The formation of the arctic biota and the impact of Quaternary climate change on the biogeographic structure of species

The Earth began to experience intense changes in climate during the late Tertiary Period with a drastic decrease in global temperatures from about 15 million years ago (Ma). This led to the conversion of the high-latitude forests to arctic tundra, which achieved a circumpolar distribution by 2 Ma.¹ The plant species of the newly formed tundra were derived from shrubby and herbaceous elements of the arctic forests of the Tertiary, as well as from high-altitude species migrating north along mountain ranges in North America and Asia as temperatures declined.²

During the Quaternary from about 0.7 Ma, climate shifts occurred on a shorter timescale with 100,000-year glacial periods and intervening interglacials lasting up to 20,000 years. During the glacial periods, large parts of Eurasia and North America were covered in ice sheets spreading from the poles and retreating during the interglacials.³

But the ice sheets did not progress evenly and some areas remained free of ice, in particular Beringia, the land on either side of the Bering Strait, spanning from northeast Russia through Alaska to the Yukon Territory and including the land bridge between them (which was exposed due to sea level drops during the major glaciations).²

After comparing the distribution maps of several species, the biogeographer Eric Hultén postulated that arctic species experienced range fragmentations due to the advancing ice sheets, but that unglaciated areas such as Beringia served as refugia, where they would persist and from where they would spread out again once the ice had retreated (Hultén 1937, cited in Abbott & Brochmann 2003²). Other proposed refugial areas were in periglacial areas to the south of the ice sheets; the exposed continental shelves of northern Greenland, Scandinavia, and northern Siberia; and nunataks, mountain peaks protruding above the ice.² (However the genetic signal of survival on nunataks would probably be lost later as these populations get overwhelmed by recolonizers from the south.¹)

Analysis of fossil pollen has confirmed that diverse tundra communities were present in Beringia during the Last Glacial Maximum; tundra also persisted to the southeast and -west of the North-American ice sheet, as well as to the south and east of the ice sheets over northern Europe and northwest Russia.² It should be noted, however, that palynological data are biased in favour of wind-pollinated species, as well as those occurring peatlands and wetlands; they also have little taxonomic resolution, especially with regard to geographic origin.³

Molecular data can provide a rich source of information for inferring the histories of species. The fragmentation of species ranges by the ice sheets would have created smaller populations, isolated from each other both by distance and by dispersal barriers, particularly ice. These conditions lead to genetic divergence through drift and selection,⁴ although the effect of selection might be slight if the populations still occupied similar habitats.¹

In warmer periods, the ranges of the populations would expand again and they may come into secondary contact. In this case, there would be greater genetic diversity than in the ancestral population but the differentiation between the populations would eventually be erased through recombination if they share the same gene pool.

On the other hand, it has been suggested that breeding between the two populations may result in offspring with lower fitness, which could lead to reinforcement and speciation. There is also the potential for speciation to occur through allopolyploidy, whereby the offspring retains a greater than haploid number of chromosomes from both parents, causing reproductive isolation from either of the parental populations.¹

Molecular data can thus be used to test biogeographic theories, as populations in refugia are expected to have higher genetic diversity, having had large and stable population sizes for a long time. Recolonized areas are expected to have low diversity due to successive founder events, as only a subsample of the population spreads into the new area. An exception is in secondary contact zones, where diversity will be higher due to the presence of somewhat-diverged lineages.²

Initially, most phylogeographic studies of arctic taxa were done on animals, rather than on plants because mitochondrial DNA (mtDNA) has a high substitution rate in animals, but is conservative in terms of the arrangement of genes. In plant mtDNA, there is a high degree of gene rearrangement and duplication, while the substitution rate is about 100 times lower. Plant chloroplast DNA (cpDNA) also has less variability than animal mtDNA.⁵

These studies on arctic animals showed that Beringian populations were more genetically diverse than those in deglaciated areas for snow geese, whitefish, and Daphnia (Quinn 1992; Bernatchez & Dodson 1994; Weider & Hobæk 1997, 2002 — all cited in Abbott & Brochmann 2003²). Arctic-alpine ground beetles were also found to have colonized Canada from Beringia.⁶ However, burbot survived glaciations in refugia to the south of the ice sheets,⁷ while a refugium for collared lemmings is suspected in the Canadian High Arctic or northern Greenland.⁸ Evidence was not found for long, stable population histories of lemmings (Fedorov et al. 1999, cited in Abbott & Brochmann 2003²) or of arctic hares,⁹ though the latter seem to have survived on the coast of the Canadian arctic adjacent to the Laurentide ice sheet.

Arctic plant biogeography was first studied using cpDNA sequences or random amplified polymorphic DNA (RAPD), as cpDNA tends to be the most variable of the plant genomes. RAPDs were superseded by AFLPs, which have higher reproducibility and have the advantage of sampling from the entire genome of an individual (as opposed to just the chloroplast).⁵ It can generate more informative data than sequencing, but since the exact nucleotide sequence remains unknown, divergence times cannot be estimated from this data.

A study of the genus Saxifraga based on RAPD from the chloroplast genome showed that deglaciated areas were rapidly recolonized by these species, with a gradient of increasing molecular diversity from south to north, as theoretically predicted.¹⁰ Subsequent studies by Abbott and colleagues of the species Saxifraga oppositifolia (purple saxifrage) used cpDNA to estimate a time of divergence of 5.4–3.8 Ma between the two major arctic clades of the species, subsequent to their dispersal from central Asia. The Eurasian and North-American clades are thought to have survived the glaciations (from about 0.7 Ma) in separate refugia, although there has since been secondary contact where their ranges have overlapped over several cycles of interglacial expansion. Proposed refugia include Alaska, south of the North-American ice sheet, the Canadian Arctic archipelago or northern Greenland, and somewhere in western Siberia, possibly the Taymyr peninsula, although sampling from this area was too sparse to greater precision.^2,1

Vaccinium uliginosum cpDNA recovered three clades, with one centered in Beringia, another on both sides of the Atlantic, and the third having a circumpolar distribution. However, as with S. oppositifolia, this main divergence was estimated to have occurred before the onset of the episodes of glaciation, and there was not enough geographical structure to infer patterns from the last 700,000 years (Alsos et al. 2005, cited in Schönswetter et al. 2007¹¹).

The distribution of Dryas integrifolia from Alaska to Hudson Bay was examined by means of cpDNA RFLPs. Four refugia were proposed in North America, although the structure in this data was also weak: Beringia and the Canadian High Arctic again seemed to be likely refugia, and the eastern interior or east coast of Canada were also possibilities. The lack of strong genetic support for these refugia was enhanced with fossil evidence for the Beringian refugium, as well as for one between New Jersey and Quebec.¹²

Earlier biogeographers, including Eric Hultén, considered the Atlantic Ocean to be an insurmountable barrier to dispersal for arctic plants, as they were thought to be maladapted to long-distance dispersal. However, several studies found identical haplotypes to occur on either side of the ocean, including S. oppositifolia,² Juncus biglumis,¹³ and Carex bigelowii s.l.¹¹ In the latter species, dispersal over long-distances reduced the geographic structuring of molecular variation over its entire range and transoceanic dispersal was evident in the occurrence of groups spanning the Bering Sea and the Atlantic Ocean.¹¹

It has been proposed that, in secondary contact zones, low hybrid fitness could lead to reproductive isolation if populations are highly differentiated (e.g. Abbott & Comes 2003¹). However so little time has passed since divergence in many of the organisms discussed that they are morphologically indistinguishable. If populations are also not ecologically differentiated, it seems unlikely that they would have significant differences in fitness. The effect of secondary contact and hybridization might, therefore, be a reduction in inbreeding depression and enhanced fitness.

Speciation might, however, occur due to allopolyploidy, which has been found to be extensive in the arctic flora.³ Abbott & Brochmann 2003² characterize polyploidization as a “simple and rapid” process that enhances the colonization ability of a group of plants, as “genetic diversity is maintained through periods of extreme inbreeding and bottlenecks.” The same allopolyploid species may be formed repeatedly through independent hybridization events, as in the case of Saxifraga cernua. The same parent species may also give rise to different hybrid species, e.g. S. opdalensis and S. svalbardensis from the crossing of S. rivularis and S. cernua individuals.²

Long-distance dispersal and frequent and recurring hybridization events obscure population histories. The former reduces molecular variation between populations through high levels of gene flow. The latter exacerbates the problems in distinguishing the effects of successive fragmentation and recolonization events in tandem with climatic cycles, by increasing genetic diversity in recently colonized areas where divergent populations come into contact again, particularly since the process is so common as to happen repeatedly over several periods of expansion.

The increasing use of AFLPs improves on that of cpDNA because, in addition to generating more variability in the data, it also allows both the organellar and nuclear genomes to be sampled. This reduces the bias in the reconstructed population history that may be present due to the fact that cpDNA is inherited maternally and it allows the population structure of the entire genome to be determined, as opposed to a single unit of inheritance.

The drawback is that it does not allow for the time to most recent common ancestor to be calculated. For this it is necessary to use direct sequence data together with a model of sequence evolution. However, the divergence times estimated by, e.g., Abbott & Comes 2003¹ were based on cpDNA sequences and thus only apply to the chloroplast gene tree. In order to relate genetic differentiation to geological and climatic events, it is necessary to have a confident estimate of the point where populations became isolated, which would require the sampling of multiple genes, including those transmitted through pollen. For biparentally-inherited nuclear genes to be used, which show lower substitution rates than chloroplasts in plants,⁵ gene regions with particularly high variability would have to be identified. Even then, the fourfold effective population size would make gene tree congruence unlikely.⁵ In this case, coalescent models should be used to estimate divergence times based on multiple genes. Only then can inferences be confidently made relating to the timing of range contractions and expansions in concert with the progression and retreat of the ice sheets.

Numerous studies on the phylogeography of the arctic biota have shown evidence for relatively large, stable populations persisting in refugia through Quaternary glaciations, followed by rapid range expansions into formerly glaciated areas after the northward retreat of the ice sheets. The existence of a major refugium for many species in Beringia, originally proposed by Eric Hultén, is supported by several analyses. Some species might have persisted in other refugia further south in Eurasia and North America, and possibly on exposed continental shelves in the Canadian High Arctic, Northern Greenland, and Scandinavia.

Nevertheless, the important finding that has emerged from a number of meta-analyses^3,2,14 is that different species can have dramatically different histories in relation to changes in their habitat. The composition of species assemblages is, therefore, liable to change, not only between plants and their associated animals, but also between plants sharing a habitat.³ This will depend on the range of ecological tolerance of the species involved and their dispersal ability, in addition to the level of genetic variation inherent in each population.