des Vereins zum Schutze der Alpenpflanzen und Tiere
 
17
:
96
133
.
Merxmüller
H
.
1953
.
Untersuchungen zur Sippengliederung und Arealbildung in den Alpen. Teil 2
.
Jahrbuch des Vereins zum Schutze der Alpenpflanzen und Tiere
 
18
:
135
158
.
Merxmüller
H
.
1954
.
Untersuchungen zur Sippengliederung und Arealbildung in den Alpen. Teil 3
.
Jahrbuch des Vereins zum Schutze der Alpenpflanzen und Tiere
 
19
:
97
139
.
Meusel
H
,
Jäger
E
,
Rauschert
S
,
Weinert
E
.
1978.
Verglei‐chende Chorologie der Zentraleuropäischen Flora. Band II
 .
Jena
:
Gustav Fischer.
Nei
M
,
Li
WH
.
1979
.
Mathematical model for studying genetic variation in terms of restriction endonucleases
.
Proceedings of the National Academy of Sciences, USA
 
186
:
5269
5273
.
Niklfeld
H
.
1972
.
Der niederösterreichische Alpenostrand – ein Glazialrefugium montaner Pflanzensippen
.
Jahrbuch des Vereins zum Schutze der Alpenpflanzen und Tiere
 
37
:
1
54
.
Pawlowski
B
.
1970
.
Remarques sur l'endemisme dans la flore des Alpes et des Carpates
.
Vegetatio
 
21
:
181
243
.
Pitschmann
H
,
Reisigl
H
.
1959
.
Endemische Blütenpflanzen der Südalpen zwischen Luganersee und Etsch
.
Veröffentlichungen des Geobotanischen Institutes der Eidgenössischen Technischen Hochschule Stiftung Rübel in Zürich
 
35
:
44
68
.
Ranker
TA
.
1994
.
Evolution of high genetic variability in the rare Hawaiian fern Adenophorus periens and implications for conservation management
.
Biological Conservation
 
70
:
19
24
.
Schneeweiss
GM
,
Schönswetter
P
.
1999
.
Feinverbreitung, Ökologie und Gesellschaftsanschluß reliktischer Gefäßpflanzen der Niederen Tauern östlich des Sölkpasses (Steiermark, Österreich)
.
Stapfia
 
61
:
1
242
.
Schneider
S
,
Kueffer
JM
,
Roessli
D
,
Excoffier
L
.
1997.
Arlequin 1.1: a software for population genetic analysis.
 
Geneva
:
Genetics and Biometry Laboratory, University of Geneva.
Schönswetter
P
,
Tribsch
A
,
Barfuss
M
,
Niklfeld
H
.
2002
.
Several Pleistocene refugia detected in the high alpine plant Phyteuma globulariifolium in the European Alps
 .
Molecular Ecology
 
11
:
2637
2647
.
Stebbins
GL
.
1942
.
The genetic approach to problems of rare and endemic species
.
Madroño
 
6
:
241
272
.
Stehlik
I
.
2000
.
Nunataks and peripheral refugia for alpine plants during quaternary glaciation in the middle part of the Alps
.
Botanica Helvetica
 
110
:
25
30
.
Stehlik
I
,
Schneller
JJ
,
Bachmann
K
.
2001
.
Resistance or emigration: response of the high‐alpine plant Eritrichum nanum (L.) Gaudin to the ice age within the Central Alps
.
Molecular Ecology
 
10
:
357
370
.
Tribsch
A
,
Schönswetter
P
,
Stuessy
TF
.
2002
.
Saponaria pumila (Caryophyllaceae) and the ice‐age in the Eastern Alps
.
American Journal of Botany
 
89
:
2024
2033
.
Van de Peer
Y
,
De Wachter
R
.
1994
.
TREECON for Windows: a software package for the construction and drawing of evolutionary trees for the Microsoft Windows environment
.
Computer Application for Bioscience
 
10
:
569
570
.
Van Husen
D
.
1987.
Die Ostalpen in den Eiszeiten
 .
Wien
:
Geologische Bundesanstalt.
Vogelmann
JE
,
Gastony
GJ
.
1987
.
Electrophoretic enzyme analysis of North American and eastern Asian populations of Agastache sect. Agastache (Labiatae)
.
American Journal of Botany
 
74
:
385
393
.
Young
AB
,
Brown
AHD
.
1996
.
Comparative population genetic structure of the rare woodland shrub Daviesia suaveolens and its common congener D. mimosoides
.
Conservation Biology
 
10
:
1220
1228
.