Libraries were prepared following the MetaFast protocol (Taberlet et al., 2018) and sequenced using the MiSeq (Bact02 and Fung02) or HiSeq 2500 (Arth02, Coll01, Euka02, Inse01, Olig01, Sper01) Illumina platforms (Illumina, San Diego, CA, USA) with a paired-end approach (2 × 250 bp for Bact02 and Fung02, and 2 × 150 bp for the others markers) at Fasteris (SA, Geneva, Switzerland). For each marker, the sequence depth corresponded to 10,000 reads per sample. USAGE NOTES: Example for Euka02 marker: The zipped folders contains raw sequence data for seven paired-end sequenced libraries. Usually two files per library are available, corresponding to the forward and reverse reads. Example of files for one library (GWM-671): 150703_SN234_A_L001_GWM-671_R1.fastq.gz 150703_SN234_A_L001_GWM-671_R2.fastq.gz When more than two files are available for the the forward and reverse reads, more than one sequencing lane was needed to guarantee the desired sequencing depth (i.e. 10,000 sequences per sample). In this case, the raw sequences of the forward and reverse sequencing were concatenated prior to bioinformatics treatment. Example of files for one library (GWM-1497): these two files were concatenated 190503_SND405_B_L001_GWM-1497_R1.fastq.gz 190503_SND405_B_L002_GWM-1497_R1.fastq.gz these two files were concatenated 190503_SND405_B_L001_GWM-1497_R2.fastq.gz 190503_SND405_B_L002_GWM-1497_R2.fastq.gz See the complete bioinformatics pipeline provided and described in the "Methods" section of the related manuscript.

Glaciers show a pattern of retreat at the global scale. Deglaciated areas are exposed and colonized by multiple organisms, but lack of global studies hampers a complete understanding of the future of these ecosystems. Until now, the complete reconstruction of soil communities was hampered by the complex identification of organisms, thus analyses at broad geographical and taxonomic scale have been so far impossible. The dataset used for this study represents the assemblages of Bacteria, Mycota, Eukaryota, Collembola (springtails), Oligochaeta (Earth worms), Insecta, Arthropoda and Vascular Plants obtained using environmental DNA (eDNA) metabarcoding. eDNA was extracted from soil samples collected from multiple glacier forelands representative of some of the main mountain chains of Europe, Asia, the Americas and Oceania. We investigated chronosequences of glacier retreat (i.e., the chronological sequence of specific geomorphological features along deglaciated areas for which the date of glacier retreat is known) ranging from recent years to the Little Ice Age (~1850). We used this newly assembled global DNA metabarcoding dataset to obtain a complete reconstruction of community changes in novel ecosystems after glacier retreat. Information on assemblages can be then combined with analyses of soil, landscape and climate to identify the drivers of community changes.