4 Discussion
4.1The genetically distinct clusters from Western to Eastern Siberia correspond well to the main larch species
The GBS resulted in a total of 14,565 variable SNPs among the 148 individuals, which is a high number in comparison to other studies with Pinaceae (e.g. Johnson et al. 2017; Eckert et al. 2010; Chen et al. 2013). With GBS the complexity of the large genome of Larix was efficiently reduced for downstream analyses (Dong et al. 2019; Deschamps et al. 2012).
The cluster analysis based on all SNPs identified four distinct genetic clusters. The boundaries of the genetically inferred main clusters correspond well to the boundaries of the main Siberian larch species described by Abaimov (2010). As expected, the Western Siberian cluster (cluster 1) can be assigned to L. sibirica and corresponds to its distribution across the territory between the Ob and Yenisey Rivers (Abaimov 2010). The easternmost site belonging to this cluster is located at Lake Khamra (59.97° N, 112.96° E). The Western Yakutian cluster (cluster 2) shows the distribution area of L. gmelinii , which extends from the basin of the Pyasina River in the west to the Lena River delta in the east. The five tree individuals from the Taymyr Peninsula region are the westernmost trees in this L. gmeliniicluster in our study. According to Abaimov (2010), the northernmostL. gmelinii forests are located on the Taymyr Peninsula. The Eastern Yakutian cluster (cluster 3) is attributable to L. cajanderi , whose western border habitat goes along the Lena River. However, in the discussion about the taxonomic status of L. cajanderi , there are different views on where to place the boundary between L. gmelinii and L. cajanderi . Some authors classify L. cajanderi Mayr as a synonym of L. gmelinii(Rupr.) Kuzen (Borsch et al. 2020). Dylis (1961), who divided L. dahurica into a western and an eastern race, drew a dividing line between them at 120–123° E longitude, which Bobrov (1972) later described as two species L. gmelinii and L. cajanderi . This also coincides with the transition zone between our separate Western Yakutian and Eastern Yakutian clusters and supports the differentiation of 4 rather than 3 clusters in our study. Further evidence for a separation has been found by molecular studies using both mitochondrial and chloroplast markers, which also reveal weak but visible genetic differentiation between L. gmelinii and L. cajanderi (Polezhaeva et al. 2010). In summary, we argue that the genetic differences between L. gmelinii and L. cajanderiare very small but existent.
Particularly striking are the topographical conditions such as mountain ranges or river valleys along the cluster and species boundaries, especially as the positions of the samples were not considered in the cluster inference. The border between the West Siberian cluster (cluster 1) and the Western Yakutian cluster (cluster 2) corresponds almost exactly with the course of the Yenisey River. Nevertheless, it can be assumed that at this point the Central Siberian Plateau serves more as a geographical barrier which is responsible for the genetic differentiation between L. sibirica and L. gmelinii . Similarly, the Verkhoyansk Range likely serves as a geographical barrier between the Western Yakutian cluster (cluster 2) and the Eastern Yakutian cluster (cluster 3) rather than the Lena River and is thus responsible for the separation between L. gmelinii and L. cajanderi . For a lowland species like larch, the mountains present a persistent geographic barrier to expansion. In addition to long-term geographical isolation, mountains can increase climatic gradients and thus further facilitate population differentiation through environmental isolation (Li et al. 2019). There may be gene flow between adjacent watersheds, mountain ranges, or valleys, but less than between populations that are not separated by landscape components (Priyadarshan and Schnell 2012). The separation could be responsible for keeping two populations genetically isolated so that they slowly diverge genetically from each other (Rowe et al. 2017). A study on the effects of the last glacial period on genetic diversity and genetic differentiation inCryptomeria japonica in East Asia (Tsumura et al. 2020) concludes that geographical barriers between populations are one of the strongest drivers for genetic variety.
The individuals in the core area of the clusters have an affiliation to exclusively one cluster, but at the periphery, individuals have a fragmented affiliation to different clusters. In the contact zone of their geographic ranges, we found hybrids from the L. gmeliniiand L . cajanderi clusters forming a stripe of transitional populations on the left bank of the Lena River (Abaimov 2010; Isaev et al. 2010). These findings meet expectations, since cross-pollination of two different plant varieties is taking place in these transitional areas (Baltunis et al.1998). Unfortunately, no tree genetic samples from the mountainous region of Krasnoyarsk Krai between cluster 1 and cluster 2 were available in this study. However, it can be assumed that the species in the contact zone between the eastern limit of L. sibirica and the western limit of L. gmelinii are involved in natural hybridization (Abaimov 2010; Isaev et al. 2010). Szafer (1913) described this stripe of hybrid populations as L. czekanowskii . From the observation of genetic clusters and hybrid zones, the modern and historical genetic influences can be deduced. Populations that have evolved independently due to isolation by geographical barriers or because they retreated into refugia during glacial periods met again when physical barriers disappeared or when the populations expanded again during warm periods (Abaimov 2010; Isaev et al. 2010).
Noticeable is the significant geographical distinction of the Chukotka cluster (cluster 4), which can be considered as a variant of L. cajanderi . The intraspecific genetic deviations between cluster 3 (Western Yakutia) and cluster 4 (Chukotka) could stem from separation by the Chersky Range and the Kolyma Mountains, where varied microclimates could be present at different elevations, which could have led to differentiation due to regional climatic differences. Genetic differentiation can also increase with geographical distance even when there are no apparent barriers to gene flow. Since very little gene flow occurs between the L. cajanderi population in Chukotka and the main distribution area of the species, then during the glacial and inter-glacial cycles, these Chukotka populations could have diverged genetically due to gene drift, a process called isolation by distance (Rousset 1999). The genetic variability of L. cajanderi is also supported by observations showing that L. cajanderi is involved in hybridization with L. kamschatica and other hybrid forms ofLarix species in the far northeast (Abaimov 2010).
4.2 Refugia likely assisted the invasion of northern areas of the forest-tundra ecotone