Climate change induced alterations to winter conditions may affect decomposer organisms controlling the vast carbon stores in northern soils. Soil microarthropods are abundant decomposers in Arctic ecosystems affecting soil carbon release through their activities. We studied whether increased snow depth affected microarthropods, and if effects were consistent over two consecutive winters. We sampled Collembola and soil mites from a snow accumulation experiment at Svalbard in early summer and used soil microclimatic data to explore to which aspects of winter climate change microarthropods are most sensitive. Community densities differed substantially between years and increased snow depth in winter had inconsistent effects. Increased snow depth hardly affected microarthropods in 2015, but decreased overall abundance and altered relative abundances of microarthropod groups and Collembola species after a milder winter in 2016. Although our increased snow depth treatment enhanced soil temperatures by 3.2 ⁰C in the snow cover periods, the only good predictors of microarthropod density changes were soil conditions around snowmelt. Our study underpins that extrapolation of observations of decomposer responses to altered winter climate conditions to future scenarios should be avoided when communities are only sampled on a single occasion, since effects of longer-term gradual changes in winter climate may be obscured by inter-annual weather variability.

The published dataset encompasses:

• Soil invertebrate community (density, ind. m-2) in 'Snoeco_microarhropods_Dryad.csv'

Microarthropods were identified to ‘group level’, ‘Collembola’, ‘Oribatid mites’, ‘Predatory mites’ (Prostigmata and Mesostigmata) and ‘Mite juveniles/other’ (nymphal Oribatids and Mesostigmata and nymphal Prostigmata) and counted. Collembola were identified to species or genus level. Microarthropods were sampled as described: 

Three cores were taken (ø 4.5 cm, 5-9 cm deep) from each fence/ambient plot from Salix polaris-dominated patches. In increased snow depth plots, approximately 10 m west of the snow fence, in the early summer of 2015 (15th of July) and 2016 (6th of July). Cores were taken so that the samples always contained the complete organic layer (on average approx. 5 cm thick (Semenchuk et al. 2019) in which most microarthropods can be found, as well as a part of the mineral soil (>1 cm), in which microarthropod densities are generally low or absent 63. In 2015, these cores were stored at 6⁰ C and transported to Abisko, Sweden for Tullgren extraction (12 bank Tullgren funnel, Burkard Scientific, Uxbridge, UK) within four days after sampling. In 2016, the cores were stored overnight at 6⁰ C and extracted in the same type of Tullgren extractor at UNIS in Longyearbyen the day after sampling. Both Tullgren extractions lasted for seven days to ensure the cores were completely dry. 

• Soil moisture (cm3 water per cm3 soil) in 'Snoeco_microarhropods_Dryad.csv'

Data has been collected according to description:

Soil moisture at time of sampling (in both sampled years) was determined gravimetrically from each plot using three replicate (ø 4.5 cm, 5-9 cm deep) cores from Salix polaris-dominated patches that were also used for microarthropod extraction. These cores were weighed upon sampling, subsequently dried at ~ 35⁰C for seven days during microarthropod extraction, and finally dried in an oven at 70 ⁰ C for 48 hours. Water weight was assumed to correspond to volume, and soil volumes were obtained by measuring the sampled  soil depth to determine soil moisture content (water volume/ volume).

• CWM: Commuity weighted mean for Collembola body size in 'Snoeco_microarhropods_Dryad.csv'

Data has been obtained using the following calculations: 

Average body size/length per Collembola species was determined for 30 randomly chosen individuals per species (or as many individuals as available, a minimum 10 individuals) by measuring Collembola length from the head to tip of the abdomen by a calibrated microscope (Leica, 40x magnification).  The body lengths obtained were used to calculate community weighted mean (CWM) body size of the community. We calculate the CWM for a whole community as:

where nj is the number of species sampled in community j, Ak,j is the relative abundance of species k in community j and FTk,j is the functional trait of interest of species k in community j. 

• Soil temperature (C) daily average in 'Snoeco_Soil_temp_Dryad.csv'

Soil temperature data as used for analyses in this manuscript. Soil temperature data was obtained as described:

Each plot (F and C) had a temperature logger (Tinytag data loggers, model TGP‐402 (Gemini)) placed just below the soil surface (in the increased snow depth treatment where snow depth reaches ~150 cm), but data from these loggers was not available for all plots in both years. Analyses have been performed only for plots in which data was available for both treatments (control and fence (deep snow area)) and years (n=6)

Presented data in file are average daily temperature means (inferred from hourly logged data) for those plots for which a full dataset was avaialble for both years. 

A complete description of methods and description of the experimental setup can be found in the published manuscript

Corrections applied to Soil temperature data:

As some temperature loggers showed a drift in sensor readings (<20%), observed temperatures were corrected before data analyses by defining the period before snowmelt when soil temperatures are constantly close to 0 for multiple days (the ‘zero curtain’) . For some loggers, readings in this period were consistently more than a degree above or below zero thus we corrected year-round temperatures for this deviation.