Abstract
Using allozymes and microsatellites we have analysed the genetic structure among European populations for several Polytrichum species to infer relevant factors, such as historical events or gene flow, that have shaped their genetic structure. As we observed low levels of genetic differentiation among populations, and no decreasing levels of genetic variation with increasing latitude within most of the examined species, no genetic evidence was obtained for a step-wise recolonization of Europe from southern refugia after the latest glacial period for P. commune, P. uliginosum, P. formosum and P. piliferum. The near absence of population substructuring within these species does indicate that extensive spore dispersal is the most important factor determining the genetic structure among European Polytrichum populations. Gene flow levels have apparently been sufficient to prevent genetic differentiation among populations caused by genetic drift, and to wipe out any genetic structure caused by the postglacial recolonization process. On the other hand, increased genetic differentiation of alpine P. formosum populations suggests that mountain ranges might restrict gene flow significantly among Polytrichum populations. In contrast to most examined Polytrichum species, P. juniperinum showed high levels of genetic differentiation and a profound genetic structure. Assuming that gene flow is not more restricted in P. juniperinum, these findings suggest that this species has recolonized Europe after the latest glacial period from two different refugia, one possibly being the British Isles.
INTRODUCTION
The partitioning of neutral genetic variation among populations is mostly determined by both the level of gene flow and historical patterns of relationship between these populations (Schaal et al., 1998). Traditionally, genetic exchange between populations has been emphasized as the most important determinant of genetic structure. However, classical models describing this structure, F-statistics (Wright, 1951), do not discriminate between recurrent and historical processes that both have shaped genetic structure within species. In the last decade, phylogeography, the study of the relationship between the genealogy of ‘variants’ and their geographical distribution within species, has become a promising field of evolutionary research, and might prove very useful in discriminating between gene flow levels and historical relationships as the main determinant of genetic structure within species (Avise, 2000).
Among the most important historical factors that have influenced the genetic structure of plant species in the temperate zone of Europe are the Quaternary cold periods (Hewitt, 1996). During these periods most species went through several contractions and expansions of geographical range, characterized by extinction of northern populations during glaciation, followed by northward expansion from the refugia in the interglacial periods. During the latest glaciation of Europe, which started ± 113 000 years ago, many species were confined to the southern refugia such as the Iberian peninsula, Italy and the Balkans. After this period, recolonization of northern Europe took place during the present interglacial that began around 13 000 years ago (Webb & Barlein, 1992). This recolonization process implies successive bottlenecks that may have led to clinally decreased levels of genetic variation in northern compared with southern populations and to high levels of genetic differentiation between populations originating from different refugia, as has been observed for several tree species (Abies alba, Picea abies, Fagus sylvatica and Quercus spp.; summarized in Taberlet et al., 1998), heather (Calluna vulagaris; Mahy et al., 1997), Arabidopsis thaliana (Sharbel, Haubold & Mitchell-Olds, 2000) and bryophytes (e.g. Leucodon sciuroides; Cronberg, 2000). On the other hand, if seed or spore dispersal is extensive among (distant) populations, the geographical structure of genetic variation within a species caused by historical events might be nullified by this recurrent gene flow. However, empirical evidence on the extent of gene flow and its influence on the genetic structure of species is inconclusive (Hamrick, 1987), and the impact of recurrent gene flow on the distribution of genetic variation among plant populations is still unresolved (Schaal et al., 1998).
In bryophytes, only a few studies have examined the genetic structure within species over larger geographical areas (e.g. Hofman, 1991; Wyatt, Odrzykoski & Stoneburner, 1992, 1993a,b; Shaw & Schneider, 1995; Shaw, 1995; Derda & Wyatt, 1999a, b; Shaw, 2000), and most of these studies concerned predominantly American populations. Even fewer studies have been performed on European bryophyte populations, mostly covering only a relatively small part of this continent (Cronberg, Molau & Sonesson, 1997; Cronberg, 1998; Van der Velde & Bijlsma, 2000; Van der Velde, Van de Zande & Bijlsma, 2001a), and consequently little is known about the distribution of genetic variation among European bryophyte populations. Moreover, in addition to several modes of asexual propagation (Van der Velde et al., 2001a, b), bryophytes can produce very large numbers of tiny spores, for example the Polytrichum species investigated in this paper can produce over 1 million, 8–15 µm small spores in just a single sporophyte (Touw & Rubers, 1989; Longton, 1997), that potentially can disperse very far (Van Zanten & Pócs, 1981). As most vascular plants generally produce a much lower quantity and heavier seeds, the level of gene flow over large distances, despite the fact that moss spores are haploid, might be magnitudes higher in mosses than in vascular plants. Therefore, we might expect the relative effect of both historical events and gene flow on the genetic structure of bryophytes to be quite different from the relative effect of these factors on the genetic structure of the plant species that have been examined so far (see examples above). Thus, clearly more research is needed to elucidate the genetic structure of bryophyte populations over large geographical scales in Europe.
To examine the effect of both historical factors and gene flow levels on the partitioning and geographical structure of genetic variation among European bryophyte populations, we analysed the genetic structure of five Polytrichum species, P. commune, P. uliginosum, P. piliferum, P. juniperinum and P. formosum, using both allozyme and, for a subset of species, microsatellite markers. We examined Polytrichum populations over a large geographical range (mainly a north–south gradient in western Europe), because a previous study comparing only Danish and Dutch populations, approximately 400 km apart, had revealed low levels of genetic differentiation between populations of these two countries for most examined Polytrichum species (Van der Velde & Bijlsma, 2000; Van der Velde et al., 2001a). We hypothesize that if gene flow is the main determinant of genetic structure within the examined Polytrichum species, we do not expect a clear genetic structure when gene flow is extensive, we expect ‘isolation by distance’ when gene flow decreases with increasing geographical distance between the populations, and greatly significant levels of genetic differentiation between populations when gene flow highly restricted. If the latest glacial period is a main determinant of genetic structure, we may expect strong genetic differentiation between populations that have been recolonized from different refugia. Moreover, if Europe has been recolonized step-wise from these refugia, we might expect reduced levels of gene diversity in northern (recolonized) populations compared to southern (refugia) populations. Examination of populations for five different species from more or less the same geographical range is especially interesting, because concordance among these taxa in their phylogeographical patterns might indicate a common historical factor shaping the genetic structure of the species (Taberlet et al., 1998), as has been shown for populations of several different species from marine, coastal and freshwater habitats in the south-eastern USA (Avise, 1992).
MATERIAL AND METHODS
SAMPLING, ALLOZYME ELECTROPHORESIS AND MICROSATELLITE ANALYSIS
From each of 33 different European locations (Fig. 1) 9–77 samples of Polytrichum formosum, P. juniperinum, P. piliferum, P. commune and/or P. uliginosum were either collected by ourselves or kindly provided by others. For nomenclature of the latter two taxa see Bijlsma et al. (2000). To ensure that within each location different individuals (genets) were sampled, samples were taken from clearly separated discrete moss-cushions at least 2 m apart, and from each cushion one sample was taken as a small clump of neighbouring shoots (see also Bijlsma et al., 2000; Van der Velde & Bijlsma, 2000). For each species 8–17 different populations, indicated in Table 1, were analysed for allozymes. For the allozyme study we used the same enzyme systems as Bijlsma et al. (2000) and Van der Velde & Bijlsma (2000), except that owing to inconsistency of the electrophoretic patterns the loci Acph and Mdh-3 were not included in this study. Detailed information about electrophoretic procedures and interpretation of the banding patterns can also be found in those papers. As for P. formosum microsatellite primers previously had been developed (Van der Velde et al., 2000, 2001a), most populations of this species (Table 1) were also analysed for eight polymorphic microsatellite loci (F-3, F-4, F-5, F-6, F-9, F-11, F-12 and F-13). For detailed information about the microsatellite loci, the method of microsatellite analysis and scoring of the banding patterns see Van der Velde et al. (2000, 2001a). As some of the microsatellite primersets developed for P. formosum also amplified a product in some of the other examined Polytrichum species (Van der Velde & Bijlsma, 2001), a number of populations of both P. commune and P. juniperinum (Table 1) were also analysed for six microsatellite loci (F-1, F-7, F-11, F-12, F-13 and F-21, and F-5a, F-5b, F-11, F-23, F-25 and F-26, respectively). To get better annealing of microsatellite primers in the latter two species, the annealing temperature was decreased by 5°C for these species for part of these primersets.
Map of sampled locations in Western Europe. Numbers refer to the populations given in Table 1.
Map of sampled locations in Western Europe. Numbers refer to the populations given in Table 1.
Names, country and coordinates of the analysed populations for each of the five examined Polytrichum species. Uli = P. uliginosum, Com = P. commune, For = P. formosum, Jun = P. juniperinum and Pil = P. piliferum. SF = Finland, N = Norway, DK = Denmark, NL = the Netherlands, D = Germany, B = Belgium, A = Austria, I = Italy, F = France, E = Spain, IRL = Ireland, GB = Great Britain. Numbers in the table indicate sample sizes of populations used for the allozyme analysis; bold numbers indicate sample sizes of populations used for the microsatellite analysis. Population numbers correspond to numbers shown in Fig. 1. Populations 24–29 are alpine populations sampled at altitudes> 1000 m, whereas the other populations are lowland populations sampled at altitudes < 500 m
| Population | Coordinates | Speciesh | |||||
|---|---|---|---|---|---|---|---|
| Uli | Com | For | Jun | Pil | |||
| 1 | Kirkkonummi (SF) | 60°05′N,24°24′E | 20 | 19/19 | – | 21/21 | 21 |
| 2 | Risør I (N) | 58°44′N,09°16′E | 9 | 27/27 | 30/29 | 23 | 25 |
| 3 | Risør II (N) | 58°44′N,09°16′E | – | – | 15 | – | – |
| 4 | Gjerrild (DK) | 56°32′N,10°44′E | – | 77 | – | – | – |
| 5 | Fjellerup (DK) | 56°31′N,10°35′E | – | – | – | 13/7 | – |
| 6 | Moesgård (DK) | 56°07′N,10°13′E | – | – | 25/32 | – | – |
| 7 | Skov Mølle (DK) | 56°07′N,10°13′E | – | – | 25/33 | – | – |
| 8 | Hjøllund (DK) | 56°04′N,09°23′E | – | – | – | 16/11 | 23 |
| 9 | Norlund I (DK) | 56°02′N,09°07′E | – | 38 | – | – | 25 |
| 10 | Norlund II (DK) | 56°02′N,09°07′E | – | 22 | – | – | – |
| 11 | Voerlade (DK) | 56°02′N,09°37′E | – | – | 40/27 | – | – |
| 12 | Hallund (DK) | 56°01′N,09°07′E | 67 | – | – | – | – |
| 13 | Grene Å (DK) | 55°43′N,09°05′E | 66 | – | – | – | – |
| 14 | Foxhol (NL) | 53°11′N,06°40′E | – | 41 | – | 25/30 | 25 |
| 15 | N.W.-kanaal (NL) | 53°09′N,06°33′E | – | 52 | – | – | – |
| 16 | Appelbergen (NL) | 53°08′N,06°34′E | – | – | 71/27 | – | – |
| 17 | De Punt (NL) | 53°07′N,06°35′E | – | – | – | 25 | 25 |
| 18 | Kniphorstbos (NL) | 53°03′N,06°36′E | 55 | 41/27 | 52/31 | – | – |
| 19 | Anloo (NL) | 53°03′N,06°42′E | 32 | – | – | – | – |
| 20 | Grolloo (NL) | 52°56′N,06°35′E | 33 | – | 29/31 | 25 | 25 |
| 21 | Bornia (NL) | 52°04′N,05°18′E | – | – | – | 18 | 25 |
| 22 | Lobenstein (D) | 50°32′N,11°36′E | – | – | 55/30 | – | – |
| 23 | Neufchateau (B) | 49°09′N,05°33′E | – | – | 28/27 | – | – |
| 24 | Annaberg (A) | 47°54′N,15°17′E | – | – | 30/30 | – | – |
| 25 | Fügen (A) | 47°23′N,11°52′E | – | – | 29 | – | – |
| 26 | Gerloß (A) | 47°16′N,12°05′E | – | 27/27 | – | 10 | – |
| 27 | Predazzo (I) | 46°18′N,11°45′E | – | – | 17/16 | 22/21 | – |
| 28 | Mont Louis (F) | 42°30′N,02°07′E | – | – | 25/24 | 13 | – |
| 29 | Lleida (E) | 41°41′N,00°32′E | 26 | – | 12/12 | – | – |
| 30 | Letterfrack (IRL) | 53°33′N,09°57′W | 13 | – | – | – | – |
| 31 | Dolgellau (GB) | 52°40′N,04°05′W | 9 | – | 13/15 | – | – |
| 32 | Chesterfield (GB) | 53°15′N,01°29′W | 12 | 9 | – | – | – |
| 33 | Matlock (GB) | 53°08′N,01°31′W | – | – | 18 | – | – |
| Population | Coordinates | Speciesh | |||||
|---|---|---|---|---|---|---|---|
| Uli | Com | For | Jun | Pil | |||
| 1 | Kirkkonummi (SF) | 60°05′N,24°24′E | 20 | 19/19 | – | 21/21 | 21 |
| 2 | Risør I (N) | 58°44′N,09°16′E | 9 | 27/27 | 30/29 | 23 | 25 |
| 3 | Risør II (N) | 58°44′N,09°16′E | – | – | 15 | – | – |
| 4 | Gjerrild (DK) | 56°32′N,10°44′E | – | 77 | – | – | – |
| 5 | Fjellerup (DK) | 56°31′N,10°35′E | – | – | – | 13/7 | – |
| 6 | Moesgård (DK) | 56°07′N,10°13′E | – | – | 25/32 | – | – |
| 7 | Skov Mølle (DK) | 56°07′N,10°13′E | – | – | 25/33 | – | – |
| 8 | Hjøllund (DK) | 56°04′N,09°23′E | – | – | – | 16/11 | 23 |
| 9 | Norlund I (DK) | 56°02′N,09°07′E | – | 38 | – | – | 25 |
| 10 | Norlund II (DK) | 56°02′N,09°07′E | – | 22 | – | – | – |
| 11 | Voerlade (DK) | 56°02′N,09°37′E | – | – | 40/27 | – | – |
| 12 | Hallund (DK) | 56°01′N,09°07′E | 67 | – | – | – | – |
| 13 | Grene Å (DK) | 55°43′N,09°05′E | 66 | – | – | – | – |
| 14 | Foxhol (NL) | 53°11′N,06°40′E | – | 41 | – | 25/30 | 25 |
| 15 | N.W.-kanaal (NL) | 53°09′N,06°33′E | – | 52 | – | – | – |
| 16 | Appelbergen (NL) | 53°08′N,06°34′E | – | – | 71/27 | – | – |
| 17 | De Punt (NL) | 53°07′N,06°35′E | – | – | – | 25 | 25 |
| 18 | Kniphorstbos (NL) | 53°03′N,06°36′E | 55 | 41/27 | 52/31 | – | – |
| 19 | Anloo (NL) | 53°03′N,06°42′E | 32 | – | – | – | – |
| 20 | Grolloo (NL) | 52°56′N,06°35′E | 33 | – | 29/31 | 25 | 25 |
| 21 | Bornia (NL) | 52°04′N,05°18′E | – | – | – | 18 | 25 |
| 22 | Lobenstein (D) | 50°32′N,11°36′E | – | – | 55/30 | – | – |
| 23 | Neufchateau (B) | 49°09′N,05°33′E | – | – | 28/27 | – | – |
| 24 | Annaberg (A) | 47°54′N,15°17′E | – | – | 30/30 | – | – |
| 25 | Fügen (A) | 47°23′N,11°52′E | – | – | 29 | – | – |
| 26 | Gerloß (A) | 47°16′N,12°05′E | – | 27/27 | – | 10 | – |
| 27 | Predazzo (I) | 46°18′N,11°45′E | – | – | 17/16 | 22/21 | – |
| 28 | Mont Louis (F) | 42°30′N,02°07′E | – | – | 25/24 | 13 | – |
| 29 | Lleida (E) | 41°41′N,00°32′E | 26 | – | 12/12 | – | – |
| 30 | Letterfrack (IRL) | 53°33′N,09°57′W | 13 | – | – | – | – |
| 31 | Dolgellau (GB) | 52°40′N,04°05′W | 9 | – | 13/15 | – | – |
| 32 | Chesterfield (GB) | 53°15′N,01°29′W | 12 | 9 | – | – | – |
| 33 | Matlock (GB) | 53°08′N,01°31′W | – | – | 18 | – | – |
Names, country and coordinates of the analysed populations for each of the five examined Polytrichum species. Uli = P. uliginosum, Com = P. commune, For = P. formosum, Jun = P. juniperinum and Pil = P. piliferum. SF = Finland, N = Norway, DK = Denmark, NL = the Netherlands, D = Germany, B = Belgium, A = Austria, I = Italy, F = France, E = Spain, IRL = Ireland, GB = Great Britain. Numbers in the table indicate sample sizes of populations used for the allozyme analysis; bold numbers indicate sample sizes of populations used for the microsatellite analysis. Population numbers correspond to numbers shown in Fig. 1. Populations 24–29 are alpine populations sampled at altitudes> 1000 m, whereas the other populations are lowland populations sampled at altitudes < 500 m
| Population | Coordinates | Speciesh | |||||
|---|---|---|---|---|---|---|---|
| Uli | Com | For | Jun | Pil | |||
| 1 | Kirkkonummi (SF) | 60°05′N,24°24′E | 20 | 19/19 | – | 21/21 | 21 |
| 2 | Risør I (N) | 58°44′N,09°16′E | 9 | 27/27 | 30/29 | 23 | 25 |
| 3 | Risør II (N) | 58°44′N,09°16′E | – | – | 15 | – | – |
| 4 | Gjerrild (DK) | 56°32′N,10°44′E | – | 77 | – | – | – |
| 5 | Fjellerup (DK) | 56°31′N,10°35′E | – | – | – | 13/7 | – |
| 6 | Moesgård (DK) | 56°07′N,10°13′E | – | – | 25/32 | – | – |
| 7 | Skov Mølle (DK) | 56°07′N,10°13′E | – | – | 25/33 | – | – |
| 8 | Hjøllund (DK) | 56°04′N,09°23′E | – | – | – | 16/11 | 23 |
| 9 | Norlund I (DK) | 56°02′N,09°07′E | – | 38 | – | – | 25 |
| 10 | Norlund II (DK) | 56°02′N,09°07′E | – | 22 | – | – | – |
| 11 | Voerlade (DK) | 56°02′N,09°37′E | – | – | 40/27 | – | – |
| 12 | Hallund (DK) | 56°01′N,09°07′E | 67 | – | – | – | – |
| 13 | Grene Å (DK) | 55°43′N,09°05′E | 66 | – | – | – | – |
| 14 | Foxhol (NL) | 53°11′N,06°40′E | – | 41 | – | 25/30 | 25 |
| 15 | N.W.-kanaal (NL) | 53°09′N,06°33′E | – | 52 | – | – | – |
| 16 | Appelbergen (NL) | 53°08′N,06°34′E | – | – | 71/27 | – | – |
| 17 | De Punt (NL) | 53°07′N,06°35′E | – | – | – | 25 | 25 |
| 18 | Kniphorstbos (NL) | 53°03′N,06°36′E | 55 | 41/27 | 52/31 | – | – |
| 19 | Anloo (NL) | 53°03′N,06°42′E | 32 | – | – | – | – |
| 20 | Grolloo (NL) | 52°56′N,06°35′E | 33 | – | 29/31 | 25 | 25 |
| 21 | Bornia (NL) | 52°04′N,05°18′E | – | – | – | 18 | 25 |
| 22 | Lobenstein (D) | 50°32′N,11°36′E | – | – | 55/30 | – | – |
| 23 | Neufchateau (B) | 49°09′N,05°33′E | – | – | 28/27 | – | – |
| 24 | Annaberg (A) | 47°54′N,15°17′E | – | – | 30/30 | – | – |
| 25 | Fügen (A) | 47°23′N,11°52′E | – | – | 29 | – | – |
| 26 | Gerloß (A) | 47°16′N,12°05′E | – | 27/27 | – | 10 | – |
| 27 | Predazzo (I) | 46°18′N,11°45′E | – | – | 17/16 | 22/21 | – |
| 28 | Mont Louis (F) | 42°30′N,02°07′E | – | – | 25/24 | 13 | – |
| 29 | Lleida (E) | 41°41′N,00°32′E | 26 | – | 12/12 | – | – |
| 30 | Letterfrack (IRL) | 53°33′N,09°57′W | 13 | – | – | – | – |
| 31 | Dolgellau (GB) | 52°40′N,04°05′W | 9 | – | 13/15 | – | – |
| 32 | Chesterfield (GB) | 53°15′N,01°29′W | 12 | 9 | – | – | – |
| 33 | Matlock (GB) | 53°08′N,01°31′W | – | – | 18 | – | – |
| Population | Coordinates | Speciesh | |||||
|---|---|---|---|---|---|---|---|
| Uli | Com | For | Jun | Pil | |||
| 1 | Kirkkonummi (SF) | 60°05′N,24°24′E | 20 | 19/19 | – | 21/21 | 21 |
| 2 | Risør I (N) | 58°44′N,09°16′E | 9 | 27/27 | 30/29 | 23 | 25 |
| 3 | Risør II (N) | 58°44′N,09°16′E | – | – | 15 | – | – |
| 4 | Gjerrild (DK) | 56°32′N,10°44′E | – | 77 | – | – | – |
| 5 | Fjellerup (DK) | 56°31′N,10°35′E | – | – | – | 13/7 | – |
| 6 | Moesgård (DK) | 56°07′N,10°13′E | – | – | 25/32 | – | – |
| 7 | Skov Mølle (DK) | 56°07′N,10°13′E | – | – | 25/33 | – | – |
| 8 | Hjøllund (DK) | 56°04′N,09°23′E | – | – | – | 16/11 | 23 |
| 9 | Norlund I (DK) | 56°02′N,09°07′E | – | 38 | – | – | 25 |
| 10 | Norlund II (DK) | 56°02′N,09°07′E | – | 22 | – | – | – |
| 11 | Voerlade (DK) | 56°02′N,09°37′E | – | – | 40/27 | – | – |
| 12 | Hallund (DK) | 56°01′N,09°07′E | 67 | – | – | – | – |
| 13 | Grene Å (DK) | 55°43′N,09°05′E | 66 | – | – | – | – |
| 14 | Foxhol (NL) | 53°11′N,06°40′E | – | 41 | – | 25/30 | 25 |
| 15 | N.W.-kanaal (NL) | 53°09′N,06°33′E | – | 52 | – | – | – |
| 16 | Appelbergen (NL) | 53°08′N,06°34′E | – | – | 71/27 | – | – |
| 17 | De Punt (NL) | 53°07′N,06°35′E | – | – | – | 25 | 25 |
| 18 | Kniphorstbos (NL) | 53°03′N,06°36′E | 55 | 41/27 | 52/31 | – | – |
| 19 | Anloo (NL) | 53°03′N,06°42′E | 32 | – | – | – | – |
| 20 | Grolloo (NL) | 52°56′N,06°35′E | 33 | – | 29/31 | 25 | 25 |
| 21 | Bornia (NL) | 52°04′N,05°18′E | – | – | – | 18 | 25 |
| 22 | Lobenstein (D) | 50°32′N,11°36′E | – | – | 55/30 | – | – |
| 23 | Neufchateau (B) | 49°09′N,05°33′E | – | – | 28/27 | – | – |
| 24 | Annaberg (A) | 47°54′N,15°17′E | – | – | 30/30 | – | – |
| 25 | Fügen (A) | 47°23′N,11°52′E | – | – | 29 | – | – |
| 26 | Gerloß (A) | 47°16′N,12°05′E | – | 27/27 | – | 10 | – |
| 27 | Predazzo (I) | 46°18′N,11°45′E | – | – | 17/16 | 22/21 | – |
| 28 | Mont Louis (F) | 42°30′N,02°07′E | – | – | 25/24 | 13 | – |
| 29 | Lleida (E) | 41°41′N,00°32′E | 26 | – | 12/12 | – | – |
| 30 | Letterfrack (IRL) | 53°33′N,09°57′W | 13 | – | – | – | – |
| 31 | Dolgellau (GB) | 52°40′N,04°05′W | 9 | – | 13/15 | – | – |
| 32 | Chesterfield (GB) | 53°15′N,01°29′W | 12 | 9 | – | – | – |
| 33 | Matlock (GB) | 53°08′N,01°31′W | – | – | 18 | – | – |
STATISTICAL ANALYSIS
In this paper we have analysed both the level and the distribution of genetic variation within species: mean number of alleles per locus (A), percentage polymorphic loci (P; a locus is considered polymorphic if more than one allele was detected, which given the sample sizes means that the frequency of the rarest allele is at least 1%) and mean gene diversity (HS; Nei, 1987) were calculated as measures of genetic variability within populations with the computer program BIOSYS-1 (Swofford & Selander, 1981), whereas FST and RST were calculated as measures of genetic differentiation among populations with the computer program ARLEQUIN (Schneider et al., 1997). The genetic structure within species was determined by constructing consensus UPGMA trees based on Nei's (1987) unbiased genetic distance (D) with the computer program PHYLIP (Felsenstein, 1993). We further investigated whether the level of genetic differentiation among populations was influenced by the geographical distance between them, resulting from reduced levels of gene flow between distant compared with nearby populations (isolation by distance). This was done by correlating the level of genetic differentiation (FST and RST) with geographical distance between all pairs of populations within each of the examined species using the Mantel-test of the computer program GENEPOP (Raymond & Rousset, 1995; Rousset, 1997). Using the latitude and longitude of each population, the ‘Great Circle Distance’ was calculated between all pairs of populations for each of the examined species as a measure of geographical distance between these populations.
RESULTS
LEVEL OF GENETIC VARIATION
Genetic variation was determined at 14 putative allozyme loci for populations of P. commune, P. uliginosum and P. formosum, and at 20 putative allozyme loci for P. juniperinum and P. piliferum. On average, low levels of genetic variability (A, P and HS) were observed within populations of P. commune, P. uliginosum and P. formosum. The mean expected heterozygosities (HS-values) ranged from 0.004 (Appelbergen-NL, P. formosum) to 0.078 (Gerloβ-A, P. commune) within these populations (Table 2). Significantly higher levels of allozyme variation were observed within P. piliferum and P. juniperinum populations (Table 2; Mann–Whitney U-test, P < 0.001). For these populations the HS-values ranged from 0.032 (Gerloβ-A, P. juniperinum) to 0.156 (Foxhol-NL, P. juniperinum).
Number of populations analysed (n), mean sample size per population (N), estimates of genetic variation (A, P and HS; averaged over all populations per species) and estimates of genetic differentiation (FST and RST) within and among European populations of five different Polytrichum species using allozymes (top) or microsatellites (bottom). A = mean number of alleles per locus, P = mean percentage polymorphic loci, HS = mean gene diversity within populations and FST or RST = fraction of the total genetic variation that can be attributed to variation between populations. Deviations of FST and RST from zero are tested for significance by permutation and if significant are denoted with an asterisk. Values in parentheses indicate the range of values observed for the parameters among the examined populations
| Species | n | N | A | P | HS | FST | RST |
|---|---|---|---|---|---|---|---|
| P. uliginosum | 11 | 31.9 | 1.2 (1.1–1.4) | 16.3 (7.1–35.7) | 0.030 (0.005–0.076) | 0.091* | – |
| P. commune | 10 | 35.5 | 1.2 (1.1–1.5) | 17.2 (7.1–42.9) | 0.037 (0.017–0.078) | 0.070* | – |
| P. formosum | 17 | 30.1 | 1.1 (1.1–1.2) | 9.1 (7.1–21.4) | 0.022 (0.004–0.042) | 0.395* | – |
| P. juniperinum | 11 | 29.2 | 1.4 (1.1–1.7) | 28.6 (10.0–55.0) | 0.102 (0.032–0.156) | 0.341* | – |
| P. piliferum | x2007;8 | 24.3 | 1.4 (1.3–1.5) | 21.6 (15.0–30.0) | 0.090 (0.057–0.129) | 0.087* | – |
| P. commune | 4 | 24.3 | 5.0 (4.2–6.2) | 100 | 0.584 (0.516–0.643) | 0.047* | 0.022 |
| P. formosum | 14 | 26.1 | 4.0 (2.3–4.6) | 99.6 (87.5–100) | 0.534 (0.346–0.604) | 0.047* | 0.062* |
| P. juniperinum | 4 | 21.3 | 3.6 (2.7–4.5) | 87.2 (83.3–100) | 0.400 (0.333–0.431) | 0.167* | 0.026 |
| Species | n | N | A | P | HS | FST | RST |
|---|---|---|---|---|---|---|---|
| P. uliginosum | 11 | 31.9 | 1.2 (1.1–1.4) | 16.3 (7.1–35.7) | 0.030 (0.005–0.076) | 0.091* | – |
| P. commune | 10 | 35.5 | 1.2 (1.1–1.5) | 17.2 (7.1–42.9) | 0.037 (0.017–0.078) | 0.070* | – |
| P. formosum | 17 | 30.1 | 1.1 (1.1–1.2) | 9.1 (7.1–21.4) | 0.022 (0.004–0.042) | 0.395* | – |
| P. juniperinum | 11 | 29.2 | 1.4 (1.1–1.7) | 28.6 (10.0–55.0) | 0.102 (0.032–0.156) | 0.341* | – |
| P. piliferum | x2007;8 | 24.3 | 1.4 (1.3–1.5) | 21.6 (15.0–30.0) | 0.090 (0.057–0.129) | 0.087* | – |
| P. commune | 4 | 24.3 | 5.0 (4.2–6.2) | 100 | 0.584 (0.516–0.643) | 0.047* | 0.022 |
| P. formosum | 14 | 26.1 | 4.0 (2.3–4.6) | 99.6 (87.5–100) | 0.534 (0.346–0.604) | 0.047* | 0.062* |
| P. juniperinum | 4 | 21.3 | 3.6 (2.7–4.5) | 87.2 (83.3–100) | 0.400 (0.333–0.431) | 0.167* | 0.026 |
Number of populations analysed (n), mean sample size per population (N), estimates of genetic variation (A, P and HS; averaged over all populations per species) and estimates of genetic differentiation (FST and RST) within and among European populations of five different Polytrichum species using allozymes (top) or microsatellites (bottom). A = mean number of alleles per locus, P = mean percentage polymorphic loci, HS = mean gene diversity within populations and FST or RST = fraction of the total genetic variation that can be attributed to variation between populations. Deviations of FST and RST from zero are tested for significance by permutation and if significant are denoted with an asterisk. Values in parentheses indicate the range of values observed for the parameters among the examined populations
| Species | n | N | A | P | HS | FST | RST |
|---|---|---|---|---|---|---|---|
| P. uliginosum | 11 | 31.9 | 1.2 (1.1–1.4) | 16.3 (7.1–35.7) | 0.030 (0.005–0.076) | 0.091* | – |
| P. commune | 10 | 35.5 | 1.2 (1.1–1.5) | 17.2 (7.1–42.9) | 0.037 (0.017–0.078) | 0.070* | – |
| P. formosum | 17 | 30.1 | 1.1 (1.1–1.2) | 9.1 (7.1–21.4) | 0.022 (0.004–0.042) | 0.395* | – |
| P. juniperinum | 11 | 29.2 | 1.4 (1.1–1.7) | 28.6 (10.0–55.0) | 0.102 (0.032–0.156) | 0.341* | – |
| P. piliferum | x2007;8 | 24.3 | 1.4 (1.3–1.5) | 21.6 (15.0–30.0) | 0.090 (0.057–0.129) | 0.087* | – |
| P. commune | 4 | 24.3 | 5.0 (4.2–6.2) | 100 | 0.584 (0.516–0.643) | 0.047* | 0.022 |
| P. formosum | 14 | 26.1 | 4.0 (2.3–4.6) | 99.6 (87.5–100) | 0.534 (0.346–0.604) | 0.047* | 0.062* |
| P. juniperinum | 4 | 21.3 | 3.6 (2.7–4.5) | 87.2 (83.3–100) | 0.400 (0.333–0.431) | 0.167* | 0.026 |
| Species | n | N | A | P | HS | FST | RST |
|---|---|---|---|---|---|---|---|
| P. uliginosum | 11 | 31.9 | 1.2 (1.1–1.4) | 16.3 (7.1–35.7) | 0.030 (0.005–0.076) | 0.091* | – |
| P. commune | 10 | 35.5 | 1.2 (1.1–1.5) | 17.2 (7.1–42.9) | 0.037 (0.017–0.078) | 0.070* | – |
| P. formosum | 17 | 30.1 | 1.1 (1.1–1.2) | 9.1 (7.1–21.4) | 0.022 (0.004–0.042) | 0.395* | – |
| P. juniperinum | 11 | 29.2 | 1.4 (1.1–1.7) | 28.6 (10.0–55.0) | 0.102 (0.032–0.156) | 0.341* | – |
| P. piliferum | x2007;8 | 24.3 | 1.4 (1.3–1.5) | 21.6 (15.0–30.0) | 0.090 (0.057–0.129) | 0.087* | – |
| P. commune | 4 | 24.3 | 5.0 (4.2–6.2) | 100 | 0.584 (0.516–0.643) | 0.047* | 0.022 |
| P. formosum | 14 | 26.1 | 4.0 (2.3–4.6) | 99.6 (87.5–100) | 0.534 (0.346–0.604) | 0.047* | 0.062* |
| P. juniperinum | 4 | 21.3 | 3.6 (2.7–4.5) | 87.2 (83.3–100) | 0.400 (0.333–0.431) | 0.167* | 0.026 |
Levels of microsatellite variation were observed to be considerably higher than those for allozymes within P. formosum, P. commune and P. juniperinum populations (Table 2). The microsatellite HS-values ranged from 0.333 (Predazzo-I, P. juniperinum) to 0.643 (N.W.-kanaal-NL, P. commune). Contrary to the findings for allozymes, the average levels of microsatellite variation (HS) were significantly higher within P. formosum and P. commune populations than within populations of P. juniperinum (Table 2; Mann–Whitney U-test, P < 0.001).
If historical (step-wise) colonization patterns had been an important factor determining the genetic structure of Polytrichum species, we might expect both the gene diversity level (HS) and the number of alleles per locus (A) to be negatively correlated with latitude. However, no significant negative correlation was observed between either mean HS or mean A within a population with latitude for any of the examined species for both types of genetic markers (data not shown). On the contrary, southern European populations of P. juniperinum (Mont Louis, Predazzo and Gerloß), P. uliginosum (Lleida) and P. formosum (Lleida, Mont Louis, Predazzo and Fügen) tend to have lower HS and A levels than northern European populations for allozymes and also, for those so far examined, for microsatellites. This trend was only significant for HS values and not for A values in P. juniperinum for allozymes (HS = 0.111 vs. 0.059, Mann–Whitney U-test, P = 0.048) and in P. formosum for microsatellites (HS = 0.557 vs. 0.401, Mann–Whitney U-test, P = 0.006). It should be noted, however, that all examined northern European populations are lowland populations, whereas the examined southern European populations are all alpine populations, sampled in the Alps, Dolomites and Pyrenees at altitudes> 1000 m. Therefore, the level of genetic variability within a Polytrichum population might rather be correlated with altitude than with latitude.
DISTRIBUTION OF GENETIC VARIATION AMONG POPULATIONS
Low levels of genetic differentiation, FST-values ranged from 0.070 to 0.091 (Table 2; allozyme data), were observed among populations of P. uliginosum, P. piliferum and P. commune. In contrast, within P. formosum and P. juniperinum relatively high levels of genetic differentiation between populations were observed (Table 2; allozyme data). In the latter two species more than one-third of the total amount of allozyme variation within species could be attributed to variation among populations. However, caution is needed with the FST-value of P. formosum (see also microsatellite data below) because for this species the calculation of FST was mainly based on only one polymorphic allozyme locus.
For microsatellite loci the levels of genetic differentiation (FST and RST) observed among both P. commune and P. juniperinum populations corroborate those observed for allozymes within these species (Table 2). In P. commune both types of markers showed similar low levels of population differentiation (FST = 0.047 and 0.070, respectively), whereas in P. juniperinum microsatellites and allozymes, although the values are quite different, both revealed substantial levels of population differentiation (FST = 0.167 and 0.341, respectively). The fact that for both species microsatellites showed lower levels of genetic differentiation among populations than allozymes may be partly explained by the fact that only a subset of four and five populations of P. commune and P. juniperinum, respectively, have been analysed for microsatellites. In contrast, for P. formosum a substantial difference was observed in the level of population differentiation as determined for either allozymes or microsatellites (FST = 0.395 vs. 0.047, respectively). This is unlikely to be explained by differences in the number of populations analysed, because in this case nearly all populations were analysed for both types of genetic markers.
If isolation by distance is an important factor determining the structure of genetic variation among Polytrichum populations, we can expect the level of genetic differentiation (FST or RST) to increase with increasing geographical distance between populations. Although the correlation between the geographical distance and FST between all pairs of P. commune populations is at the border of significance, a significant positive correlation between the geographical distance and both FST and RST was only observed for P. formosum (Fig. 2; RST-values for microsatellites not depicted). The correlation coefficients indicate that 16–23% of the variation observed in pairwise FST- or RST-values between P. formosum populations is explained by differences in geographical distance between pairs of populations (Fig. 2E,F). However, when these populations were subdivided into lowland (sampled at altitudes <1000 m) and alpine populations (sampled at altitudes>1000 m), the FST- or RST-values between pairs of populations seem rather to be affected by the type of populations compared (i.e. a comparison of lowland–lowland, lowland–alpine or alpine–alpine populations) than by the geographical distance between them (see closed circle, open circle and open triangle signs in Fig. 2E,F). For allozymes, pairwise comparisons between lowland and alpine populations showed significantly higher FST-values than comparisons between lowland populations (Mann–Whitney U-test, P < 0.0001) and alpine populations (Mann–Whitney U-test, P = 0.0004). For microsatellites, pairwise comparisons between lowland and alpine populations also showed significantly higher FST-values than comparisons between lowland populations (Mann–Whitney U-test, P < 0.0001), but they showed significantly lower FST-values than comparisons between alpine populations (Mann–Whitney U-test, P = 0.0128).
Correlations between geographical distance and genetic differentiation (FST) for all pairwise combinations of the examined populations of (A) P. commune, (B) P. uliginosum, (C) P. piliferum, (D) P. juniperinum and (E,F) P. formosum. FST-values are based on allozyme (A–E) or microsatellite (F) data. Correlation coefficients (r) and their significance (P) are also denoted. Different symbols in E and F indicate pairwise comparisons of lowland–lowland (•), lowland–alpine (○) and alpine–alpine (▵) populations. For further explanation see text.
Correlations between geographical distance and genetic differentiation (FST) for all pairwise combinations of the examined populations of (A) P. commune, (B) P. uliginosum, (C) P. piliferum, (D) P. juniperinum and (E,F) P. formosum. FST-values are based on allozyme (A–E) or microsatellite (F) data. Correlation coefficients (r) and their significance (P) are also denoted. Different symbols in E and F indicate pairwise comparisons of lowland–lowland (•), lowland–alpine (○) and alpine–alpine (▵) populations. For further explanation see text.
GEOGRAPHICAL STRUCTURE OF GENETIC VARIATION
By means of a cluster analysis, using Nei's (1987) unbiased genetic distance (D), we examined whether the distribution of genetic variation among populations within each of the examined Polytrichum species was structured according to geographical location. Within Europe the examined populations of P. formosum (Fig. 3E,F) and P. juniperinum (Fig. 3D) showed a genetic structure that seemed to correlate with geographical location, whereas in the other examined species no particular pattern could be discovered (Fig. 3A–C). Although the geographical pattern of genetic variation was different for P. formosum and P. juniperinum, for both species a clear bifurcation of populations was observed (Fig. 3D,E).
Consensus UPGMA trees expressing overall levels of Nei's unbiased genetic distance (D) between populations of (A) P. piliferum, (B) P. uliginosum, (C) P. commune, (D) P. juniperinum and (E,F) P. formosum. Genetic distances are based on allozyme (A–E) or microsatellite data (F). Country codes and numbers correspond to those in Table 1. Numbers in the tree denote bootstrap values (percentage) after 1000 resamplings. Bars in each figure indicate the scaling of genetic distance, note that the scale is larger in D and F. In D, *this bootstrap value becomes 65% when population NL-20 is omitted from the analysis. E: the identical bootstrap values in this tree are explained by the fact that only one of the examined allozyme loci showed substantial variation within P. formosum.
Consensus UPGMA trees expressing overall levels of Nei's unbiased genetic distance (D) between populations of (A) P. piliferum, (B) P. uliginosum, (C) P. commune, (D) P. juniperinum and (E,F) P. formosum. Genetic distances are based on allozyme (A–E) or microsatellite data (F). Country codes and numbers correspond to those in Table 1. Numbers in the tree denote bootstrap values (percentage) after 1000 resamplings. Bars in each figure indicate the scaling of genetic distance, note that the scale is larger in D and F. In D, *this bootstrap value becomes 65% when population NL-20 is omitted from the analysis. E: the identical bootstrap values in this tree are explained by the fact that only one of the examined allozyme loci showed substantial variation within P. formosum.
In P. juniperinum (Fig. 3D) the Danish and Dutch populations were genetically distinct from both the northern European and the southern European populations, which, conspicuously, were clustered together. The division of the European populations into two groups was mainly based on five allozyme loci: Gpi-2, Gpi-3, Idh, Me and Pgm-1. Alleles at these loci were private to either the Danish–Dutch cluster or to the other cluster of populations, and showed relative high frequencies: 0.15–0.54. Additionally, three other alleles at these loci showed also relative high frequencies (0.13–0.68) in one cluster, but were found to be rare (<0.025) in the other.
For P. formosum, populations sampled in the southern half of Europe (Lleida, Mont Louis, Predazzo, Fügen and Neufchateau) formed genetically a distinct group from those sampled in northern Europe (Fig. 3E). However, it should be noted that the absolute level of divergence between these two groups is rather low, as, in addition to three very rare alleles at other loci, only one of the examined allozyme loci (6-Pgd-2) showed substantial levels of variation in this species. In contrast to the allozyme data, the UPGMA cluster analysis based on eight polymorphic microsatellite loci did not reveal a bifurcation of P. formosum populations within Europe (Fig. 3F). However, the latter analysis did reveal that the three examined alpine populations of this species, Lleida, Mont Louis and Predazzo (sampled at altitudes>1000 m), seemed to be moderately isolated from both the other examined populations and each other.
DISCUSSION
GENETIC STRUCTURE AMONG POPULATIONS
For several Polytrichum species, questions like ‘what factors determine the distribution and geographical structure of genetic variation within these species?’, and ‘how can the observed differences in the distribution and geographical structure of genetic variation among the species be explained?’ can be easily answered: there seems to be no apparent structure for P. commune, P. uliginosum and P. piliferum. Therefore, the low FST-values (Table 2), the lack of clustering of regional populations in the UPGMA consensus trees (Fig. 3A–C) and the absence of isolation by distance (Fig. 2A–C) indicate that abundant gene flow is the most important factor determining the genetic structure among populations within these species. The level of spore dispersal apparently has been sufficient to prevent substantial genetic differentiation among populations by genetic drift, and to wipe out an initially present genetic structure that may have resulted from the latest glacial period. In contrast, when compared to our values (FST = 0.070 and 0.087, respectively), studies on the genetic structure among American populations of P. commune and P. piliferum revealed much higher levels of genetic differentiation (FST = 0.51 and 0.72, respectively; Derda & Wyatt, 1999a, b) within these species. This might reflect the different evolutionary histories of the species on the different continents, but may also be (partly) due to differences in methodology or taxonomic treatment of the examined taxa between the studies (see Discussion in Van der Velde & Bijlsma, 2000).
At first sight allozymes indicated a substantial level of genetic differentiation (FST = 0.395) among populations of P. formosum, mainly due to a division between northern and southern European populations (Fig. 3E). However, for allozymes the genetic differentiation within this species is almost totally due to only one polymorphic locus (6Pgd-2). In contrast, the analysis of P. formosum populations for eight polymorphic microsatellite loci showed low levels of genetic differentiation (FST = 0.047), and no such clustering of northern or southern European populations in the UPGMA consensus tree (Fig. 3F). It is therefore likely that the high level of genetic divergence among P. formosum populations as observed for allozymes is biased due to the presence of only one polymorphic locus. One of several possible explanations that might account for this result is clinal selection at the 6Pgd-2 locus of P. formosum, which has also been observed in other plant species (e.g. Silene acaulis; Gehring & Delph, 1999, and references therein). Thus, in spite of the genetic differentiation for the 6Pgd-2 locus in P. formosum we conclude that this species also shows low levels of genetic differentiation for neutral markers, and hence that sufficient gene flow is also a dominant factor explaining the genetic structure observed within this species.
Contrasting all other studied Polytrichum species, P. juniperinum showed both high FST-values and a profound geographically correlated genetic structure among European populations for a considerable number of polymorphic allozyme loci (Table 2, Fig. 3D). Moreover, in contrast to P. formosum, the allozyme data are corroborated by the microsatellite analysis for a subset of these P. juniperinum populations (Table 2; UPGMA consensus tree not shown). This observation, in combination with the high number of allozyme alleles that were (nearly) private to either Danish and Dutch populations or to the populations Risor, Kirkkonummi, Mont Louis, Predazzo and Gerloß, indicates that, compared with the other species, there is much more genetic differentiation among populations of P. juniperinum. This conspicuous difference between P. juniperinum showing strong genetic differentiation among populations and the other examined species revealing low genetic differentiation should either reflect different gene flow levels and/or differences in the evolutionary history between the studied Polytrichum species (Schaal et al., 1998; Taberlet et al., 1998). As spores of P. juniperinum are not larger than those of the other examined species (Touw & Rubers, 1989), lower gene flow levels in P. juniperinum are unlikely to explain the observed high level of genetic differentiation in this species. A more likely explanation therefore is that P. juniperinum differs from the other studied Polytrichum species in its evolutionary history. A possible scenario might be that most Polytrichum species have recolonized Europe from one refugium, whereas P. juniperinum has recolonized Europe from two different refugia. For example, most of the sampled P. juniperinum populations may have been recolonized from one refugium (such as Spain or Italy), whereas the Danish and Dutch populations mainly originate from a different refugium (such as southern England). Comparable patterns have been observed for heather (Calluna vulgaris; Mahy, Ennos & Jacquemart, 1999) and the moss species Sphagnum rubellum (Cronberg, 1998). For both these species populations from the British Isles were genetically differentiated from populations of other parts of the continent (Spanish/French populations of heather and Swedish populations of S. rubellum, respectively). These results were also explained by recolonization of the examined populations from different refugia (Cronberg, 1998), in such a manner that the north-western European populations were recolonized from the (southern) British Isles (Mahy et al., 1999, and references therein). Whether this scenario is also applicable to P. juniperinum has to be tested by including samples from the UK and other unsampled areas in our analysis for this species.
Although the geographical distribution of P. formosum is somewhat more restricted, i.e. confined to the temperate zones of the northern hemisphere, than that of the other examined Polytrichum species, the distributions of the studied Polytrichum species overlap to a great extent within Europe. For this reason, we can assume that they have undergone similar historical events in terms of contraction and expansion during glacial and interglacial periods, and it is therefore not immediately clear why the recolonization patterns of Europe differ between P. juniperinum and the other Polytrichum species. However, differences in life history or ecology of these species may have caused the glacial periods to affect the genetic structure of the studied Polytrichum species differently. For example, a lower tolerance to low temperatures for P. formosum and a higher tolerance to drought for P. juniperinum and P. piliferum may have caused different geographical distributions of refugia for the examined Polytrichum species during the cold and dry glacial periods. Nevertheless, neither cold nor drought resistance is unique to P. juniperinum, therefore these traits do not explain why only P. juniperinum shows a genetic structure that is different from the other Polytrichum species, particularly P. piliferum. Alternatively, as P. juniperinum is the species with the most allozyme variability (Table 2), a low resolving power of this genetic marker in the other examined species might have masked existing genetic differentiation among the populations within these species. However, despite the presence of five substantially polymorphic allozyme loci, P. piliferum does not show sizeable levels of genetic differentiation between locations that show significant levels of differentiation for P. juniperinum. Moreover, as highly polymorphic microsatellite loci did not reveal substantial genetic differentiation among populations of both P. formosum and P. commune (Table 2), we do not think that a low resolving power of the allozymes is responsible for the low levels of genetic differentiation that are observed within most of the examined species. At the moment, we do not know why the evolutionary history of P. juniperinum in Europe might have been different from that of the other Polytrichum species.
ALPINE POPULATIONS OF P. FORMOSUM
Isolation by distance has been reported for many plant species (e.g. Arabidopsis thaliana; Sharbel et al., 2000), indicating that limited dispersal capacity has played a significant role in shaping the genetic structure within these species. In contrast, as we observed only for P. formosum a significant correlation between the geographical distance and the level of genetic differentiation (FST) between all pairs of populations (Fig. 2F), isolation by distance seems to be of limited importance in shaping genetic structure within Polytrichum species. Moreover, when the data of P. formosum are analysed in more detail, it becomes clear that also for this species the level of genetic differentiation between populations is not simply explained by isolation by distance. When populations were divided into lowland and alpine populations, the level of pairwise FST seemed consistently more influenced by the lowland or alpine origin of both populations than by the geographical distance between them (Fig. 2F). A GLM analysis (procedure in SPSS, univariate analysis with FST as dependent variable and ‘geographical distance’ and ‘type of comparison’, i.e. comparison of lowland–lowland, lowland–alpine or alpine–alpine populations, as covariate and fixed factor, respectively) revealed that the pairwise FST-values of Fig. 2(F) were significantly less influenced by the geographical distance between the populations (F = 1.9, P = 0.174) than by the type of comparison of the populations involved (F = 95.9, P < 0.001). As significantly higher levels of population differentiation were observed among alpine (triangles in Fig. 2F) and between alpine and lowland populations (open circles in Fig. 2F) than among lowland populations (closed circles in Fig. 2F), this might indicate that within P. formosum gene flow is more reduced by the mountains between the populations than by large geographical distances. This is also clearly seen in the UPGMA consensus tree based on microsatellites (Fig. 3F), where the three alpine populations (Lleida, Predazzo and Mont Louis) sampled in the mountains of southern Europe were genetically isolated from all other sampled populations and also from each other. Cronberg (1998) reported similar results in a study on two Sphagnum species, in which he observed the level of genetic differentiation among lowland populations to be smaller than among alpine populations, although in his case the higher level of differentiation among alpine populations was rather attributed to founder events due to the rare establishment of new genotypes in alpine populations.
We observed conspicuous differences in level of genetic variation both between species and among populations within species. For allozymes, P. juniperinum and P. piliferum showed significantly higher levels of HS than P. commune, P. uliginosum and P. formosum (HS = 0.102, 0.09 and 0.022–0.037, respectively). These findings confirm the observation of Van der Velde & Bijlsma (2000), who also reported a 3- to 5-fold difference in the level of genetic variation between those two groups of species for a much more limited dataset. As P. commune, P. uliginosum and P. formosum are often found in more stable habitats, like peat-bogs and forests (Touw & Rubers, 1989), where clonal growth might outweigh sexual reproduction, whereas P. piliferum and P. juniperinum are more often observed in exposed and dynamic, sandy habitats (Touw & Rubers, 1989), where sexual reproduction might be more important, this difference in the level of genetic variability between the studied Polytrichum species might be attributed to differences in habitat and life history between these species (Van der Velde & Bijlsma, 2000). Also among populations within species interesting differences in the level of genetic variability have been observed. Alpine populations of P. formosum seem not only more isolated than lowland populations, but they also have lower levels of gene diversity than lowland populations (microsatellites: HS = 0.56 vs. 0.40). The latter is also observed for comparisons of lowland and alpine populations of P. uliginosum, P. juniperinum (this study) and two Sphagnum species (Cronberg, 1998). Although others have reported similar or even higher levels of genetic variation in alpine compared with lowland populations for both other mosses (Plagiomnium ciliare–Wyatt, Odrzykoski & Stoneburner, 1989; Hylocomium splendens–Cronberg et al., 1997) and higher plant species (Picea abies–Gugerli et al., 2001; Campanula rotundifolia–Bingham & Ranker, 2000), our findings are consistent with the general view that owing to the more harsh conditions of alpine habitats asexual reproduction is predominant and sexual reproduction is rare, and, consequently, levels of genetic variation might be lower in alpine plant populations. However, if higher levels of asexual reproduction (i.e. recruitment of identical multilocus genotypes in the population) are responsible for the lower levels of genetic variation in alpine populations, we would expect the genotypic diversity (i.e. number of different multilocus genotypes) to be lower in alpine compared with lowland populations. As microsatellites, which previously have been shown to be adequate markers for inferring genotypic diversity within natural populations (Van der Velde et al., 2001a, b), did not reveal lower genotypic diversity in alpine populations of P. formosum (data not shown), increased levels of asexual reproduction most probably cannot explain the lower levels of genetic variation in these populations. Alternatively, the degree of isolation of the alpine populations itself may explain the lower levels of genetic variation within these populations. As argued above, alpine populations experience, owing to the mountain ranges, lower levels of long-distance gene flow than the sampled lowland populations of north-western Europe. This suggests that aerial spore dispersal, which has been proposed as a common mode of long-distance dispersal for moss spores (Van Zanten & Pócs, 1981), is less effective in alpine areas. Alpine populations are therefore likely to have received less new genotypes/alleles via long-distance spore dispersal (from geographically distant and genetically different populations) than the sampled lowland populations of north-western Europe, which might have caused the difference in gene diversity levels observed between both types of populations.
CONCLUSIONS
For most Polytrichum species genetic differentiation among populations is low. This indicates that levels of spore dispersal have been sufficient to counteract genetic drift and to erase, if any, genetic traces of historic events in these species. This result agrees well with Van Zanten & Pócs (1981) who showed that spores can disperse over large distances. Even though geographical distance is not easily restrictive for spore dispersal in Polytrichum species, mountain ranges seem to limit spore dispersal significantly in some species. Additionally, although not addressed in this paper, preliminary data obtained for P. commune and P. juniperinum indicate that populations from Japan and North America are differentiated to a considerable extent from European populations. This suggests that also very large geographical distances and oceans do limit spore dispersal between populations substantially. In contrast to most examined Polytrichum species, only for P. juniperinum have high levels of genetic differentiation and clustering of geographically nearby populations been observed, which are most likely explained by historical events. It is as yet unclear why the impact of historical events on the genetic structure of P. juniperinum has been different from the impact of these historical events on the genetic structure of the other Polytrichum species.
ACKNOWLEDGEMENTS
We thank Heinjo During, Wilke van Delden and one anonymous reviewer for helpful comments on the manuscript, and Jaakko Hyvönen, Paco Lloret, Janet Marr/Janice Glime and Takeshi Ueno for providing us with samples from several populations analysed in this study.
