Amplicon library preparation and sequencing

The amplicon libraries were prepared with a two-step barcoding approach as described by Herbold et al. [79] with some modifications. First, the target markers were PCR-amplified with diagnostic primers synthesized with a 16 bp head sequence (5′-GCTATGCGCGAGCTGC-3′, modified from Rudi et al. [80]) at the 5′ end for 25 cycles. After the 25 cycles, the PCR was paused, and a second set of primers consisting of the 16 bp head sequence and a library-specific 8 bp barcode [81] was added and amplified for 5 more cycles. Each PCR reaction (45 μl in the first step, and 50 μl in second step) consisted of 10 ng of DNA template, 1 unit of Titanium Taq DNA Polymerase (Takara Mirus Bio Inc., WI, USA), 100 ng of bovine serum albumin, 1× Titanium Taq PCR buffer, 0.2 mM dNTP mix, 0.2 μM of forward and reverse diagnostic primers, and 5 μM of the library-specific barcodes (added during the last 5 PCR cycles). Thermocycler conditions were 95 °C for 3 min; 95 °C for 30 s, 60 °C (for diagnostic primers), or 52 °C (barcodes) for 30 s; 73 °C for 5 min. Obtained PCR products were inspected by gel electrophoresis, purified using magnetic beads following the protocol for magnetic purification described in the DNA extraction section, and quantified using a Qubit fluorometer. Products were equimolarly combined and concentrated by bead purification to create sequencing libraries, which consisted of 200 samples at a time (200 out the 600 samples), yielding a total of 3 libraries. One microgram of each pooled library was used for the ligation of adapters for Illumina sequencing using the NEBNext Ultra II DNA Library Prep kit for Illumina (New Englands Biolabs). Each adapter-ligated library was quantified by qPCR using the NEBNext Library Quantification Kit. Each library was spiked with 10% phiX and sequenced on an Illumina Miseq using the Miseq Reagent kit v3.

Before amplifying all our samples, we tested two-sets of primers, targeting the V1V2 and V9, 18S rRNA regions, for the characterization of soil protist communities in our soils. The V1V2 primers were those published by Fonseca et al. [82], FO4 (5′-GCTTGTCTCAAAGATTAAGCC-3′) and R22 (5′-CCTGCTGCCTTCCTTRGA-3′). The V9 primers were previously published by Amaral et al. [83], 1380F (5′-CCCTGCCHTTTGTACACAC-3′), and 1510R (5′-CCTTCYGCAGGTTCACCTAC-3′). The results showed that, when used in our soil samples, the V1V2 amplified more non-target sequences than the V9 primers (Wilcoxon test p = 4.9 × 10⁻⁷), with 46.6% and 26.4% of the total belonging to fungi (for the V1V2 and V9 markers, respectively; Supplementary Fig 9). Our analysis also showed that the V9 primers amplified significantly more sequences (Wilcoxon test p = 7.3 × 10⁻⁹) belonging to the protist division Alveolata (27.8% for V9 vs. 4% for V1V2), and also detected sequences belonging to the division Apusozoa, Hacrobia, Protalveolata, and the phyla Mesomycetozoa and Rhodophyta which were not amplified by the V1–V3 primers. Since the V9 primers outperformed the V1V2 pair in representing protists and discriminating fungal sequences (Supplementary Fig. 9), our subsequent analyses were conducted with the V9-18S rRNA primers.