Peatlands, in particular groundwater-fed fens of the temperate zone, have been drained for agriculture, forestry and peat extraction for a long time and on a large scale. Drainage turns peatlands from a carbon and nutrient sink into a respective source, diminishes water regulation capacity at the landscape scale, causes continuous surface height loss and destroys their typical biodiversity. Over the last decades, drained peatlands have been rewetted for biodiversity restoration and, as it strongly decreases greenhouse gas emissions, also for climate protection.

With the dataset published here, we quantified restoration success by comparing 320 rewetted fen peatland sites to 243 near-natural peatland sites of similar origin across temperate Europe with regards to biodiversity (vegetation), ecosystem functioning (hydrology, geochemistry) and land cover characteristics based on remote sensing.

Vegetation data comes as species-specific cover values. Hydrology data covers on average 2.3 years and minimally one full year and comes as median, minimum, and maximum water table depth. Geochemistry consists of pH and electrical conductivity of the pore water (0-60 cm), bulk density and organic matter content of the top soil layer (0-30 cm), all sampled in summer for all sites included here alongside the vegetation data sampling. Land cover characteristics contain 208 spectral-temporal metrics for a full annual time series of Copernicus Sentinel-2 A/B data for 2018.

See methods section of the accompanying paper for details about data collection and processing; see ReadMe-file for parameter explanation.

Potential sites were found through literature search and contacting the respective authors. All such authors providing data were included as co-authors and we included all data from fen ecosystems of temperate Europe which were drained and had a dateable rewetting action and all sites without direct drainage history as confirmed by local experts and remote sensing. We included all sites that provided data for at least two of the following four response clusters in order to obtain comparable datasets for these clusters: (1) vegetation, (2) hydrology, (3) geochemistry, (4) land cover characteristics.

We included all available datasets fitting to the definitions laid out above. Sampling for vegetation and geochemistry occurred in summer for all sites. Vegetation sampling consisted of complete lists of vascular plants and bryophytes (539 species in total) based on 16 m² (median, ranging between 12 and 25 m²) with estimates of individual plant species cover. All vegetation data collections included in this study aimed at full species lists and used comparable methodologies, i.e. estimating species-specific cover values. Studies focusing on specific taxa or just reporting the dominant species were excluded from the analyses. Geochemical sampling quantified pH and electrical conductivity of the pore water (0-60 cm) and bulk density and organic matter content of the top soil layer (0-30 cm). Hydrological data relied on on continuous are at least monthly manual sampling for on average, 2.3 years, and a minimum of at least one full year. Land cover characteristics were sampled after the fact for all sites for which the required remote sensing prodcuts were available in the year 2018.

Data was collected for different purposes over different years. The data owners are included as co-authors.
Vegetation data is the estimated aboveground cover of all vascular plants and bryophytes (539 species in total) within a 16 m² (median, ranging between 12 and 25 m²) plot noted down by experts with pen on paper.
Hydrological data is based on 269 piezometers with dataloggers, 91 piezometers related to a datalogger in a transect, 216 piezometers with manual measurements of at least one year and biweekly or monthly readings of the water table depth relative to the peat surface.
Geochemical data consisted of pH and electric conductivity of the pore water extracted in the field and measured directly with portable pH-sensors and conductivity sensors. Bulk density was quantfied based on volumetric field samples (0-30cm depth) in relation to their dry weight after drying to constant weight in a drying cabinet. Organic matter was quantified as the loss on ignition of these dry samples.
Land cover characteristics: spectral-temporal metrics for a full annual time series of Copernicus Sentinel-2 A/B data for 2018. The Sentinel-2 A/B constellation provides optical imagery of the Earth’s surface between ~0.49 - ~2.2 µm in ten spectral bands and at 10 – 20 m ground sampling distance at a theoretical acquisition frequency of 2.5 – 5 days. We here acquired all available Sentinel-2 A/B imagery for 2018 with cloud cover <70% from the ESA API Hub. We used all valid observations to derive spectral-temporal metrics from the time series. Spectral temporal metrics are statistical measures (e.g. average, minimum, maximum, quartiles, …) per spectral band or index (e.g. MNDWI = (green - short wave infrared)/(green + short wave infrared)) using all available cloud- and shadow-free observations over time. The median count of clear-sky-observations per pixel across the sites is 45, while 90% of all sites featured 27 clear-sky observations or more. Both data processing to Analysis Ready Data as well as calculating spectral-temporal metrics was performed through the Framework for Operational Radiometric Correction for Environmental monitoring. Our analysis included data averaged over 3x3 pixels around the center plot location of the site. Different spatial aggregations (e.g. single pixels, 5x5 pixels around the center plot) led to highly similar results, implying that the intra-site variability was robust around locations of the vegetation survey. The inclusion of an annual series of Sentinel-1 synthetic aperture radar data (temporal metrics for VV and VH polarization, IW swath at 10 m resolution) for the same year did not affect the results.

Spatial scale: temperate fen ecosystems of Europe.
Timing: Data was collected between 1994 and 2019 with sampling for vegetation and geochemistry occurring once per site with known year and time since rewetting for the rewetted sites but different years between sites. Hydrology was monitored for >1 year at each site (see above for details and rationale), again with known time periods per site and different timing for different sites. Land cover characteristics were sampled for all sites for the year 2018 as decribed above.

Several taxa included in this dataset are at risk from a harmful human activity, in accordance to Chapman 2008 (https://docs.gbif.org/sensitive-species-best-practices/master/en/) we therefore report the georeferences denatured to 0.1 degrees (~10 km). Data may be supplied at finer scales on request under the conditions of a written data agreement.

Missing values are coded as NA, zeros are true and measured values.