Materials and Methods
Study area
We selected altogether 15 sand pastures (Table 1) for study in the
Nyírség region, East Hungary where there is a high proportion of
man-made habitats such as croplands and tree plantations (Botta-Dukát,
2008). The Nyírség region is characterised by an annual rainfall ranging
between 530–680 mm, while the average temperature is between 9.4–9.8°C
(Dövényi, 2010). In some years, annual rainfall is even less than 400 mm
and some serious drought events occur (Négyesi, 2018). The physical soil
type of the selected sites is typically coarse sand. The pH of the
studied sites ranged from 4.45–5.71, except for site 12 which is
characterised by a pH of 7.26 (site codes are found reported in Table
1). Soils of the studied pastures are rather poor in humus (0.6–2.6
m/m %). Some further characteristics of the soil of the study sites are
summarised in Table 2.
Sampling
We sampled the biomass of 15 pastures from late May to early June 2021
(Table 1). The pastures were managed by seasonal pastoral sheep grazing
(Merino breed, typically from early April to the end of October). Two
sites had been fenced for 13 years (since the summer of 2008) to exclude
livestock grazing (Aszalósné Balogh et al., 2023). With the selected
pastures, our aim was to cover a broad range of grazing intensity. For
the pasture and site selection beside livestock unit/hectare (LU/ha),
the proximities of drinking and/or resting places of livestock, the
number of droppings and other tracks of livestock grazing were
considered (see further details for intensity classification in Table 1
and Table 3). Information on current and past livestock type and
intensity levels were provided by National Park rangers, and current
grazing intensities have also been refined during the field sampling by
inspecting herders and livestock herds. National Park rangers also
helped in site selection and supported our assumption that stocking rate
(LU/ha) in itself is not sufficient for evaluating grazing intensity
levels, which was also stressed by some former studies (e.g., Tonn et
al., 2019). Considering this information, we classified our sampling
sites into five grazing intensity categories explained more in detail in
Table 3.
In each pasture, we designated a 10 m × 10 m sampling site to ensure
uniform biomass sample heterogeneity. Before the biomass sampling, we
recorded the complete list of vascular plant species in the sites to
ease the biomass sorting in the laboratory. In each sampling site, we
harvested the total aboveground biomass in ten, 20 cm × 20 cm plots
(altogether 150 samples) using secateurs. Standing litter and the litter
layer were also included in the samples. The samples were dried using a
laboratory oven (65°C for 48 hours). After drying, biomass was sorted to
main fractions such as moss, lichen, litter (including both the litter
layer and standing litter) and green biomass. Green biomass was further
sorted to vascular plant species while moss and lichen fractions were
not sorted further. The sorted biomass fractions were measured by a tare
balance (accuracy: 0.01 g).
We also collected pooled soil samples during the biomass sampling (at
least 500 g air-dried soil in total for each sampling site, pooled soil
samples collected from the ten biomass sampling plots) from the upper 5
cm soil layers in each site where biomass was harvested to characterise
the average site properties,. Soil samples were analysed in an
accredited laboratory; the physical soil type, pH, humus content,
NO2- and
NO3- contents, K2O,
P4 O10, CaCO3, and
water-soluble salt contents were assessed (Table 2).
Data processing and analyses
Supporting the functional analysis of the sorted biomass of vascular
plant species, we obtained regional plant trait data from the Pannonian
Database of Plant Traits (PADAPT, Sonkoly et al., 2023). We also
classified the species into simplified morpho-functional groups of
short-lived forbs, short-lived graminoids, perennial forbs, and
perennial graminoids using PADAPT and Király (2009). We classified the
detected species to Social Behaviour Types (SBT, a refined CSR
classification adapted to the Hungarian flora by Borhidi, 1995). Using
the SBT classification system, we grouped the species into three
categories along an increasing disturbance tolerance of the species: 1)
sand grassland species including the categories competitors (C),
specialists (S), generalists (G) and natural pioneers (NP) of sand
grasslands, 2) natural disturbance tolerant species (DT), and 3) ruderal
weedy species including the categories ruderal competitors (RC),
adventive competitors (AC), and weeds (W). For each plot, we calculated
community-weighted means (CWMs) of this ordinal variable (groups 1, 2,
and 3, respectively) weighted by biomass values and used it as an
ecological indicator value of disturbance (= disturbance value) in the
analyses. We used generalized linear mixed-effect models (GLMMs) to
assess the impact of sheep grazing (intensity level included as a fixed
factor, site identity as a random factor) on dependent variables (see
listed variables in Table 4, SPSS 26.0 program package; IBM Corp.,
2019). We plotted the species richness along increasing green biomass
and analysed their relationship by fitting second order polynomial fit.
The biomass composition of sites and grazing intensities were explored
by Canonical correspondence analysis (CCA) calculated by CANOCO 5.0
program package (Šmilauer & Lepš, 2014). We included seven variables
(soil phosphorous content, pH, soil compactness, soil nitrogen content,
soil potassium content, disturbance value, soil humus content) to
secondary explanatory matrix of the CCA and selected significant
predictors by a Monte-Carlo permutation test. Only significant
explanatory variables are shown in the figure (499 permutations,p <0.008).