Fungal microbiomes associated with Lycopodiaceae during ecological succession

Abstract Lycopodiaceae species form an early‐diverging plant family, characterized by achlorophyllous and subterranean gametophytes that rely on mycorrhizal fungi for their nutrition. Lycopodiaceae often emerge after a disturbance, like in the Hochfeld reserve (Alsace, France) where seven lycopod species appeared on new ski trails following a forest cut. Here, to better understand their ecological dynamic, we conducted a germination experiment of lycopod spores following an anthropogenic disturbance and examined their associated fungi. Only 12% of the samples germinated, and all gametophytes were abundantly colonized by a specific clade of Densosporaceae (Endogonales, Mucoromycotina), which were also present in the roots of lycopod sporophytes, but absent from the ungerminated spores and the roots of surrounding herbaceous plants, suggesting high mycorrhizal specificity in Lycopodiaceae. In addition, ungerminated spores were profusely parasitized by chytrid fungi, also present in the surrounding lycopod gametophytes and sporophytes, which might explain the low spore germination rate. Altogether, the requirement of specific mycorrhizal Mucoromycotina fungi and the high prevalence of parasites may explain why Lycopodiaceae are often rare pioneer species in temperate regions, limited to the first stages of ecological succession. This illustrates the primordial roles that belowground microbes play in aboveground plant dynamics.

Back in the lab, the plastic slides were opened under a binocular, in sterile conditions, only up to a few hours after their extraction from the soil. If present, gametophytes were collected for microscopic observations and/or stored in 70% ethanol for molecular characterization of the associated fungi. If several gametophytes were present in the same slides, they were pooled together as the same sample for molecular analyses. We also collected some ungerminated spores in 28 slides selected at random as well as 23 root samples of some herbaceous plants that managed to enter inside the slides (and were therefore in contact with gametophytes and/or ungerminated spores). We performed microscopic observations of some gametophytes and spores using both optical microscopy and scanning electron microscopy.

Supplementary Methods 2: Molecular analyses and bioinformatics
Samples stored in ethanol, including gametophytes, were rinsed using sterile water before starting DNA extractions. For dried roots, 30 mg of tissue was crushed using sterile tungsten beads in the TissueLyser II (Qiagen). DNA was extracted using the DNeasy Plant Mini kit (Qiagen) following the manufacturer's instructions.
The sequencing results were processed using VSEARCH (Rognes et al., 2016) with a pipeline available in GitHub (https://github.com/BPerezLamarque/Scripts/) following Perez-Lamarque, Krehenwinkel, et al. (2022). Paired-end reads were assembled, quality checked, demultiplexed with cutadapt (Martin, 2011), and clustered into operational taxonomic units (OTUs) using Swarm (Mahé et al., 2015), a clustering approach based on local thresholds and amplicon abundances. Chimeras were removed de novo and we assigned taxonomy to the OTUs using Silva and UNITE databases (Quast et al., 2013;Nilsson et al., 2019). We filtered out the contaminants of the OTU tables using the decontam pipeline (Davis et al., 2018). Non-fungal OTUs were then discarded for subsequent analyses. The few samples having fewer than 10 reads were discarded in the following analyses. In addition, we also performed a classical 97% OTU clustering using VSEARCH. Yet, because they gave qualitatively similar results, only the results obtained with Swarm are presented in the main text.

Supplementary Methods 3: Statistical analyses:
To assess whether different sample types and different lycopod species were associated with different fungi, we performed PCoA (principal coordinate analysis) and PermANOVA (permutational analysis of variance) from Bray-Curtis dissimilarity matrices between pairs of samples. The PCoA enables the visualization of differences in composition of the fungal communities per sample. PermANOVA tests whether samples of different types and/or different species tend to have significantly different fungal compositions. It gives the fraction (R) of the variance that can be explained by the sample type and/or by the lycopod species. We performed PermANOVA using the 5 adonis function from the R-package vegan (Oksanen et al., 2016) with 10,000 permutations.
Then, to investigate whether the fungal OTUs were shared or not between different samples, we built plant-fungus interaction networks. We considered an association occurs between a plant sample and a fungus if the fungal OTU is represented by at least 1% of the total fungal reads of the sample, following the approach of Toju et al. (2014). By using a threshold of 1%, we thus corrected the heterogeneous number of reads per sample and avoided counting crosscontamination and spurious interactions occurring in samples with high coverage.