This data set contains information on plant community properties, grazing intensity, elevation, soil properties and density of soil decomposers collected in 2006 and 2007 from the 31 semi-natural grasslands exposed to cattle grazing along elevation gradient at large topographic scale ranging from the Carpathian Mountains, across the adjacent foothills to the plain areas. Plant community properties are represented by the following variables for each of the 31 study grassland: number of species (species number 100 m^-2), number of functional groups 100 m^-2 (that is, legumes, grasses, rushes and sedges, and non-legume forbs), proportion of undesirable weeds (%), number of species of legumes (species number 100 m^-2), number of species of grasses (species number 100 m^-2), number of species of rushes and sedges (species number 100 m^-2), and number of non-legume forbs (species number 100 m^-2). Grazing intensity is measured as cattle density (livestock units h^-1). Soil properties are represented for each 31 study grassland by bare soil exposure (%) , soil organic carbon content (%), and soil pH. Density of soil decomposers are represented by biomass of earthworms g m^-2 and abundance of soil microorganisms cells × 10⁸  g (dry soil)^-1. Plant community composition is based on individual species’ canopy cover. This dataset also contain information on the following plant traits for each plant species (n=175 species) found across the 31 study pastures across 2006 and 2007 sampling years: family of the plant species, belonging to legumes, grasses, rushes and sedges, non-legume forbs, and undesirable weeds. All data are averages across the two sampling years. 

Study area and site selection

We studied 31 semi-natural grasslands (in 2006 and 2007), that have been used for decades as public pastures for cattle grazing. The study grasslands are distributed throughout the Chernivtsi Region (47°43' – 48°41' N × 24°55' – 27°30' E) located in the south-west of Ukraine along the rivers Dniester, Prut and Siret. 

Of the 31 grasslands sites, 12 were selected within the plains, 12 within the foothills and 7 within the mountains. All grasslands have been used as common grazing lands for cattle pasturing by private households, which typically have 2-3 livestock units per household. Grazing season depends on the growing season of vegetation and varies for the three physical-geographical zones. The grazing season for mountain grasslands lasts nearly 120-150 days, for the foothills around 180 days and for the plains zone 210-220 days.

Sampling was performed identically for each of the compared ecosystems during June–July in 2006 and 2007. The same grassland sites were sampled twice, i.e. one time per year. A handheld GPS-12 Garmin^® was used to identify the geographic coordinates and average elevation (m asl) for each study ecosystem. Four plots (10 m × 10 m) were selected within each of the 31 grassland sites. Placement of plots was random but constrained by the edges and size of the field site. The average closest distance between two neighboring plots was 10 m and was chosen to minimize the potential for spatial autocorrelation influencing the results. The distance from plot to the edges of the grassland was 10 m to prevent edge effects. Within each plot, a transect was positioned diagonally through the plot. Three 1 m by 1 m subplots for estimation of bare soil, and earthworm sampling were randomly selected along the transect with a minimum distance to the next subplot of 1 m. Soil samples for microbial and chemical analysis were taken from one of the three subplots within each plot. Locations of plots and subplots within each grassland site differed across the two years.

Cattle density and bare soil exposure

Cattle density was measured as the number of livestock units per hectare of grassland area (livestock units h^-1). For this, the number, type and age of cattle were recorded for private households, which used the pastures during the grazing seasons of the two study years. We transformed the number of cattle to livestock units based on the widely used conversion factors for Europe, as follows: 1 for dairy cows, and males (2 years old and over);  0.8 for other cows (2 years old and over), including cows that have not yet calved (heifers); 0.7 for animals from 1 to 2 years old; and 0.4 for animals under 1 year old. We used this cattle density measure as a proxy for the degree of grazing intensity of the study pastures. For each 1 m² subplot, the fraction of bare soil was visually estimated. The mean of the three subplots and then the four plots was taken to approximate the percent bare soil per grassland in each year. Cattle density and the fraction of bare soil were averaged across the two sampling years.

Vegetation

       Vegetation was recorded within each plot during the peak growing season for the different physical-geographical zones (June–July) in the two study years. All plant species were determined within each plot. We used the number of species as a measure of plant species richness 100 m^-2. Within each plot three 1 m × 1 m randomly selected subplots were used to estimate the relative cover of each species by vertically projecting canopy cover (%) for each species within each subplot. Averages were then taken across the three replicate subplots. All recorded species were classified into four functional groups: legumes (Fabaceae), grasses (Poaceae), other monocots (rushes and sedges), and forbs (other than legumes). The number of functional groups was used as the measure of plant functional diversity per 100 m². Further, species were classified as undesirable weeds if they were known to reduce grazing efficiency, forage yield, palatability and quality, therefore contributing to lower forage and animal production of grassland ecosystems. The number of undesirable weed species was quantified on the 100 m² plots. Data on plant species richness, functional diversity and richness of undesirable weeds were averaged across the four plots level for each year and further averaged across the two sampling years. Further, we measured the proportion of undesirable weeds as the ratio of the species number of the undesirable weeds to the total plant species number. Data on canopy cover (%) for each species were averaged across the four plots within each year and scaled to unitless relative cover measures (C) ranging from 1 to 5 in accordance with Braun-Blanquet: C = 1 for cover from 1 to 5 %; C = 2 for cover from >5 to 25 %; C = 3 for cover from >25 to 50 %; C = 4 for cover from >50 to 75 %; C = 5 for cover >75%. Further, the relative cover measures for each species were averaged across the two sampling years.

Soil sampling and analysis

Before soil sampling, vegetation and upper litter layer were removed in a small area. One soil sample per plot (for a total of four samples per grassland) was collected at 0−10 cm depth during the vegetation sampling campaign within each of the grassland sites using a soil corer with a diameter of 5 cm. The soil samples were immediately stored at -5 ^oC for microbial and chemical analysis. For the examination of organic carbon content and soil pH analysis the soil samples were air-dried, sieved (mesh width 2 mm) and homogenized. Soil organic carbon (%) was determined using a Tyurin's wet combustion technique, which is based on organic carbon oxidation by potassium dichromate (0.4 N) in acid solution (K₂Cr₂O₄: H₂SO₄ in a 1:1 ratio). The soil pH was determined by a standard glass electrode pH meter using a potassium chloride solution in a 1:2 ratio (soil : 0.1-N KCl).

Density of soil biota

For the soil microbiological analysis, we counted cells of three microbial groups: heterotrophic bacteria, micromycetes, and actinomycetes. Cells were cultured on group-specific substrates under controlled temperature conditions: heterotrophic bacteria were cultured on meat-peptone agar between 28 and 30 °C, micromycetes were cultured on modified Czapek-Dox substrate with streptomycin at 20 to 25 °C, and actinomycetes were cultured on starch-ammonium agar at 28 to 30 °C. The total number of cells of all three groups was used as abundance measure for the soil microbial community (cells × 10⁸ g^-1 dry soil). We took averages across the four plots to approximate the abundance per grassland for each year. Further, the abundance data were averaged across the two sampling years. To sample earthworms we used a standard Quantitative Hand Sorting method. For this, 30×30 cm² soil blocks with a depth of 15 cm were excavated from each of three subplots in each of four plots (leading to 12 samples per grassland), and earthworms were immediately separated manually in the field and sampled into empty vials. At the same day specimens were counted in the lab, their fresh weight was determined, then they were oven-dried and their dry weight was determined. Earthworm dry weight data were calibrated to an area of 1 m². Averages were taken across the four plots to calculate earthworm biomass per grassland (g m^-2) for each year and further averaged across the two sampling years.