CAA1, CSC2, CSC3, CIE4)

The abundance of seven OTUs representing phagotrophs was correlated to the total abundance of phototroph OTUs (fig. F.4). Six of these could be assigned taxonomically to genus level according to a threshold of 97% identity (Table S3). Amongst these species, four (Rhogostoma, Platyreta, Trinema and Pseudochilodonopsis) are large sized protists - and thus potential predators of micro-eukaryotes, including phototrophs. Our pictures illustrate three of these species in the act of predating algae (fig. F.5). Rhogostoma spp. (Thecofilosea) (X2) is closely related to recently characterized fungi- and algivores that have been shown to despise bacteria (Dumack et al., 2016b). Although Rhogostoma spp. are known to feed on bacteria (Howe et al., 2009), we could show

Rhogostoma to feed on algae as well. Platyreta (X64), which is described to feed on fungi is a member of

the exclusively eukaryvorous Vampyrellida, which are best known as algal predators (see fig. F.5, represented by the closely related Leptophrys vorax) (Bass et al., 2009a; Hess et al., 2012). Our results indicate that the terrestrial Vampyrellida, although described as predominantly fungivorous, might indeed feed on algae.

Figure F.5 – Pictures of three selected organisms, closely related to the found OTUs correlating to the phototroph sequence abundances (Rhogostoma sp. (a), Trinema sp. (b), Leptophrys vorax (c)). The scale bar represent 10µm.

(cyanobacteria and/or pigmented eukaryotes) (Meisterfeld, 2000a). Our observations confirmed the ingestion of algal material (fig. F.5). Pseudochilodonopsis (X321) are considered as exclusive algivores specialized on diatoms (Hamels et al., 2004). Labyrinthulomycetes from the Amphifilidae clade (X343) are a diverse group including bacterivores such as Sorodiplophrys stercoraria and Amphifila marina (Anderson et al., 2011). The taxonomic as well as functional diversity of this group is however only marginally documented, and the existence of algivorous forms is thus possible. The group of Spumella-like chrysophyte (X12) is composed of small phagotrophic flagellates having lost their photosynthetic abilities secondarily. However, it has been shown that transitions between phagotrophic and phototrophic strategies occurred often in the evolutionary history of chrysophytes. It is possible therefore that the Spumella-like chrysophyte X12 is actually mixotrophic like many chrysophyceae (Boenigk et al., 2005b), and therefore shares higher light requirements with other phototrophs. Alternatively, it is possible that the Spumella-like chrysophyte X12 feeds preferentially on bacteria that are associated to phototrophs and their exudates. Bacterial communities associated to algae are highly influenced by the host in aquatic systems (Sapp et al., 2007). A similar explanation could possibly be given for Allapsa (X54), a genus of small cercozoan flagellates formerly collectively classified under the name "Heteromita globosa" (Howe et al., 2009).

To the contrary, OTU X34 is assigned to the Viridiraptoridae, a family of highly specialised organisms feeding as yet known exclusively on phototrophic organisms (Hess and Melkonian, 2013). The LM obtained for this OTU did not respect the conditions of homoscedasticity because of its high abundance in two samples. Such high sequence abundance may correspond to local blooms of these small flagellates, which are reported as frequent (Hess and Melkonian, 2013).

Altogether, phagotrophic sequences belonging to an OTU co-occurring with phototrophs reaches 26.9% of all phagotrophs (28.1% if X34 is considered). Under a pessimistic scenario, if only those organisms that have been observed eating algae are playing actually that role, 19.8% of all phagotrophs actually correlate with phototrophs in a putative trophic way. The total amount is estimated between one fifth and one third of all phagotrophic sequences, an amount which is far from being negligible. This fifth to third of phagotroph sequences (139’453 to 197’911 sequences) represent from three to five time the amount of phototrophic sequences. This ratio suggests that the standing biomass of soil microalgae is lower than that of their predators (Giner et al., 2016). By analogy to aquatic ecosystems, this can be explained by the faster turnover of phototrophs. It is therefore possible that the phagotrophs whose abundance is significantly correlated to that of phototrophs indeed primarily feed on them, although this correlation does not constitute a formal proof.

Our data suggest that eukaryotic phototrophs represent an alternative nutrient source for phagotrophic protists in soils, in addition to the traditional “soil microbial loop". As in marine systems (De Vargas et al., 2015), metabarcoding of soil micro-eukaryotes shows that trophic interactions are much more complex than previously thought, and integrate components that have hardly been studied before. These trophic relationships inferred from correlative analysis of metabarcoding data need to be further explored. Combining traditional

Alternative pathway for the soil microbial loop 117 culture-based approaches, direct microscopic observation, high throughput sequencing of bulk soil samples or isolated phagotrophs, as well as new statistical tools will be the next steps for unravelling other potential relationships between protists. Nevertheless, we argue that what is now most needed is to characterise the many unknown OTUs, and conducting good observations on these organisms to provide useful natural history background needed for sound interpretation of HTS data. As suggested by our study, we believe, that future studies providing exact identities of the huge amount of unknown OTUs and suggesting their life styles and ecology, will provide sound interpretation of the ever-increasing massive sequencing data.

Acknowledgements

We would like to thank Kathleen Hasler for the field work, and Marion Cartier and Christophe Paul for the laboratory analysis. We would also want to thank the BDM program for providing basic data about the sites. This study was partly funded by Swiss National Science Foundation projects no. 310003A 143960 to EL. LB is supported by the European Union’s Seventh Framework Programme under grant agreement 245268 (ISEFOR). The authors declare no conflict of interest.

Appendix G