Results
Distribution of taxonomic and
functional groups across the ecotone
In the 18S dataset, Fungi was the overall richest group (1083 OTUs;
85,101 reads), followed by Metazoa, which included the highest number of
reads (1015 OTUs; 94,166 reads). These two groups dominated the soil
with about 80% of all reads at kingdom level. We observed a slight
increase in the proportion of Metazoa from the forest to the alpine
heath vegetation (Fig 2a). At phylum level, Nematoda (422 OTUs; 29,096
reads) and Filosa-Sarcomonadea (411 OTUs; 13,278 reads) included highest
richness, while Annelida (52 OTUs; 38,922 reads) possessed the highest
number of reads. We observed no clear trends for the distribution of
Annelida and Nematoda at phylum level, but the amount of reads belonging
to Arthropoda and Rotifera increased slightly towards the low-alpine
vegetation (Fig. 2b). At both kingdom and phylum level, none of the
protists, represented by Ciliophora, Cercozoa, Apicomplexa, Lobosa and
Conosa, showed any clear distribution patterns across the ecotone (Fig.
2a).
Only assessing fungi in the 18S data, the majority of OTUs belonged to
the phylum Cryptomycota (318 OTUs; 5804 reads), followed by
Basidiomycota (226 OTUs; 16,810 reads). However, the majority of reads
belonged to Mucoromycota (47 OTUs; 30,137 reads) and Ascomycota (181
OTUs; 26,096 reads), together making up about 60% of the fungal reads
(Fig. 2b and c). We observed that the relative abundance of Ascomycota
increased from the mountain birch forest to the low-alpine vegetation,
while we saw an opposite trend for Mucoromycota (Fig. 2b). The relative
abundance of Archaeorhizomycetes, which belongs to Ascomycota, increased
distinctly towards the low-alpine vegetation, with an opposite trend for
Mucoromycetes (Fig. 2c). We observed no clear trend for the distribution
of Agaricomycetes (Basidiomycota).
The ITS2 dataset (Fig. 2d) was dominated by the phyla Ascomycota (1522
OTUs, 380 072 reads) and Basidiomycota (903 OTUs, 200 356 reads) and
showed a different pattern in relative abundance, compared to the 18S
data. In the ITS2 data, the proportion of Basidiomycetes (especially
Agaricomycetes) decreased along the ecotone towards the low alpine
vegetation, while Ascomycetes (especially Leotiomycetes and
Eurotiomycetes) showed an opposite trend (Fig. 2d). Leotiomycetes made
up the highest proportion of Ascomycetes at class level. In contrast to
the 18S data, Archaeorhizomycetes and Mortierellomycetes/Mucoromycetes
only made up a small proportion of the ITS2 reads.
For the fungal ITS2 dataset, we assigned OTUs to functional guilds (Fig.
3a), which revealed that saprotrophs made up the largest group (815
OTUs, 130 986 reads), followed by unassigned ascomycetes (626 OTUs, 148
884 reads), EcM fungi (321 OTUs, 137 428 reads), yeasts (149 OTUs, 43
932 reads), pathotrophs (135 OTUs, 7612 reads), root-associated
ascomycetes (115 OTUs, 94 454 reads) and lichens (108 OTUs, 20 586
reads). About half of the 40 most common OTUs were root-associated
ascomycetes and most of the other were EcM fungi. The abundance of EcM
and saprotrophs decreased towards the low-alpine vegetation, while
root-associated ascomycetes showed an opposite trend of being far more
abundant in the low-alpine vegetation compared to the subalpine mountain
birch forest (Fig. 3a). Furthermore, we observed a strong correlation
between the percent C content in the dry mass of the samples and
ergosterol (R2=0.73), and percent C content and read
abundance of root-associated ascomycetes (R2=0.45;
Fig. 4) across the ecotone.
Drivers of community composition
GNMDS analyses on both datasets demonstrated a clear gradient across the
ecotone in community composition of all micro-eukaryotes (18S dataset,
Fig. 5a) and fungi alone (ITS2 dataset, Fig. 5b). In both diagrams, the
first ordination axis (GNMDS1) identified the ecotone stretching from
subalpine mountain birch forest to low-alpine vegetation as the main
gradient driving the compositional changes in soil communities. Soil
edaphic factors, together with the plant groups, were largely structured
along the same main gradient as the soil biota (Fig. 3b and c, Fig. 5a
and b). Percent C in dry mass soil and ergosterol content increased from
the birch forest to the low-alpine vegetation, while pH and P content
decreased (Fig. 3b). We observed no systematic trend for % total N
content in the soil samples. The amount of ErM forming plants increased
above the forest line, while AM plants decreased (Fig. 3c). By
definition, the EcM forming Betula pubescens was only present
below the forest line, while there was a slight increase in other
understory EcM plants towards the low-alpine vegetation (Fig. 3c). In
both datasets, the second axes (GNMDS2) were structured by the
site-specific environmental variables (Fig. 5a, b), in addition to soil
N in the ITS2 diagram (Fig. 5b).
The CCA analyses demonstrated that site-specific factors, which account
for regional (between-site) variability, explained 4.69% of the
compositional variation in the 18S dataset, and 13.68% in the ITS2
dataset (Table 1). Plot-specific factors, which also account for
variability within the individual gradients, explained 11.27% in the
18S dataset and 13.44% in the ITS2 dataset, respectively. Thus, site-
and plot-specific variables were about equally important in explaining
variation in community composition in the ITS2 dataset. Interaction
effects, both explained at site and plot level, were 3.3 (18S) and 4.2%
(ITS2). Total variation explained, as a fraction of total variation, was
25.96% in the 18S dataset, and 27.12% in the ITS2 dataset.
Species score GNMDS ordination of ITS2 OTUs (Fig. 5c) revealed the same
trends as outlined above, where OTUs distributed along the ecotone with
relatively more root-associated ascomycetes and lichen-forming fungi in
the low-alpine vegetation and relatively more EcM in the birch forest.
The largest proportion of root-associated ascomycetes was associated
with the heath part of the low-alpine vegetation.