Towards rainy high Arctic winters: how experimental icing and summer warming affect tundra plant phenology, productivity and reproduction
Data files
Dec 27, 2025 version files 48.49 MB
-
JoE_IcingExp_Biomass_PIM.txt
89.92 KB
-
JoE_IcingExp_Flower_Count.txt
128.95 KB
-
JoE_IcingExp_HOBO.txt
31.80 KB
-
JoE_IcingExp_Leaf_trait.txt
97.30 KB
-
JoE_IcingExp_Moisture.txt
408.67 KB
-
JoE_IcingExp_NDVI_Daily_Prediction.txt
760.51 KB
-
JoE_IcingExp_NDVI_metrics.txt
19.30 KB
-
JoE_IcingExp_NDVI.txt
125.42 KB
-
JoE_IcingExp_Phenology_Binary.txt
3.01 MB
-
JoE_IcingExp_Temperature.txt
43.79 MB
-
README_JoE_LeMoullec2025.txt
9.69 KB
-
README.md
18.40 KB
Abstract
The Arctic is warming rapidly, and much faster in winter than in summer. Warm spells in winter lead to more frequent extreme rain-on-snow events that alter snowpack conditions and can encapsulate plants in ‘basal ice’ (‘icing’) for months. Yet, how icing affects plant communities, especially over multiple winters and under warmer summers, remains largely unstudied.
We investigated winter icing and summer warming effects on vascular plants’ productivity, reproduction and phenology in mesic dwarf-shrub heath, an important reindeer habitat in high Arctic Svalbard, where winter temperatures have been rising particularly fast. In a full factorial field experiment, rain-on-snow and resultant icing were simulated in five consecutive winters, and each followed by experimentally increased summer temperatures. Vascular plant responses at the community-level, with particular attention to the dominant dwarf shrub Salix polaris, were assessed throughout each subsequent growing season.
Icing alone increased community-level primary productivity, but only late in the growing season, and reduced inflorescence production. Accordingly, S. polaris showed delayed early leaf phenophases, but accelerated subsequent development, resulting in smaller, thinner leaves. This compensatory growth response apparently occurred at the cost of delayed seed maturation. The phenological delay was associated with icing-induced delays in spring soil warm-up, possibly favouring resource allocation to primary productivity over reproduction. Experimental summer warming (on average 0.8 °C) largely counteracted the effects of icing, enhancing community productivity throughout the growing season, offsetting S. polaris leaf size reductions and turning around its delayed phenophases, including seed dispersal. Effect sizes could be larger than under warming alone. Yet, summer warming did not negate the reduction in community inflorescence production caused by icing.
Synthesis. Extreme rain-on-snow events encapsulating plants in ice can influence high Arctic plant communities in mesic habitats to similar extents as – the better-studied – summer warming. Nevertheless, the absence of magnified icing effects over the years indicates community resistance to icing, particularly under warmer summers, which contrasts with earlier documented ice-induced die-offs in communities dominated by evergreen shrubs. As warm spells during winter become the rule rather than exception, we call for similar experiments in coordinated circumpolar monitoring programmes across the tundra biome.
Dataset DOI: 10.5061/dryad.9ghx3ffxx
Description of the data and file structure
The datasets include all raw data from the manuscript:
Towards rainy high Arctic winters: how experimental icing and summer warming affect tundra plant phenology, productivity and reproduction.
Mathilde Le Moullec, Anna-Lena Hende, Matteo Petit Bon, Ingibjörg Svala Jónsdóttir, Øystein Varpe, René van der Wal, Larissa Teresa Beumer, Kate Layton-Matthews, Ketil Isaksen and Brage Bremset Hansen.
Published in Journal of Ecology, Accepted on 31 October 2025.
Data contains information measured at the plot level on: soil temperature, soil moisture, air temperature, Normalized Digital Vegetation Index (NDVI), point intercept biomass quantification, flower count and Salix polaris leaf traits and phenology.
Common column names related to the study design, included within each datasets:
year: 2016 to 2020
block: 3 replicated units across the landscape (A1, A3, A4) with 12 plots each
plot: 36 plots with unique IDs followed throughout the years. Plots are named 1-12 in each block, so a unique plot name is e.g. A1-2
IDfull: unique information of each plot, combining block_plot_treatment_year info, e.g. A3_2_W_2018
treatment: the 4 different categories include C (control plots), I (icing plots), W (warming plots), IW (icing + warming plots). The icing treatment encased the vegetation into a layer of basal ice for several months in winter. The warming treatment consisted of placing an Open Top Chamber on the plot for the snow-free season. Each block has 3 plots with the same treatment.
icing: binomial recording of the icing treatment on a plot; 0 (no icing), 1 (icing).
warming: binomial recording of the warming treatment on a plot; 0 (no warming/Open top Chamber), 1 (warming/Open top Chamber)
round: the repetition number of a measurement throughout one same summer
DOY or day-of-year = The day of year since 1st of January that year.
Note that year, block, plot, treatment and round should be used as factors! None of these variables have a numerical meaning.
A R script is available on request: mathilde.lemoullec@ntnu.no and malm@natur.gl
Files and variables
File: JoE_IcingExp_Biomass_PIM.txt
Description: Community-level vascular plant species abundance estimation by Point Intercept Methodology (Bråthen & Hagberg, 2004). Measruements were conducted shortly after peak growth (end of July) in 2016–2019 at 25 intersections per plots. We recording all ‘hits’ of live vascular plant species and dead tissue classified as standing dead or litter.
Variables
- year: 2016–2019
- block: 3 replicated units across the landscape (A1, A3, A4) with 12 plots each.
- plot: 36 plots with unique IDs followed throughout the years. Plots are named 1-12 in each block, so a unique plot name is e.g. A1-2.
- treatment: the 4 different categories include C (control plots), I (icing plots), W (warming plots), IW (icing + warming plots). The icing treatment encased the vegetation into a layer of basal ice for several months in winter. The warming treatment consisted of placing an Open Top Chamber on the plot for the snow-free season. Each block has 3 plots with the same treatment.
- icing: binomial recording of the icing treatment on a plot; 0 (no icing), 1 (icing).
- warming: binomial recording of the warming treatment on a plot; 0 (no warming/Open Top Chamber), 1 (warming/Open Top Chamber).
- species: Corresponding to the species and groups presented in Fugure S1. alobor = Alopecurus borealis, bisviv = Bistorta vivipara, dryoct = Dryas octopetala, equarv = Equisetum arvens, luzcon = Luzula confuse, poaarc = Poa arctica, salpol = Salix polaris, other_vasc = other live vascular plants, standing dead = sdead, litter = litter, tot_vasc = total live vascular plants, vascular_noSalixAlop = total live vascular plants minus salpol and alobor.
- Count: summed number of hits per plots.
File: JoE_IcingExp_HOBO.txt
Description: Daily surface air temperature recorded 5 cm above the bryophyte layer, in summer 2016-2018. A total of six HOBO loggers were deployed: 2 per per block, one in a Warming and one in a Control plot. Data are averaged per treatment.
Instrument: HOBO loggers (type U23-003/UA-001; ± 0.2°C accuracy)
Variables
- DOY: The day of year since 1st of January
- treatment: the 4 different categories include C (control plots), I (icing plots), W (warming plots), IW (icing + warming plots). The icing treatment encased the vegetation into a layer of basal ice for several months in winter. The warming treatment consisted of placing an Open Top Chamber on the plot for the snow-free season.
- Date: Date corresponding to the DOY
- Time: Not relevant
- Temp: Temperature in degree Celsius
- RH: Relative Humidity in %, not used and not explored in this study
- year: Year of recording, 2016, 2017 or 2018.
File: JoE_IcingExp_Flower_Count.txt
Description: Flowering counts was assessed in mid-July each year by counting total inflorescence numbers within each plot (50 × 50 cm), subdivided into 16 sub-squares- Recorded species are S. polaris (males and females [catkins]), B. vivipara, A. borealis, L. confusa and P. arctica.
Variables
- year: 2016 to 2020.
- block: 3 replicated units across the landscape (A1, A3, A4) with 12 plots each.
- plot: 36 plots with unique IDs followed throughout the years. Plots are named 1-12 in each block, so a unique plot name is e.g. A1-2.
- square: the sub-square within the plot recording frame 1-16.
- treatment: the 4 different categories include C (control plots), I (icing plots), W (warming plots), IW (icing + warming plots). The icing treatment encased the vegetation into a layer of basal ice for several months in winter. The warming treatment consisted of placing an Open Top Chamber on the plot for the snow-free season. Each block has 3 plots with the same treatment.
- icing: binomial recording of the icing treatment on a plot; 0 (no icing), 1 (icing).
- warming: binomial recording of the warming treatment on a plot; 0 (no warming/Open Top Chamber), 1 (warming/Open Top Chamber).
- salix_tot: total number of S. polaris (males and females) catkins per plot.
- Bis_tot: total number of Bistorta vivipara reproductive stems (inflorescences and/or bulbils) per plot.
- gram: total number of graminoid inflorescences, suming species A. borealis, L. confusa and P. arctica.
- total: total number of inflorescences.
File: JoE_IcingExp_Leaf_trait.txt
Description:
Variables
- year: 2018 to 2020
- block: 3 replicated units across the landscape (A1, A3, A4) with 12 plots each.
- plot: 36 plots with unique IDs followed throughout the years. Plots are named 1-12 in each block, so a unique plot name is e.g. A1-2.
- Branch_id: Unique branch ID within plot.
- Leave_id: Unique leaf ID within branch or within sampling sub-square.
- Leave_id_br: Unique leaf ID within branch ID.
- treatment: the 4 different categories include C (control plots), I (icing plots), W (warming plots), IW (icing + warming plots). The icing treatment encased the vegetation into a layer of basal ice for several months in winter. The warming treatment consisted of placing an Open Top Chamber on the plot for the snow-free season. Each block has 3 plots with the same treatment.
- icing: binomial recording of the icing treatment on a plot; 0 (no icing), 1 (icing).
- warming: binomial recording of the warming treatment on a plot; 0 (no warming/Open Top Chamber), 1 (warming/Open Top Chamber).
- Dry_weight_mg: Dry mass per leaf. At the nearest 0.01 mg.
- SLA_mm2_mg: Specific leaf area (SLA) was calculated as the ratio of leaf area to dry mass.
- Area_mm2: Leaf area in mm^2. Measured on ImageJ software.
File: JoE_IcingExp_NDVI.txt
Description: NDVI handheld raw measurements at the plot level in summer 2016-2020.
Instrument: Skye SpectroSense2+ handheld device
Variables
- Date: Date of sampling, corresponding to DOY
- Day_of_year: The day of year since 1st of January that year
- year: 2016 to 2020
- block: 3 replicated units across the landscape (A1, A3, A4) with 12 plots each
- round: the repetition number of a measurement throughout one same summer
- plot: 36 plots with unique IDs followed throughout the years. Plots are named 1-12 in each block, so a unique plot name is e.g. A1-2
- treatment: the 4 different categories include C (control plots), I (icing plots), W (warming plots), IW (icing + warming plots). The icing treatment encased the vegetation into a layer of basal ice for several months in winter. The warming treatment consisted of placing an Open Top Chamber on the plot for the snow-free season. Each block has 3 plots with the same treatment.
- warming: binomial recording of the warming treatment on a plot; 0 (no warming/Open top Chamber), 1 (warming/Open top Chamber)
- icing: binomial recording of the icing treatment on a plot; 0 (no icing), 1 (icing).
- NDVI: the raw value of NDVI between 0 and 1, 4 digits.
- plotID: Combination of block and plot number to give a unique plot ID.
File: JoE_IcingExp_NDVI_metrics.txt
Description: NDVI metrics calculated based on the daily NDVI predictions from the file 4b. JoE_IcingExp_NDVI_Daily_Prediction.txt . Note that 2017 does not has sufficient monitoring rounds to estimates accurately these metrics.
Variables
- year: 2016 to 2020 (but do not use 2017).
- plot: 36 plots with unique IDs followed throughout the years. Plots are named 1-12 in each block, so a unique plot name is e.g. A1-2.
- block: 3 replicated units across the landscape (A1, A3, A4) with 12 plots each.
- treatment: the 4 different categories include C (control plots), I (icing plots), W (warming plots), IW (icing + warming plots). The icing treatment encased the vegetation into a layer of basal ice for several months in winter. The warming treatment consisted of placing an Open Top Chamber on the plot for the snow-free season. Each block has 3 plots with the same treatment.
- icing: binomial recording of the icing treatment on a plot; 0 (no icing), 1 (icing).
- warming: binomial recording of the warming treatment on a plot; 0 (no warming/Open top Chamber), 1 (warming/Open top Chamber).
- time_maxNDVI: day-of-the-year of when the maximum NDVI is reached.
- maxNDVI: maximum NDVI value between 0 and 1, use max. 4 digits.
- NDVIcumStart: early-season cumulative NDVI (‘cumulative start’) obtained by integrating daily predicted NDVI values from the first day of measurements that summer to the day when Control plots, on average, reached maximum NDVI. Use max. 4 digits.
- NDVIcumEnd: late-season cumulative NDVI (‘cumulative end’) obtained by integrating daily values from the day when Control plots, on average, reached the maximum NDVI to the last measurement day of that season. Use max. 4 digits.
- NDVIcumEntireSeason: cumulative NDVI across the growing season (‘cumulative total’) obtained by integrating daily values from the first to the last day of measurements that season. Use max. 4 digits.
File: JoE_IcingExp_Moisture.txt
Description: Volumetric water content (i.e. soil moisture) measured at 5-10 cm depth, repeated 3-15 times per summer (round) in 2017-2020.
Instrument: ML3 ThetaProbe Sensor (HH2 Soil Moisture Meter from Delta-T Devices Ltd, UK, % volume with 1% accuracy)
Variables
- Date: Date of measurement, corresponding to DOY.
- Day_of_year: The day of year since 1st of January that year
- year: 2016 to 2020.
- block: 3 replicated units across the landscape (A1, A3, A4) with 12 plots each.
- round: the repetition number of a measurement throughout one same summer.
- plot: 36 plots with unique IDs followed throughout the years. Plots are named 1-12 in each block, so a unique plot name is e.g. A1-2.
- treatment: the 4 different categories include C (control plots), I (icing plots), W (warming plots), IW (icing + warming plots). The icing treatment encased the vegetation into a layer of basal ice for several months in winter. The warming treatment consisted of placing an Open Top Chamber on the plot for the snow-free season. Each block has 3 plots with the same treatment.
- square: replicated measurement at five points per plot (in each corner [soil_moisture_1; soil_moisture_2; soil_moisture_3; soil_moisture_4] and centre [soil_moisture_5]).
- moisture: soil moisture in %.
File: JoE_IcingExp_NDVI_Daily_Prediction.txt
Description: Daily estimated NDVI values in summer fitting a generalized additive model (GAM) to each plot-year combination. We used the R function ‘gam’ from the R-package mgvc (Wood et al., 2015). Note that 2017 does not have enough replicated measurement rounds across the summer (and do not cover the peak growing season) for estimating daily NDVI values with gam.
Based on data from file JoE_IcingExp_NDVI.txt
Variables
- ID: specific ID for each plot each year, used for the gam fit.
- Day_of_year: The day of year since 1st of January that year.
- NDVI: daily estimated NDVI based on gam for each plot seperately. gam(NDVI ~ s(Day_of_year,k=K),data=PI,method="REML") with k representing the maximum number of measurement rounds.
- year: 2016 to 2020
- treatment: the 4 different categories include C (control plots), I (icing plots), W (warming plots), IW (icing + warming plots). The icing treatment encased the vegetation into a layer of basal ice for several months in winter. The warming treatment consisted of placing an Open Top Chamber on the plot for the snow-free season. Each block has 3 plots with the same treatment.
- color: or plotting purpose.
File: JoE_IcingExp_Phenology_Binary.txt
Description: Phenology of S. polaris was recorded for the most advanced vegetative and reproductive phenological stages (‘phenophases’) in each of the 16 sub-squares of the 50 × 50 cm sampling frame.
Records are binary, 0 (stage not reached) or 1 (stage reached).
Variables
- squareID: unique sub-square ID combining Block_Plot_Treatment_Square IDs.
- day_of_year: The day of year since 1st of January that year.
- block: 3 replicated units across the landscape (A1, A3, A4) with 12 plots each.
- plot: 36 plots with unique IDs followed throughout the years. Plots are named 1-12 in each block, so a unique plot name is e.g. A1-2.
- square: the sub-square within the plot recording frame 1-16.
- treatment: the 4 different categories include C (control plots), I (icing plots), W (warming plots), IW (icing + warming plots). The icing treatment encased the vegetation into a layer of basal ice for several months in winter. The warming treatment consisted of placing an Open Top Chamber on the plot for the snow-free season. Each block has 3 plots with the same treatment.
- year: 2016 to 2020.
- PhenoPart: female = female catkins development, male = male catkins developments, veg = vegetative leaves development.
- icing: binomial recording of the icing treatment on a plot; 0 (no icing), 1 (icing).
- warming: binomial recording of the warming treatment on a plot; 0 (no warming/Open Top Chamber), 1 (warming/Open Top Chamber).
- phase0: female = NA, male = NA, veg = buds not open.
- phase1: female = distinct inflorescence buds visible, male = distinct inflorescence buds visible, veg = leaves starting to unfurl.
- phase2: female = buds recognisable as female, male = buds recognisable as male, veg = leaves fully expanded.
- phase3: female = receptive stigmas, male = open anthers releasing pollen, veg = start of senescence.
- phase4: female = stigmas withered, male = anthers withered, veg = leaves fully senesced.
- phase5: female = seed dispersal, male = NA [No 5th stage], veg = NA [No 5th stage].
File: JoE_IcingExp_Temperature.txt
Description: Soil temperature recorded at different depth from January to August in each experimental plot, in different years (2016-2020).
Instrument: iButton loggers (type DS1921G-F5, ± 1.0 °C accuracy, 0.5 °C resolution)
Variables
- datetime: the exact recording time of the iButton, each 6 to 12 hours per day. Date format year-month-day
- temp: Temperature in degree celsius.
- block: 3 replicated units across the landscape (A1, A3, A4) with 12 plots each
- treatment: the 4 different categories include C (control plots), I (icing plots), W (warming plots), IW (icing + warming plots). The icing treatment encased the vegetation into a layer of basal ice for several months in winter. The warming treatment consisted of placing an Open Top Chamber on the plot for the snow-free season. Each block has 3 plots with the same treatment.
- plot: 36 plots with unique IDs followed throughout the years. Plots are named 1-12 in each block, so a unique plot name is e.g. A1-2
- depth: specific depth where the iButton logger is placed in the soil; S (2cm depth, below moss layer), D (5cm depth, soil subsurface), 10d (10cm depth, only in 2018), 20d (20cm depth, only in 2018), NA (in winter 2016, 2-5cm depth, soil subsurface).
- DOY: The day of year since 1st of January that year.
- year: 2016 to 2020
- month: month of year
- day: day of the month
- IDfull: ID composed of block_plot_treatment
- ID_year: ID composed of block_plot_treatment_year
- icing: binomial recording of the icing treatment on a plot; 0 (no icing), 1 (icing).
- warming: binomial recording of the warming treatment on a plot; 0 (no warming/Open top Chamber), 1 (warming/Open top Chamber)
File: README_JoE_LeMoullec2025.txt
Summary of this README file, to download and keep with the datafiles.
Code/software
A R script is available on request: mathilde.lemoullec@ntnu.no and malm@natur.gl
Access information
This data is associated to the open access publication:
Le Moullec et al. (2025) Towards rainy high Arctic winters: how experimental icing and summer warming affect tundra plant phenology, productivity and reproduction.
Published in Journal of Ecology, Accepted on 31 October 2025.
Please refer to the methods of the open access publication:
Le Moullec et al. (2025) Towards rainy high Arctic winters: how experimental icing and summer warming affect tundra plant phenology, productivity and reproduction.
Published in Journal of Ecology, Accepted on 31 October 2025.
