Discussion
Species richness and green biomass
We found a humped-back relationship between species richness and green biomass that supports the intermediate disturbance hypothesis (Connell, 1978; Gao & Carmel, 2020). Moderate levels of disturbance provide favourable conditions for a wider range of species (Metera et al., 2010). The highest species richness were detected at the fourth and fifth grazing intensity levels. In accordance with our results, several studies detected that species richness is higher under moderate grazing pressure than, for example, in ungrazed pastures (Sasaki et al., 2009; Fensham et al., 2011; Deng et al., 2013). According to the results of Deng et al. (2013), the density, the height, and the cover of vegetation were the highest in the ungrazed parcel. Furthermore, the density of dominant, good competitor species had the highest scores in the ungrazed parcel. When grazing intensity is low, better competitor plants can reach greater height as they acquire more resource, and can overcome resource limitation more effectively in dense, ungrazed vegetation and grow efficiently (Westoby et al., 1999; He et al., 2021). Livestock grazing decreases green biomass and litter (Magnano et al., 2019) which is obviously more pronounced under more intensive grazing. These findings can explain the higher proportion of biomass samples belonging to the first and second level of intensity at higher biomass values (Figure 2). Although we detected a humped-back relationship between species richness and green biomass in the studied pastures, the possibility of a decreasing trend should not be excluded in unusually arid years (Milchunas et al., 1988; Gao & Carmel, 2020).
One would expect that the samples originating from the same grazing intensity levels would cluster in the ordination, but this was not validated by the results (see Figure 5). The likely reason for this is that there might be a fine-scale heterogeneity of vegetation. Fine-scale heterogeneity may have several reasons we wish to detail on below. Godó et al. (2017), studying the vegetation composition of alkaline and sand pastures at multiple-scales, observed that there is a significant relationship between the plot size and grazing effects: with increasing plot size they found decreasing levels of differences in species composition, i.e. the small-sized plots showed higher beta diversity than larger ones. This is likely the reason why our fine-scale samples (harvested in 20 cm × 20 cm plots) were not clearly separated along the gradient of green biomass and species richness. Beside of the selective defoliation effects of grazing, fine-scale heterogeneity of vegetation is also supported by multiple fine-scale side effects provided by grazing: uneven patterns of trampling, defecation, and seed input with dung or fur (Ruifrok et al., 2014). Grazing also supports the formation of bare soil patches (Eichberg & Donath, 2018) and provides higher availability of light (Ruifrok et al., 2014). But the bare soil surfaces can also act as safe sites for weedy species to colonize. In pastures intensively grazed by sheep, we can expect higher amounts of bare soil surfaces than under lower grazing intensity based on the results of former research (Watt & Gibson, 1988; Süss & Schwabe, 2007; Teuber et al., 2013). However, the seeds of several species cannot successfully germinate if they fall on bare soil surface, unless they get buried to an optimal depth subsequently, which might be ensured by the trampling of sheep (Eichberg et al., 2005).
Plant species with variable characteristics are also able to diversify the fine-scale species richness and biomass: in grazed pastures, coexistence between plant species occurs more frequently if the grazing intensity is suitably low (Vázquez-Ribera & Martorell, 2022), which contributes to a vegetation consisted of more variable plants. Briske (1996) explains in detail how the different plant characteristics contribute to a better grazing resistance, for example herbivore accessibility is limited if plant height is lower. Based on this argument, it is reasonable that plant height is more varied at moderate levels of grazing where short and tall plants are alike occurred. This idea, the distribution of our samples, and the humped-back relationship are confirmed also by Oksanen (1996) who argued and visualized that humped-back relationship can even be artificially produced if following the rules that the number of plant individuals increases below a ”crowding point” (where grazing resistant species are more frequent), but above it (where better competitor species are more frequent), plant size is what rather increases, therefore (in the sense of intermediate disturbance) the highest species richness is at intermediate biomass, if using small, fixed plot size. Beyond these assumptions, it should be added that the spatiotemporal patterns of grazed and ungrazed fine-scale patches is another important variable that reasonably becomes less marked with the extremely increased or decreased grazing intensity.
Main biomass fractions and the biomass of life forms
We found significant differences along the grazing intensity gradient for main biomass fractions and biomass of life forms as well. Both the green biomass + litter fraction and the litter fraction were significantly lower at higher grazing intensities. Consumption of plants by livestock contributes to a decreased green biomass and litter (Magnano et al., 2019). According to Kemp et al. (2000), perennial graminoids are the most sensitive to grazing, and with increasing grazing disturbance, subordinated species are enabled to spread (Grime & Mackey, 2002) which might explain why we detected significantly lower amounts of perennial graminoids at higher grazing intensities. Green biomass showed significant differences and though the biomass of short-lived forbs was significantly higher at the highest grazing intensity, total green biomass fraction decreased with increasing grazing intensity. Three explanations may be behind this pattern. First, grazing may be less selective at higher intensities (Golodets et al., 2009; Tóth et al., 2018), which causes a net loss of total green biomass, but at the same time it favours short-lived forbs, due to their fast regrowth rate and colonization ability in gaps (Westoby et al., 1999; Hofmann & Isselstein, 2004). Second, the mean height of species is typically lower at higher grazing intensities (Deng et al., 2013; Török et al., 2016), therefore, they may be represented by less biomass. Third, specific leaf area (SLA, mm2/mg) is higher, and leaf dry matter content (LDMC, mg/g) is lower in case of annuals, thus, their dry weight is lower than those with higher LDMC and lower SLA (E-Vojtkó et al., 2020). In the meta-analysis of plant responses to grazing, Díaz et al. (2007) found that the abundance of perennial plant species decreased with increasing grazing intensity. This was also confirmed by our results to an extent, but a striking leap can be observed at the fourth level of intensity for perennial forbs. Among the detected perennial forbs, Thymus glabrescens had remarkably high values at this intensity level. Without this species, there would be a continuous decline of perennial forb biomass along the gradient of grazing intensity. Their extensive woody stems increase the biomass of perennial forbs, moreover, their large amount of biomass also shapes the differences of the total green biomass scores. Phytochemicals (monoterpenes) produced by Thymus vulgaris means protection against grazers (Cappuccino & Carpenter, 2005), therefore it is reasonable to think that Thymus glabrescens in our studied pastures is similarly not favoured by grazers. The meta-analysis mentioned above (Díaz et al., 2007) details the response of annuals to grazing; it was found that they increased together with the grazing intensity which is confirmed by our results in case of forbs. Another explanation for successfully spreading short-lived plants is their generally higher SLA values compared to perennials. High SLA is linked to fast re-growth ability (Helm et al., 2019) and higher SLA scores at higher grazing intensities were confirmed by a former study addressing livestock grazing in sand grasslands (Kovacsics-Vári et al., 2023). We found no significant differences for short-lived graminoids which is presumably due to their low species number and the even distribution of the species along the gradient of grazing intensity. For example,Apera spica-venti was represented by higher amounts at lower intensity levels, meanwhile Bromus hordeaceus had higher biomass at higher intensities.
Impact of grazing disturbance on species composition
Significantly higher biomass scores of disturbance tolerant and ruderal species were detected at the highest grazing intensity level. A process leading to overgrazing is written in a simply summarized form by Schulze et al. (2019): First, the vegetation composition changes; second, vegetation cover decreases, third, bare soil surfaces are formed, and fourth, soil erosion becomes more severe. Our results might support this description, but we presume that these stages are not reached step by step but occur in parallel. For example, when bare soil surfaces are formed, further changes can be expected also in vegetation cover as gap strategists take the advantage of disturbance and they establish quickly. When grazing tolerant species with a good colonising and/or fast-regrowth ability (Westoby et al., 1999) cannot keep up with the severity and frequency of grazing the fourth step may take effect. This fourth step might be in accordance with the slightly lower species richness at the fifth level of grazing intensity compared to the fourth level. In our study, the disturbance value did not differ strikingly between the first and the fourth intensity level, however, at the fifth level, scores were significantly higher compared that of other levels. Therefore, we can conclude that the proportions of disturbance tolerant and ruderal species were considerable at the highest grazing intensity, and this is well in accordance with the proportion of short-lived forb biomass as they typically represent these groups. The high disturbance values and proportions within the simplified morpho-functional groups indicated a stronger disturbance and a process leading to overgrazing. Midolo et al. (2021) valued plant species by disturbance categories, and they detected that annuals are favoured by disturbance as their ability to grow fits to circumstances which do not provide stable biotic and abiotic features that otherwise would be favoured by better competitors using resources efficiently on the long-run (Salguero-Gómez, 2017; Schulze et al., 2019). In contrast, Pettit et al. (1995) and Farmilo et al. (2023) found no significant differences between ungrazed and grazed samples for native annual forbs.
At the highest level of grazing intensity, the highest disturbance value confirms also the observation of Botta-Dukát (2008) according to whom open sand grasslands are among the most exposed habitats to disturbance in Hungary. The Nyírség region has been densely populated for centuries, and despite the low productivity of sand grasslands, large areas were cultivated with high cover of disturbance tolerant and ruderal weedy species. Thus, the vegetation composition of pastures with opening surfaces can be likely colonised by these species having typically a good dispersal ability (Schulze et al., 2019).
The fourth and fifth level of grazing intensity included sampling sites with relatively high stocking rates (1.1 to 4 LU/ha), and the highest scores of species richness were observed at these levels which is due to the higher number of short-lived species. Increasing species richness in more intensively grazed sites was also confirmed by Kiss et al. (2006). In their assessment of grazing intensity, they considered the proximity of study sites to stables, and similarly to us, they detected a higher proportion of disturbance tolerant and weedy species contributing to the higher species richness. We detected significantly higher species richness at the fifth and fourth level of intensity compared to the second level. At the fourth level, presence of grassland species is still substantial, in addition disturbance tolerant and weedy species appeared more frequently. One can find significantly different scores if second and fourth intensity levels are compared to the third and fifth intensity levels which implies the possibility that flocks of sheep graze and trample the sites representing the third and fifth level more frequently. These significant differences suggest that stocking rate (LU/ha) and proximity jointly drive changes in vegetation. A higher grazing frequency and intensity are presumed with the higher proximity to resting and watering points. The joint impact of proximity and stocking rate on plant characteristics can be similarly explored in the study by Kovacsics-Vári et al. (2023) for flowering period and life forms. In addition, the findings by Kiss et al. (2006) and Tonn et al. (2019) suggest that the effect of livestock should be studied on finer scales than the scale of a pasture, since LU/ha, as an important metric in grazing regimes, is determined for the entire pasture area.