Discussion
We found that the ecotone, stretching from subalpine mountain birch forest to treeless low-alpine vegetation, was the primary structuring gradient shaping the soil biota. Both the fungal communities, as well as the overall micro-eukaryotic communities, were structured primarily along the ecotone. Variation in community composition between sites was largely accounted for by regional variation in climate, as well as other site-specific factors, such as slope, aspect and bedrock, the latter affecting nutrient content.
We observed that the belowground community composition largely reflected the aboveground shift in vegetation. However, for fungal communities, the 18S and ITS2 markers showed substantial differences in abundance and distribution of fungal groups across the ecotone. Most noteworthy, Archaeorhizomycetes (Ascomycota) and Mucoromycota dominated in the 18S dataset, and Chytridiomycota and Cryptomycota were also abundant across the entire ecotone. In contrast, these groups were poorly represented in the ITS2 dataset. On the contrary, the subphylum Pezizomycotina, represented mainly by Leotiomycetes, Eurotiomycetes, Lecanoromycetes and Dothideomycetes, was largely absent in the 18S dataset, but made up a large proportion of the ITS2 dataset. We suggest that these contrasting results can largely be explained by primer biases (Nilsson, Anslan, et al., 2019; Rosling et al., 2011; Tedersoo et al., 2015). The employed ITS primers have been shown to discriminate against e.g. Archaeorhizomycetes (Ihrmark et al., 2012; Rosling et al., 2011; Tedersoo et al., 2015) and they also seemingly amplify Mucoromycota, Chytridiomycota and Cryptomycota poorly. The 18S primers we used in this study, was expected to amplify all eukaryotes (Hadziavdic et al., 2014) and provide a more general overview of the fungal community composition compared to the ITS2 dataset. However, the low amount of Pezizomycotina in the 18S data indicated that primer biases may be present. Indeed, after a closer inspection, a mismatch in the reverse primer was observed in numerous Pezizomycotina lineages. Irrespective of the biases, there was a general trend of more Ascomycota above the forest line, most notably Archaeorhizomycota (18S data), and Leotimycetes and Eurotiomycetes (ITS2 data).
Several groups of Ascomycota in the ITS2 data represented plant root-associated fungi. For example, the overall most common OTU showed high sequence similarity to Pezoloma ericae , a common ErM fungus (Smith & Read, 2008) . In the species ordination plot, most of the root-associated ascomycetes, representing ErM and DSE fungi, clustered in the alpine heath along the ecotone, where ericaceous plants, especially Empetrum nigrum , dominated. The gradual increase of Leotiomycetes and Eurotiomycetes towards the low alpine vegetation (ITS2 data) is in line with the pattern showed in Tedersoo et al.(Tedersoo et al., 2014), where these groups had highest relative abundances in arctic tundra. ErM and DSE fungi living under stressful conditions typically have melanized hyphae. Melanin is a recalcitrant polymer of the hyphal cell wall that may protect fungi against desiccation (Fernandez & Koide, 2013). Melanin induces slower decomposition and the retention time of highly melanized fungi in the soil may thus be longer, in comparison with non-melanized fungi (Fernandez, Heckman, Kolka, & Kennedy, 2019; Fernandez & Koide, 2014). Fungal necromass (i.e. dead biomass) is an important part of boreal soil carbon pools (Clemmensen et al., 2015), and the longer retention time of melanised mycelia may explain the strong positive correlation between root-associated ascomycetes and percent C content in this study. We also observed an increase of Archaeorhizomycetes towards the alpine heath. Little is known about the ecology of the Archaeorhizomycetes (Rosling et al., 2011), but is has been hypothesised that some Archaeorhizomycetes may be root-associated mutualists (Menkis, Urbina, James, & Rosling, 2014). The high abundance of Archaeorhizomycetes in the low alpine vegetation may indicate that they are linked to ericaceous plants, along with ErM fungi and DSEs. Other studies confirm that Archaeorhizomycetes are relatively abundant in stressful environments (Sterkenburg, Bahr, Durling, Clemmensen, & Lindahl, 2015), e.g. in high altitude and latitude ecosystems (Pinto-Figueroa et al., 2019; Schadt, Martin, Lipson, & Schmidt, 2003).
In parallel to the root-associated ascomycetes, we observed that percent soil C and ergosterol content had a strong positive correlation towards the low-alpine vegetation, in line with our hypotheses 3. Clemmensenet al. (Clemmensen et al., 2015) showed a similar correlation between ergosterol and total soil C stock, and argued that this relationship is due to slower mycelial turnover rate leading to a long-term build up and increased soil C sequestration (Clemmensen et al., 2013; Clemmensen et al., 2015; Hagenbo et al., 2017; Kyaschenko, Clemmensen, Karltun, & Lindahl, 2017). These processes may also account for the high C and ergosterol content in the low-alpine vegetation. Percent C content correlated strongly with the community composition in the ordination diagrams, but with respect to decomposition and cycling processes, C content is not regarded as a driver, rather a consequence of the fungal community composition (Clemmensen et al., 2015). However, it must be emphasised that in this study we only have data on percent C content and not overall C stock measurements. Hence, to conclude about overall C stock dynamics, additional data must be collected.
In accordance with hypothesis 2, the distribution of EcM fungi across the ecotone reflected their host plants distributions, with higher abundances of EcM fungi below the mountain birch forest line. Our observations are in line with what observed by Vašutová et al . (Vasutova, Edwards-Jonasova, Baldrian, Cermak, & Cudlin, 2017), where EcM fungi declined with altitude. Many EcM fungi are associated with relatively fast soil organic matter turnover (Bodeker et al., 2014; Clemmensen et al., 2013; Lindahl & Tunlid, 2015) and may, together with the saprotrophic fungi, account for the lower amount of soil C observed below the forest line in our study.
In general, we observed a higher proportion of saprotrophic fungi in the alpine birch forests compared to the heath, which may account for the lower fraction of soil C observed here, due to faster decomposition. Based on the 18S dataset, Mucoromycetes dominated the fungal community below the forest line. These fungi are foremost known as saprotrophs, and there is likely a higher abundance of available litter in the forest compared to the alpine heath. However, it has also been observed that Mucoromycotina, as fine root endophytes, are important symbionts with cryptogams (Field et al., 2015; Hoysted et al., 2019), receiving plant C in exchange of soil N. Hence, Mucoromycetes may play different functional roles, which may explain the high relative abundance of Mucoromycota across the entire ecotone.
Although our study was mainly focused on fungi, we found that,Metazoa was about equally abundant as fungi in the soil measured as sequence reads. As pinpointed earlier, some fungi Ascomycota might have been poorly amplified by the 18S marker, so it remains unclear whether the proportions reflect an unbiased picture of the belowground biota. The two most abundant Metazoa groups, Annelida and Nematoda, were common across the entire ecotone, whereas Arthropoda and Rotifera were relatively more abundant above the forest line. Other studies have shown that annelids and nematodes typically are more abundant at high latitudes and altitudes (Phillips et al., 2019; Procter, 1990). This is in line with the result of this study, where annelids and nematodes are the dominant groups. Rotifera has also been shown to be abundant in soils at high latitudes and altitudes, and are often present in mosses and lichens (Bielańska-Grajner, Ejsmont-Karabin, & Yakovenko, 2011; Fontaneto & Ricci, 2006), which corresponds well with their relatively higher abundance above the forest line. Studies have shown that soil moisture strongly influence protist communities (Bates et al., 2013; de Araujo et al., 2018; Stefana, Cornelia, Jorg, & Michael, 2014), while others regard soil pH as more important (Heger, Derungs, Theurillat, & Mitchell, 2016; Shen et al., 2014). Heger et al. (Heger et al., 2016) also showed a correlation between protist communities and altitude. In general, Cercozoa is commonly observed in various soil types, whereas Ciliophora and Apicomplexa have been shown to be relatively more abundant in humid soils (Bates et al., 2013). Due to the parasitic lifestyle of Apicomplexa, their distribution is likely largely dependent on the presence of host species (Arthropodes) (Mahe et al., 2017; Seppey et al., 2020). However, at kingdom and phylum level, none of the protists showed any clear distribution pattern across the ecotone.
The ecotone, stretching from subalpine mountain birch forest to treeless low-alpine vegetation, represents a corresponding shift in belowground fungal communities, from soils dominated by EcM and saprotrophic fungi to soils dominated by root-associated ascomycetes, respectively. This study has shown there is a parallel shift in belowground ergosterol content across the ecotone, which is strongly associated with the abundance of root-associated ascomycetes in the low-alpine vegetation. Our findings corroborate the view that despite higher aboveground productivity, shrubification and raise in forest line in northern ecosystems may lead to soil C loss because of higher soil respiration and C cycling (Parker et al., 2015; Sjogersten & Wookey, 2009; Sorensen et al., 2018), through corresponding shifts in fungal communities. However, C stock data is needed to conclude on this matter. On a technical note, our study also underlines that in future community studies of fungi, a wider set of DNA-markers and primers should be considered to obtain a comprehensive picture of soil fungal communities.