3 RESULTS

3.1 Sequencing and bioinformatic pipeline

We obtained 550 OTUs from 1.381.839 sequences (Cercozoa) and 331 OTUs from 1.610.374 sequences (Oomycota). The average number of cercozoan OTUs was 516 ± 15 and 546 ± 3 per microhabitat and tree species, respectively, while the average number of oomycete OTUs was 236 ± 25 and 304 ± 4 per microhabitat and tree species, respectively. Tree canopies contained a substantially unknown diversity of oomycetes, with 57 oomycete OTUs with less than 70% percent identity to any reference sequence, while this was the case for only three cercozoan OTUs (Figure 1). While most of the reads and OTUs showed a similarity of 97-100% to any known reference sequence, oomycete reads revealed two additional peaks at about 76% and 87%, indicating that a small number of highly abundant OTUs in oomycetes is still not taxonomically recorded (see Supplementary Figure 1 and Supplementary Tables 1 and 2 for taxonomic composition and annotation).

3.2 Alpha diversity

The used taxon-specific primers recovered the majority of OTU richness from canopy and ground samples (soil and litter) as indicated by rarefaction curves (Supplementary Figure 2). The extrapolation revealed that doubling the sequencing depth would have yielded no more cercozoan and oomycete OTUs, respectively, for all samples combined. All sampled microhabitats showed high alpha diversity (Figure 2), except for oomycetes in the ground leaf litter (ANOVA F value = 10.79, p <0.001; Figure 2B).

3.3 Beta diversity

Most variation in protistan beta diversity was explained by microhabitat differences (permANOVA, Cercozoa: R 2 0.45, p = 0.001; Oomycota: R 2 0.30, p = 0.001). Another significant proportion of beta diversity was explained by the differences between the two strata ground and canopy (permANOVA; Cercozoa: R 2 0.13, p = 0.001; Oomycota:R 2 0.14, p = 0.001). Only the beta diversity of the oomycete communities, but not that of the cercozoan, differed between tree species, yet with only a small amount of explained variance (permANOVA; Cercozoa: R 2 0.04, p = 0.1; Oomycota: R 2 0.06, p = 0.01) (Supplementary Table 5). Non-metric multidimensional scaling of protistan community profiles revealed a strong separation of canopy and ground strata (Figure 3). In Cercozoa, communities inhabiting fresh leaves in the canopy were most distinct to those in mineral soil on the ground (Figure 3A). Leaf litter communities on the ground scaled closer to fresh canopy leaves than to the underlying soil communities. A clear transition in beta diversity occurred from fresh canopy leaves to deadwood, bark and lichen, and finally the moss communities (Hypnum sp. andOrthotrichum sp.). Communities of arboreal soil were highly variable, ranging from samples with high similarity to mosses to samples closely resembling the mineral soil communities underneath the litter layer. Also in oomycetes (Figure 3B), canopy and ground communities were most distinct. Again, the two ground samples, mineral soil and leaf litter, showed no overlap, with leaf litter having a low alpha diversity. In contrast to Cercozoa, oomycete communities of canopy microhabitats showed a high overlap, and communities of fresh canopy leaves were clearly distinct from litter on the ground. The only common pattern between cercozoan and oomycete communities in the canopy was their high variability in the arboreal soil.
A db-RDA showed the same pattern, but revealed a bit clearer the similarity of communities of fresh canopy leaves to deadwood in both protistan taxa (Figure 4). For the Cercozoa (Figure 4A), both axes explained approximately the same amount of variance. Bark and epiphyte communities were again more similar to each other; thus these factors explained the same kind of variance in the community compositions. The soil communities were clearly different, while leaf litter clustered together with fresh leaves and deadwood communities. Little variance was found between tree species and arboreal soil. In the Oomycota (Figure 4B) a similar pattern could be observed. Bark and epiphytes communities were similar, and soil - together with leaf litter communities - appeared as separated from the other microhabitats. Tree species communities appeared further away along the second RDA axe, which however only explained approx. 5% of the variance.

3.4 Shared OTUs

Despite high differences in beta diversity, the majority of OTUs were shared between all microhabitats irrespective of the protistan phylum (Figure 5). Only few combinations yielded more than ten unique OTUs shared between distinct microhabitats, which is negligible given the high OTU richness per sampled microhabitat, which varied between 498 (deadwood) and 537 OTUs (fresh leaves) for the Cercozoa (Figure 5A) and between 189 (leaf litter) and 270 (Orthotrichum moss) for the Oomycota, respectively (Figure 5B). Because almost all OTUs were shared between all microhabitats with the species accumulation curve showing only a flat increase, communities revealed no patterns of nestedness (Supplementary Figure 3, Supplementary Figure 4).