During the last couple of decades, the small, arctic-alpine Saxifraga rivularis L. (Fig. 1, 3) has been studied using molecular markers in order to learn more about its origin, genetic variation and distinctiveness. The results have indicated where it may have had its refugia during the last glaciation, and how it has reached its current distribution. Many researchers have been involved, and several international papers have been published. Here, I summarize some of what we now know about this rather anonymous, but quite remarkable little herb.

Fig1 Saxrivularis ssp rivularis Narvik KBWestergaard
Fig.1. Saxifraga rivularis ssp. rivularis growing in a typical late snow bed in Vesterskaret, Narvik, northern Norway.
Photo K.Bakke Westergaar

Linnaeus first described Saxifraga rivularis in 1737 from Lapland in Sweden. The plants are small, with loosely to densely tufted growth. The flowers are small, with white to pink petals, and the leaves are palmate. The species is strongly autogamous, and reproduces by both seeds and short rhizomes. Seeds are dispersed away from the parent plant by short-range ballistic dispersal, but they lack any obvious morphological adaptations to long-distance dispersal except being small and light. 

It is part of the circumpolar Saxifraga rivularis-complex, which consists of four species traditionally being designated as diploids (2n = 26; S. hyperborea, S. bracteata, and S. flexuosa) or tetraploid (2n = 52; S. rivularis). The taxonomy and evolutionary history of this complex were investigated by Jørgensen et al. (2006) by flow cytometry, morphological characters and high-resolution molecular markers called amplified fragment length polymorphisms (AFLPs). They suggested that the tetraploid S. rivularis originated in Beringia sometime before the last glaciation as the result of a single allopolyploidization event (hybridization and chromosome doubling) between the diploids S. bracteata and S. hyperborea (see also Brochmann et al., 1998 and Guldahl et al., 2005).

Saxifraga rivularis has two described subspecies with somewhat different morphology and ecology: the Atlantic ssp. rivularis, and the Beringian ssp. arctolitoralis (Jørgensen et al., 2006). Subspecies rivularis has a relatively long, glabrous or sparsely hairy flowering stem, while ssp. arctolitoralis has a short, sparsely to densely hairy flowering stem. Further, the hypanthium of ssp. rivularis is sparsely covered by short glandular hairs with non-colored or weakly colored partition walls, while long glandular hairs with purple partition walls densely cover the hypanthium of ssp. arctolitoralis. And whereas ssp. rivularis prefers moist and nutrient-rich habitats such as snow-beds, springs and bird-manured cliffs, ssp. arctolitoralis seems to be limited to clay and silt on seashores

Saxifraga rivularis ssp rivularis distribution map
Fig.2: Distribution map of Saxifraga rivularis

The oceanic S. rivularis is distributed on both sides of the North Atlantic Ocean, and also narrowly on both sides of the Bering Strait (Fig. 2). In the eastern Atlantic region, it is rather common in the Norwegian and Icelandic mountains, frequent in the arctic archipelagos of Svalbard and Franz Josef Land, on Novaya Zemlya and Jan Mayen, while it is rare along the arctic coast from Kola Peninsula to Taimyr, and also in the Scottish Highlands. In the western Atlantic region, it is found along the Greenland coast, and in the eastern parts of Nunavut, Quebec, and Labrador. Its Beringian range is restricted to parts of the arctic Pacific coast of Alaska, Chukotka and Wrangel Island. The highly disjunct Atlantic and Beringian range of S. rivularis is one of the most extreme disjunctions known in the arctic flora.

Genetic variation and conservation

To potentially provide information for conservation biologists, the genetic variation in the southernmost European stations of S. rivularis in the Scottish Highlands has been studied using a variety of molecular markers (Hollingsworth et al., 1998; Westergaard et al., 2008). The small and fragmented Scottish populations are found in naturally unstable habitats, and some populations have been observed to decline. Although Hollingsworth et al. (1998) only detected low levels of genetic variation, they suggested that a variety of the Scottish populations should be protected. This would guard against stochastic events such as rock falls or local environmental stress, and also help conserve any undetected genetic variation that may exist. 

Hollingsworth et al. only had two Icelandic plants for comparison with the Scottish material of S. rivularis. Thus, they were not able to assess whether or not the Scottish populations were genetically depauperated compared to populations in the main range of the species. We later raised this question in a meta-analysis of genetic diversity in Scottish alpine plants (Westergaard et al., 2008), where the levels of genetic diversity and distinctiveness in the Scottish population of S. rivularis at Lochnagar were compared with 21 populations from the total range of the species. Our results suggested that the Scottish population was not genetically depauperate compared to other populations in northern Europe (excluding Svalbard), and further, it was not genetically distinctive, and could be regarded as a southern outpost population. The ancestors of the Scottish populations probably shared refugia with other European populations of S. rivularis during the last glacial period, and have managed to persist in an increasingly sub-optimal habitat as the climate became warmer after the deglaciation. Such small, fragmented and geographically isolated populations may be vulnerable to processes such as genetic drift and inbreeding depression. Although we did not find any genetic distinctiveness in the Scottish populations using AFLPs, we cannot exclude the possibility that other unique genetic components are present. To prevent a reduction in overall genetic diversity, we concluded that it is important to maintain present distributions (Westergaard et al., 2008).

Long-distance dispersal of Saxifraga rivularis

We have investigated the glacial history, migration routes, and dispersal abilities of S. rivularis as part of a large project at the National Centre for Biosystematics (University of Oslo, Norway). This project was designed to investigate the past and future colonization of the geographically isolated arctic archipelago of Svalbard. Saxifraga rivularis was included as one of three study species that are present in Svalbard even though they are lacking obvious morphological adaptations for long-distance dispersal, except that the seeds are small and light.

Apparently, S. rivularis must have colonized Svalbard repeatedly from several source regions in order to bring the observed genetic variation to the archipelago (Alsos et al., 2007). Along with other cold-hardy species like Cassiope tetragona, Salix herbacea, and Dryas octopetala, S. rivularis had the highest estimates of colonizing propagules of all included study species, and they also had a level of genetic diversity in Svalbard similar to that in their primary source regions. We concluded that dispersal might not be the limiting factor for long-distance colonization in the Arctic, even for species like S. rivularis, which lack special adaptations for long-distance dispersal. It is more likely the establishment phase, including germination with subsequent survival and successful reproduction, which is limiting plant colonization in the Arctic (Alsos et al., 2007).

During an expedition along the west coast of Greenland in 2004, I came across two populations of S. rivularis that caught my eye and made me collect them. The first one was growing inside an old Norse stone construction “polar bear trap”(fig. 3) on the western part of the Nuusuuaq peninsula (72ºN, 52 ºW), just north of Qeqertarsuaq Island (Disko Island). The second one was growing in a wet area in connection to the dog yards in the small village Tasiusaq (73ºN, 56 ºW). Both populations had a more loosely tufted growth form than I was used to seeing, and the flowering stems generally had about the same height as the basal leaves. A closer inspection revealed that the hypanthium was densely covered with long glandular hairs. Could it be the supposedly Beringian endemic ssp. arctolitoralis?

When investigating widespread species like S. rivularis, sampling of populations takes time – a lot of time. Our work with S. rivularis has been work in progress for the last decade or so, and we have expanded the data set after each field season. We started out collecting mostly west-Atlantic populations, then we included Greenlandic and Canadian populations, and finally we included populations of ssp. arctolitoralis from Beringia. It was not until we had proper collections of ssp. arctolitoralis that we were able to figure out the origin of the interesting populations from Greenland.

We did a phylogeographic analysis of 45 populations of S. rivularis, covering most of the distribution area of the species (Westergaard et al., 2010). Unfortunately, in spite of considerable efforts, we were not able to obtain leaf material from Russia for our analyses. Based on their geographic origin, we referred to 35 of the populations as ssp. rivularis, and 10 as ssp. arctolitoralis. First we sequenced five regions of the chloroplast DNA from 12 individuals in search for variation, but we did not find any. Then we tested a high number of AFLP primer combinations (64), and selected three that yielded the most variation. The AFLP data clearly revealed a distinct division of the populations into two groups, one amphi-Atlantic and one mainly Beringian, corresponding with the two subspecies. However, four of the western Atlantic populations grouped with the Beringian ones, just as expected from the observed morphological characters. Thus, ssp. arctolitoralis could no longer be regarded as a Beringian endemic, and notably, the four Atlantic populations contained different subsets of the genetic variation found in Beringia, suggesting several recent dispersal events across the vast Canadian Arctic.

Our new data demonstrate that the extreme Beringian/Atlantic disjunction in S. rivularis has formed at least twice. The first expansion from Beringia probably happened before the last glaciation, and was followed by allopatric differentiation into the two subspecies in two separate glacial refugia. The second expansion probably occurred by several long-distance dispersals after the last glaciation, resulting in co-occurrence of the two subspecies in the Atlantic region. We argue that the last expansion happened by means of long-distance dispersal rather than vicariance, because the continental climate of Arctic Canada during the current inter-glacial makes it unsuitable for the oceanic S. rivularis, and because the species evidently has high dispersal abilities. 

Subspecies arctolitoralis is not rare in the western Atlantic region. After a revision by Marcel Blondeau (Herbier Louis-Marie, Université Laval, Quebec) and us of Canadian and Greenlandic herbarium specimens designated as S. rivularis and the closely related S. hyperborea in several Canadian and Norwegian herbaria, we found 75 specimens of ssp. arctolitoralis. Even though the two subspecies co-occur in e.g. Ittoqqortormiit, East Greenland, our AFLP data do not indicate that they hybridize.

Glacial history of Saxifraga rivularis

During the Late Weichselian period (25,000-10,000 years ago) the North Atlantic region was extensively glaciated, while Beringia remained only partially glaciated, and was an important refugium for arctic-alpine plants. We did not find much geographically structured genetic diversity within each subspecies of S. rivularis (Westergaard et al., 2010), and our results suggest that they originate from two main Weichselian refugia, and that the Svalbard populations may possibly originate from a separate one. For ssp. arctolitoralis, the main refugium was probably in Beringia, while for the Atlantic ssp. rivularis, the main refugium might have been located just outside the ice sheet that covered North America or Europe. Interestingly, the refugium for the Svalbard populations could have been in small, ice-free areas within the ice sheet in Svalbard. This is controversial, but the species is obviously very cold-hardy; it has been observed growing in bryophyte cushions deposited on a glacier lobe in Krossfjorden, Svalbard (I.G. Alsos, personal observation; (Colour Fig. 3). Unfortunately, since we lack reference material from Russia, we cannot assess whether the Svalbard populations originate from a separate refugium, and if so, where this was located. 


  • Alsos IG, Eidesen PB, Ehrich D, Skrede I, Westergaard K, Jacobsen GH, Landvik JY, Taberlet P & Brochmann C (2007) Frequent long-distance plant colonization in the changing Arctic. Science 316: 1606–1609.
  • Brochmann C, Xiang Q-Y, Brunsfeld SJ, Soltis DE & Soltis PS (1998) Molecular evidence for polyploidy origins in Saxifraga (Saxifragaceae): the narrow endemic S. svalbardensis and its widespread allies. American Journal of Botany 85: 135–143.
  • Guldahl AS, Gabrielsen TM, Scheen A-C, Borgen L, Steen SW, Spjelkavik S & Brochmann C (2005) The Saxifraga rivularis species complex in Svalbard: molecules, ploidy and morphology. Flora 200: 207–221.
  • Hollingsworth PM, Tebbitt M, Watson KJ & Gornall RJ (1998) Conservation genetics of an arctic species, Saxifraga rivularis L., in Britain. Botanical Journal of the Linnean Society 128: 1-14.
  • Jørgensen MH, Elven R, Tribsch A, Gabrielsen TM, Stedje B & Brochmann C (2006) Taxonomy and evolutionary relationships in the Saxifraga rivularis complex. Systematic Botany 31: 702-729.
  • Westergaard KB, Alsos IG, Ehrich D, Eidesen PB, Hollingsworth PM & Brochmann C (2008) Genetic diversity and distinctiveness in Scottish alpine plants. Plant Ecology & Diversity 1: 329-338.
  • Westergaard KB, Jørgensen MH, Gabrielsen TM, Alsos IG & Brochmann C (2010) The extreme Beringian/Atlantic disjunction in Saxifraga rivularis (Saxifragaceae) has formed at least twice. Journal of Biogeography 37: 1262-1276.