In this paper we demonstrate that, by investigating polyploid complexes in d.\p/enium, it is possible to locate the areas in Europe that are southern glacial rcfugia, and arc likely to have been so since the tjcginning of the Plcistoccnc during the consecutive cold and warm periods in Europe. Identification and consenation of thrsc specific areas that s e n e as safe havens for plants, and perhaps animals, is of paramount importance for the maintcnancc of European liiodiversity because hlan's activities arc resulting in an ever-increasing loss of natural habitats and putting diversity at risk. The genus A.\pl~nium in Europe comprises some 50 taxa: half of these are diploid while the other half are polyploids derived from thr diploids. All aspleniums in Europe are (small) rock ferns with high suhstratc specificity. Today, most of mainland Europe, Scandinavia and the British Isles has been colonized by polyploid dsplenium species, while the diploids that gavc rise to these polyploids are distributed around (and more or less confined to) the hfcditcrrancan Basin. In the tetraploids genetic variation is partitioned mostly between sites, whereas diploids show a high dcp-cc of genetic variation both within and between sites. The tctraploid taxa seem capable of single spore colonization via intragametophytic selling, but the diploid taxa appear to he predominantly outbreeding, For most diploids at least two gametophytes, produced by different spores, have to tic prcscnt to achieve fertilization and subsequent sporophyte formation for the successful colonization of a new site. This results in a slower rate of colonization. The formation of auto-and allopolyploid taxa from diploid communities appears to have been a recurrent and common fcaturc in Europe. Minority cytotypc exclusion is likely to prevent the establishment of tetraploids within the diploid communities, but spores from tetraploids can cstaldish populations outside the diploid communities. The differences between colonization abilities of tetraploid and ancestral diploid taxa, resulting from their differcnt hrccding systems, has prevented thc merging and mingling of their ranges and led to the establishment of contact/ hybrid zoncs. This has resulted in the restriction of diploid populations to ancient glacial rcfugia and the colonization of the rest of Europe by polyploids. Mapping the rurrrnt distribution of these diploid communities and comparing the genetic diversity within and between outbrccding diploid i-l.cp/zninm taxa allows us to dcfinc the area, age and historical biogeography of thcsc rrfugia and to assess thrir importance for present day genetic and species diversity in Europe. 0 lW9 'l'tir Liiiiiriiii Socirt! 111 1.11ii1i11ii

Trichomanes speciosum is a threatened species restricted to sheltered, very humid sites. Uniquely amongst European ferns, differing ecological tolerances of the gametophyte and sporophyte generations are manifested as widely differing distributions. The perennial, vegetatively propagating gametophyte persists in drier, colder, darker habitats than the sporophyte. In sites where the gametophyte grows, light availability was found to be 1 µmol m −# s −" for approx. 85% of daylight hours, rarely or (in some sites) never rising above 10 µmol m −# s −" . Much of the time, light was 0.01% of full sunlight. Measurements of gas exchange and chlorophyll fluorescence yield show that these plants have optimal photosynthesis at light intensities c. 5-10 µmol m −# s −" , the highest light to which they are normally exposed to in their natural environment. The absence of any capacity for reversible nonphotochemical fluorescence quenching means that there is little or no protection of the photosynthetic apparatus from light-induced damage. We conclude that these plants are able to create what are essentially monocultures in their extreme environments only because of a combination of low metabolic rate (at low temperatures) and an ability to make efficient use of what little light is available to them by morphological and physiological means.

Genetic diversity in the Killarney fern, Trichomanes speciosum Willd. has been investigated in south‐western Scotland, the northern‐most limit of the distribution of the sporophyte. T. speciosum is unique amongst European pteridophytes in that both phases of the life cycle are perennial and capable of vegetative propagation. Within sites no variation was revealed by allozyme electrophoresis, even where both generations were growing together. In contrast, diversity was observed among sites, with seven different multilocus phenotypes (MLPs) present in the area. Two of these MLPs covered large areas while the others were restricted to one, or few localities. Asexual reproduction of the gametophyte via gemmae is assumed to be the main means of dispersal in recent times, allowing single clones to become widespread, while the overall genetic variability may be attributed to sexual reproduction and spore dispersal in historic times under more favourable climatic conditions. We suggest that it is not inbreeding, nor lack of genetic variation that limits sporophyte production, but rather the prevailing climatic conditions. The sporophyte is extremely rare and vulnerable. However, when the gametophyte is considered, the species is neither threatened with extinction, nor does it appear to face the danger of marked genetic erosion, because the long‐lived gametophyte stage contains all of the genetic variability present in the area and can be regarded as a valuable ‘seed‐bank’.

The potential for environmental heterogeneity to generate spatial structuring of genotypes in seedplant populations that occupy patchy habitats has been demonstrated by several studies, but little is known about the population structure of pteridophytes occupying patchy environments. In this study we have examined the genetic structure of isolated populations of the rock fern Asplenium csikii, an ecological specialist, growing almost exclusively on perpendicular walls of natural rock outcrops. All genetic variation observed in this taxon was partitioned between localities; no allozyme variation was found within a site and each site was colonized by a single multilocus phenotype (MLP). In total, ®ve di erent MLPs were recorded from the nine localities, with two MLPs present at more than one site. Previous examination of population structure and genetic diversity in another rock fern, A. rutamuraria, showed that the genetic diversity increases through multiple colonization over time. However, we cannot ®nd any such correlation for A. csikii. All populations are genetically uniform, despite the probably considerable age of the populations and sites. Earlier studies concluded that the ample production of wind-borne propagules would lead to multiple colonization of sites and that reproductive features, such as single-spore colonization and subsequent intragametophytic sel®ng, would lead to very little genetic structuring of fern populations. In contrast to this prediction, it appears that ecological specialization and the scarcity of the narrowly de®ned niche contribute strongly to the pronounced partitioning of genetic variability observed in populations of A. csikii.

The Killarney fern Trichomanes speciosum Willd. (Hymenophyllaceae) is unique in possessing both extensive sexual (sporophyte and gametophyte generations present) and asexual (gametophyte only) ranges. It was first discovered in central Europe in 1993 and is represented in this area only by its perennial, vegetatively propagating gametophyte generation. Genetic variation has been investigated at 35 sites. Allozyme diversity is partitioned primarily between, not within, sites. Although genetic variation exists at a fine scale « 5 m) within some populations, the results suggest that clones were not intimately associated in these cases. The majority of sites support unique multilocus phenotypes. Where phenotypes were present at more than one site they tended to recur at the next closest site. However, similar phenotypes link eastern and western Pfiilzerwald sites up to c. 70 km apart. This pattern of diversity suggests that colonisation was not solely of a "stepping stone" or "leading edge" type. We suggest that during a climatically favourable period, probably the Atlantic hypsithermal, there may have been an explosive colonisation by long-distance dispersal from refugial areas. This was followed by a short period during which sporophyte production, sexual reproduction and local spread were possible. With climatic change, reduction in the available habitat and the loss of the sporophyte generation, different individual genets became fixed within small, favourable, but scattered, sites. The possibility that some central European sites north of the Alps acted as periglacial refugia cannot be discounted, but would appear less likely than (re-)colonisation from the Atlantic fringe.