.

We strongly encourage contacting Kevin Newsham (kne@bas.ac.uk ) for more information regarding this dataset, including guidance on its reuse.

Methodology:

The field experiment was deployed in November 2007 at Mars Oasis on the south-eastern coast of Alexander Island in the southern maritime Antarctic. It consisted of 64 circular plots of 1 m diameter deployed on barren fellfield soil in a periglacial habitat. The soil at Mars Oasis is free of snow and ice cover for c. 10 weeks (early December - mid February) and is permanently frozen for c. 39 weeks of each year. It has a mean pH value of 7.9, mean water potentials measured within a month of snowmelt of between -1 MPa and -7 MPa, and mean organic C and N concentrations of 0.26 percent and 0.02 percent, respectively. Three substrates were applied to the soil. The soil in 48 plots was amended yearly with either powdered glucose, glycine or tryptone soy broth (TSB), with soil in 16 plots receiving each substrate. Within each group of 16 plots, eight were warmed for the duration of the experiment with conical polycarbonate open top chambers (OTCs), and eight were irrigated annually with 1.5 L of water applied to a 0.05 m2 central area in each plot, raising soil water concentrations to 100 percent of water holding capacity. The experimental layout resulted in 16 OTC-irrigation-substrate treatments, each replicated four times in a randomised design.

Soil was sampled from the central region of each plot into sterile plastic tubes in early December 2007, 2009, 2010, 2011 and 2012, and was stored at c.-3 deg C for 96 h (2010) or 16-24 h (other years), before being frozen at -20 deg C and being transported, at the same temperature, to the UK. There, the soil from each plot was defrosted and passed through a 2mm sterile mesh under aseptic conditions. Total DNA was then extracted from 1.1g (fwt) sub-samples of soil and was dried under aseptic conditions prior to shipping to Norway, after which it was resuspended in 30 micro-litres of sterile distilled water. The internal transcriber spacer region or ribosomal RNA genes was first amplified with the primers ITS1F (5'-CTTGGTCATTTAGAGGAGTAA-3') and ITS4 (5'-TCCTCCGCTTATTGATATGC-3'). The ITS1 region was then tagged in a second, nested PCR using fusion primers consisting of the primers ITS1F and ITS2 (5'-GCTGCGTTCTTCATCGATGC-3') and one of 96 unique 6 basepair multiple identifier combinations. PCR reactions were conducted in 25 micro-litre reactions. Final concentrations of reagents were as follows: 1x gold buffer, 0.2 mM dNTPs, 1.5 mM MgCl2, 0.2 micro-metre each of the forward and reverse primers, 2 micro-litres of template DNA and 0.75 units of AmpliTaq Gold DNA Polymerase (ThermoFisher Scientific). Reaction conditions were as follows: initial denaturation at 95 deg C for 5 minutes, followed by 30 cycles (initial amplification) or 20 cycles (nested amplification) of denaturation at 95 deg C for 30 seconds, annealing at 55 deg C for 30 seconds, elongation at 72 deg C for 1 minute, and a final elongation step of 72 deg C for 7 minutes. Amplicons were purified with Agencourt AMPure beads (Agencourt Bioscience, Beverly, United States) and DNA concentrations were measured by Qubit assays. Samples were combined into equimolar pools of 96 uniquely tagged samples that were then submitted to StarSeq GmbH (Mainz, Germany) where each of the four pools of 96 samples were further indexed using the Illumina Nextera Library Prep kit before sequencing in a paired end run on an Illumina MiSeq platform.

Samples were demultiplexed and primer sequences were identified and removed from both the 5' and 3' ends of forward and reverse reads using cutadapt v.1.9.1, allowing up to 15 percent mismatch across the length of the primer. The DADA2 v1.18 package in R was used for quality filtering, error correction and chimera detection. Reads were quality filtered to remove all sequences with ambiguous bases, more than 2 expected errors in the forward direction, more than 4 expected errors in the reverse direction, and length of les...(18)

Data collection:

Instrumentation:

1. Qubit dsDNA HS assays: Life Technologies, Carlsbad, CA, USA

2. Amplicon sequencing: Illumina MiSeq platform, Illumina Inc., San Diego, USA

3. cutadapt v. 1.9.1: Martin (2011) doi: 10.14806/ej.17.1.200

4. DADA2 v. 1.18: Callahan et al. (2016) doi: 10.1038/nmeth.3869

5. RDP classifier v. 2.13: Wang et al. (2007) doi: 10.1128/AEM.00062-07

6. UNITE v. 8.3: Nilsson et al. (2018) doi: 10.1093/nar/gky1022

7. FUNguild v. 1.0: Nguyen et al. (2016) doi: 10.1016/j.funeco.2015.06.006

Data quality:

Instruments were calibrated according to laboratory practices, with standards being used where necessary. No data losses and no data cleaning. Columns MB-MG and MO contain missing values (entered as asterisks) when an OTU could not be confidently assigned to a phylum, class, order, family, genus or species, or a guild or growth form.