Ecological Research

, Volume 33, Issue 1, pp 213–223 | Cite as

Riverine wood-pasture responds to grazing decline

  • Dušanka Krašić
  • Elli Groner
  • Minucsér Mészáros
  • Tijana Nikolić
  • Dimitrije Radišić
  • Stanko Milić
  • Marko Kebert
  • Dubravka Milić
  • Ante Vujić
  • Zoran Galić
Original Article


There is insufficient available information on structural changes within wood-pastures including their relationship to abiotic influences such as livestock grazing, flooding and available soil nutrients. In this paper, we address the links between important environmental variables and different stages of the wood-pasture cycle, with the aim of understanding fluctuations in this relationship and processes that follow changes in wood-pasture condition. We used satellite and aerial image interpretation to identify structural vegetation shifts over 44 years under significantly declining livestock numbers. We used ground truthing of 24 plots to assess the current field scenario and employed canonical correspondence analysis (CCA) to evaluate the relationship between plant communities and environmental influences. Three dominant structural vegetation types grassland, transitional vegetation with thorny shrubs and woody encroachment were surveyed and the following set of variables was chosen: grazing intensity, inundation frequency, elevation, soil total nitrogen, soil available phosphorus, soil potassium, soil magnesium, soil calcium, soil pH and soil carbon to nitrogen ratio. Interpretation of satellite images revealed dominance of wood-pasture in the past, which alternated structurally between more open and more closed physiognomies. CCA with ground truthing data and forward selection revealed grazing intensity as the predominant ecological driver modifying vegetation structure, as well as transitioning vegetation patterns between open herbaceous and closed woody cover. Each structural vegetation type demonstrated a collective distribution pattern and a close relationship to certain abiotic drivers, indicating strong interactions between soil parameters, grazing pressure and vegetation composition.


Wood-pasture Cannonical correspondence analysis Satellite images Vegetation composition Soil properties 


Lowland wood-pastures of Europe provide valuable ecosystem services to the human societies while having exceptional biodiversity (Bergmeier and Roellig 2014). Apart from perhaps representing modern analogues to the primaeval natural vegetation in landscapes where large, now-extinct herbivores once lived (Vera 2000; Pokorný et al. 2015), centuries-long development of wood-pasture utilization as grazing grounds places these among the oldest land-use types in Europe (Bergmeier et al. 2010). Riverine wood-pastures represent a special type of wood pastures due to their highly dynamic nature with grazing and seasonal flood dynamics as occurring disturbances. Rich biodiversity, especially regarding plant and bird communities, makes these wood-pastures a rare component of European historic landscapes with exceptional conservation value.

At present, understanding the underlying drivers of spatial patterns of heterogeneity in plant species composition is an important stepping stone towards restoration, conservation and management of riverine wood-pasture.

Many species of conservation importance in Europe depend on traditional land use practices for habitat preservation (Miller 1996; Vos and Meekes 1999; Hampicke 2006; Plieninger et al. 2006; Falk 2014). As such they are sensitive to land abandonment as well as land use intensification, with both of these scenarios increasing in European landscapes. Simultaneously, the number of preserved riverine wood-pastures in Europe has seen a significant decline with constantly changing plant species compositions (Peterken and Hughes 1998; Bergmeier et al. 2010; Cízková et al. 2013). The process of wood-pasture degradation is largely the result of neglect, most often due to the cessation of management (Hartel et al. 2014; Carboni et al. 2015). Being highly management-dependent, these areas of grazed pasture with open-grown veteran trees require carefully tailored grazing regimens since fluctuations in livestock numbers and grazing intensity strongly affect ecosystems by altering the composition of plant assemblages (Augustine and McNaughton 1998; Hendricks et al. 2005), the spatial heterogeneity of plants (Adler et al. 2001), succession pathways (Briske et al. 2003; Milchunas and Vandever 2014) and soil properties (McNaughton et al. 1997; Teague et al. 2011). The observed trend of decreasing livestock numbers in Europe will inevitably bring about some of the aforementioned changes, though the specifics are yet unknown.

In these highly dynamic ecosystems, where flooding occurs anywhere from periodically to frequently, disturbance is regarded as the principle driver of vegetation composition rather than resource competition (Grime 1974). Nevertheless, many studies have confirmed nutrient-limited plant growth in floodplain grasslands (Van Oorschot et al. 1998; Joyce 2001; Olde Venterink et al. 2001; Ogden et al. 2002; Beltman et al. 2007; Roos et al. 2009). Since different nutrients limit the growth of different plant species within a plant community (Braakhekke and Hooftman 1999; Güsewell et al. 2003), prevalence of a certain nutrient will favor certain plant species, thus influencing species composition (Güsewell et al. 2003; Güsewell 2004). This means that soil properties and soil nutrient gradient may affect variation in plant species composition (Stewart and Kantrud 1972; Sollins 1998; Tzialla et al. 2005; Lorenzo et al. 2007; Ren et al. 2013). Beltman et al. (2007) suggest that flooding and nutrients alternate in dominating the governance of plant species composition in floodplains, and is synchronized with the flooding cycle (i.e. the disturbance frequency).

At sites under existing land use, such as wood-pastures, the intensity of land use may outcompete abiotic factors, such as flooding events, in affecting the distribution and composition of plant communities (Adams and Perrow 1999; Dangol and Shivakoti 2001; Veen et al. 2008; Kalusova et al. 2009; Semmartin et al. 2010; Yoshihara et al. 2010).

Identification of structural changes in vegetation for riverine wood-pastures under decreasing livestock numbers may provide a useful tool for developing and implementing restoration strategies and to facilitate sustainable management of these wood-pastures. Such studies can be used to postulate the mechanisms controlling plant competition, help in understanding the dynamics of riverine wood-pasture ecosystems, aid in projecting future succession pathways, as well as inform management of these ecosystems in a sustainable way. The objective of this study was to improve our understanding on wood-pasture dynamics and processes that follow changes in the shifting mosaic of open landscape (grassland), shrub and mature trees (grove) i.e. wood pasture cycle. Besides inspecting historical changes in the vegetation structure we focused on the current impact of environmental factors on vegetation composition on Krčedinska ada.

By using satellite and aerial image analysis, we aimed to capture the effect of grazing decline on a wood-pasture system over a period of 44 years, providing an insight into shifting vegetation physiognomies to assess the dynamics of the two main vegetation types in wood-pastures, specifically on grasslands and woody encroachment. Additionally, we examined correlations between plant species composition and ground-truthed site data considering grazing intensity, environmental variables such as elevation, inundation frequency and soil characteristics.

Materials and methods

Study area

Krčedinska ada is a Danube river island located in North Serbia, in the southeast of the Bačka region (Fig. 1) and is part of a larger floodplain complex—Special Nature Reserve Koviljsko-petrovaradinski rit—that was designated a Ramsar site in 2012. It is one of the biggest river islands in the Serbian Danube. It is semi-circular in shape, the base of which is 4.5 km long and the arc is 10.2 km long. Its total surface area is 870 ha (Fig. 1). The region is characterized by a moderate continental climate, with an annual precipitation of 550–600 mm, with 12–13% of the total annual precipitation received in June. The island is periodically and partially flooded. Geologically, the whole alluvial plain is a deposit of recent sand and silt accumulated from river runoff and floods, with spatially variable sedimentation rates that constantly change the microrelief of the island.
Fig. 1

Location of the study area (Krčedinska ada, Danube river island), Northern Serbia

The island has several vegetation types, but is dominated by grasslands, with scattered shrubs and woodland fragments covering the island, thus creating a mosaic of vegetation with different successional stages. The presence of open-growing, veteran willow trees on Krčedinska ada demonstrate that the island has been historically shaped as wood-pasture.

Using the classification by Plieninger et al. (2015), we distinguished two types of wood-pasture on Krčedinska ada: type I—pastures in open woodlands, with trees as primary cover and grassland as secondary cover (tree crown density is > 10%); and type II—pastures with sparse open-growing veteran willow trees where grassland is the primary cover (tree crown density is 5–10%). Natural woody vegetation is dominated by Salix alba L., Salix cinerea L., Salix fragilis L., Populus alba L., Ulmus minor Miller, Crataegus nigra Waldst. & Kit., Rubus caesius L., and Rosa canina agg. The woodland understorey has been invaded by the indigo bush (Amorpha fruticosa L.), green ash (Fraxinus pennsylvanica Marshall.) and box elder (Acer negundo L.), all of which are invasive alochtonous species that prefer flooded and other disturbed habitats (Rejmánek et al. 2005). The first records on the presence of these aforementioned invasive species in the area (5–7 km from Krčedinska ada) come from 1972.

Historical records indicate the nearby village of Krčedin as being an inhabited settlement with livestock farming since the sixteenth century, meaning that livestock management has been practiced in the area for more than 400 years. According to historical records and interviews with landowners, a century ago, grazing was carried out by sheep, pig, horse and cattle, with an intensity of ca 10,000 animals (i.e. 11 ha−1). The number of livestock has significantly diminished over the past few decades and, currently, stands at 1770 heads (2 heads/ha), represented by cows, donkeys, horses and pigs roaming freely on the island. The majority of livestock is held on the island for a period of 7–8 months (between March and October), with the exception of horses which, typically are kept permanently on the island in the dry years. There are no fences and livestock roams freely on the island.

Sampling and field investigation

The study was conducted over 3 years, from 2012 to 2014. All field measurements were made during the peak growing period (May and June).


We arranged plots following a stratified random sampling scheme to allow comparison of different successional phases consisting of the main variations in the change of wood-pasture condition by inspecting changes in soil properties and the aboveground plant community. Data on our chosen set of environmental variables were obtained from plots (n = 24) comprised of three vegetation forms, each comprised of 8 plots: grassland (GR), woody cover (WC) and transitional plots (TR). GR plots were those identified as wood-pasture type I, with open-growing veteran trees and a high percentage of herbaceous land cover; WC plots were coppice-like plots with a predominance of tall shrubs and tree species with a height exceeding 2 m; and TR plots were designated as transitional between grassland and thorny shrub cover with a height less than 2 m (Table 1). Both WC and TR plots were identified as wood-pasture type II in satellite and aerial images from 1969.
Table 1

Description of investigated vegetation types in Krčedinska ada

Vegetation type

No. of species per plot (mean ± SD)

Dominant species

Companion species


8.5 (± 3.3)

Salix alba

Rubus caesius, Vitis riparia, Amorpha fruticosa, Fraxinus pennsylvanica, Crataegus nigra


13.8 (± 2.7)

Cynodon dactylon

Potentilla reptans, Mentha aquatica, Agrostis alba, Taraxacum officinale, Potentilla anserina


9.8 (± 2.9)

Crataegus nigra

Rubus caesius, Potentilla reptans, Rorippa sylvestris, Polygonum hydropiper, Lysimachia nummularia

WC woody cover, GR grassland, TR transitional vegetation with thorny shrub

For each plot, we recorded plant species abundance using the Braun Blanquet scale (Braun-Blanquet 1964) and recorded the coordinates and elevation (E) by GPS (GARMIN etrex20). In order to account for plant life forms (i.e. grass, shrubs and trees) we used two different plot sizes to assess the vegetation: 2 × 2 m2 for GR and 10 × 10 m2 for both WC and TR plots. Concerning the impact of differently sized plots on our results, plots in our study are with quite large variability in species composition therefore do not distort the real vegetation differentiation in ordination analysis (Otypková and Chytry 2006).

Soil samples

In each plot, a composite soil sample was made by mixing three randomly chosen corner samples from 0.5 m × 0.5 m sized sampling frames. Soil samples were collected using a soil core sampler with a 2.5 cm core diameter (30 cm in depth), after which they were placed in plastic bags and transported to the laboratory within 24 h. Soil samples were analyzed for carbon to nitrogen ratio (C/N), total nitrogen (TN), soil acidity (pH), plant available phosphorus (AP), plant available potassium (AK), magnesium (Mg), and calcium (Ca). Total nitrogen and carbon content were determined using a CHNS analyser (AOAC 2000). Soil pH was determined in a 1:5 soil water suspension (volume fraction) using a potentiometer. Plant available phosphorus was determined spectrophotometrically and plant available potassium content was evaluated by flame photometry. Magnesium and calcium contents in the soil samples were determined by microwave-assisted digestion, in accordance with USEPA Method 3051A using the Milestone Ethos 1 microwave sample preparation system. Analysis was subsequently performed using ICP-OES (Varian Vista Pro-axial). Quality control was periodically carried out with the IRMM BCR reference materials CRM-141R and CRM-142R, with recoveries being within 10% of the certified values. The methods for analysing concentrations of Mg and Ca are given in Gulan et al. (2013).

Grazing pressure

Vegetation response to grazing differs between plant communities implying that a heavily grazed community may have been recovered at the time of sampling (Milchunas et al. 1988; Westoby 1989). Additionally, vegetation biomass may be influenced by flooding frequency. Taking these factors into account we have repeated plot visitations for 3 years during May and June and recorded four categories of grazing pressure (GI); high, moderate, light and no grazing. High utilization was designated to plots with approximately 60% of the palatable herbs utilized by grazers; moderate utilization when 30–40% of vegetation was removed; and light where 10–20% of vegetation had been removed by grazers.

Inundation frequency (Ind)

Inundation frequency (Ind)—expressed as the average number of inundation days per year, was estimated using flooding maps showing various degrees of flooding and a hydrological database for the period of 1969–2013 received from the Hydrometerological service of the Republic of Serbia. Inundated areas were vectorized from a series of available high resolution images from Google Earth. Data on inundation frequency were categorized according to a five-level ordinal scale (see Table 2).
Table 2

Ordinal scaling of data on inundation frequency

Ordinal scale

Inundation frequency (days year−1)


< 15








> 90

Satellite image interpretation

To perform the geospatial evaluation of land cover change on Krčedinska ada, we used satellite and aerial images from different periods, which were verified by field data. The analysis was performed in ArcGIS 10.1. The research area was divided into a grid, with cell sizes of 50 m (covering an area of 2500 m2 and a total of 3513 cells). The percentage of land cover and vegetation type in each cell was determined by visual interpretation of images. Seven categories of land cover were defined: water, woody cover, wood-pasture type I, wood-pasture type II, plantation, sandy shore and transitional shrubbery. The first available image suitable for analysis was obtained from the archives of the CORONA program. The panchromatic CORONA image was acquired on 7 February 1969, with a resolution of 2.5 m. The dense snow cover visible on the image facilitated further distinctions of different surface types. Another available scanned aerial photograph from a 1969 geodetic survey was used only for partial verification purposes, as it was heavily damaged. For the assessment of recent land cover conditions, a high resolution image from Google Earth from October 2013 was selected. The land cover grid datasets from 1969 and 2013 were compared using the raster overlay function, by subtracting values for different vegetation types, with resulting values ranging between 100 and − 100, thereby indicating the intensity of change for the particular vegetation type. The resulting raster showing the intensity of vegetation type change was reinterpolated for better visual representation using the centroid points of each grid cell.

The summary land cover maps for 1969 and 2013 were visualized using the point density cartographic method (Kaim et al. 2016). The six land cover categories were represented with different symbols on the map. The map was divided in a grid, with a cell size of 50 m. One symbol depicts 10% of the total land cover within one grid cell. In each grid cell a total of 10 symbols show the land cover composition.

Data analysis

In order to detect relationships between plant species and selected environmental variables, we performed multivariate analysis in the Canoco 5 software (Lepš and Šmilauer 2003). First, detrended correspondence analysis (DCA) was performed to detect gradients of compositional change and select a proper ordination method (Ter Braak and Šmilauer 2002). The length of the first DCA axis (4.33) reflected a unimodal relationship between plant species and our chosen set of variables. Multicollinearity among explanatory variables was checked by examining Variance Inflation Factor (VIF) provided in Canoco 5 (Ter Braak and Šmilauer 2002). All VIF values were below 10, indicating the lack of multicollinearity among explanatory variables (Neter et al. 1996; Montgomery et al. 2001). In order to evaluate the relative importance of grazing intensity and soil variables in explaining community composition we performed canonical correspondence analysis (CCA), which is highly compatible with our data allowing us to quantify the amount of variation that could be explained by each matrix and arranges the graph so that the first axis is the most important one. A Monte Carlo permutation procedure with 499 unrestricted permutations was done to test for significance of variables in explaining the plant species composition. In order to underline the best subset of predictors that explain most of the inertia in the dataset, we performed forward selection (Ter Braak and Šmilauer 2002) that excluded variables explaining < 10% of the variation in the data (cut-off point P = 0.10). Variables were added sequentially until there was no improvement in model statistics.


Satellite image interpretation

Satellite image interpretation revealed that the percentage of total area under wood-pasture (types I and II) in 1969 was 73% or 648 ha but by 2013 that had decreased to 30% or 263 ha representing a loss of wood-pasture of 54% (Figs. 2, 3). In contrast, the percentage of land covered by shrub and tree species had increased over time. Former large expanses under adult tree cover and a significant proportion of open areas (wood-pasture type II) had been converted into coppice-like land cover comprised of thorny shrubs.
Fig. 2

Differences in the percent vegetation cover over a period of 44 years (1969–2013) under declining livestock abundance. WP I—wood pasture with trees as primary cover and grassland as secondary cover (tree crown density is > 10%); WP II—wood pasture with grassland as primary cover with sparse open-growing veteran willow trees (tree crown density is 5–10%)

Fig. 3

Study area map showing intensity of change including wood-pasture loss from 1969 to 2013. Land cover change for each 50 × 50 m raster was assessed from panchromatic CORONA image in 1969 and a high resolution Google Earth image in 2013

Relationship between plant species composition and environmental drivers

The overall number of plant species recorded in 24 plots was 48, with 26 families, 4 invasive plant species and 11 weed species. The most common family was Rosaceae, with Potentilla reptans L. and Potentilla anserina (L.) Rydb. as the dominant species on GR plots. The second-most common family was Poaceae, with Cynodon dactylon (L.) Pers. and Festuca pratensis Huds.

The CCA revealed a significant correlation betweeen the vegetation and the total set of environmental variables (P = 0.002), explaining 56.2% of species variance (i.e. showing that both the test on the first axis and the test on all axes were highly significant, P = 0.002 with 499 permutations). The CCA diagram (Fig. 4) displays the positions of 24 plots and 10 explanatory variables. GR, WC and TR plots are clearly separated with very little overlap in characteristics indicating an impact of examined variables on segregation in plant communities. In terms of species segregation, invasive tree species grow where grazing and waterlogging are absent, while herbaceous weed species are concentrated in areas with high grazing intensity and a high degree of waterlogging (Fig. 5).
Fig. 4

The CCA biplot diagram showing plot variations along environmental gradients (open circle GR plots, open square WC plots, times TR plots). Arrows represent environmental variables. GI grazing intensity, Ind Inundation frequency, E elevation. Soil variables: AP available phosphorus, TN total nitrogen, K potassium, Mg magnesium, Ca calcium, pH soil pH, C/N carbon to nitrogen ratio. Underlined variables explained the highest proportion of variation in species data in forward analysis

Fig. 5

The CCA biplot diagram showing species variations along environmental variables. GI grazing intensity, Ind Inundation frequency, E elevation. Soil variables: AP available phosphorus, TN total nitrogen, K potassium, Mg magnesium, Ca calcium, pH soil pH, C/N carbon to nitrogen ratio. Plant species abbreviations: AcerN = Acer negundo, AgrpR = Agropyron repens, AgrsA = Agrostis alba, AlthO = Althaea officinalis, AmbrA = Ambrosia artemisiifolia, AmorF = Amorpha fruticosa, AmarL = Amaranthus lividus, ArctL = Arctium lappa, CarxD = Carex distans, ChenA = Chenopodium album, CirsA = Cirsium arvense, ConvA = Convolvulus arvensis, CratN = Crataegus nigra, CyndD = Cynodon dactylon, DiplM = Diplotaxis muralis, EuphL = Euphorbia lucida, FestP = Festuca pratensis, FraxP = Fraxinus pennsylvanica, GaliP = Galium palustre, GratO = Gratiola officinalis, InulB = Inula britannica, JuncC = Juncus compressus, MentA = Mentha aquatica, MentP = Mentha pulegium, MyosS = Myosotis scorpioides, PoaP = Poa pratensis, PlanL = Plantago lanceolata, PlanM = Plantago media, PolgA = Polygonum aviculare, PolgH = Polygonum hydropiper, PotnR = Potentilla reptans, PotnA = Potentilla anserina, RanuR = Ranunculus repens, RorpS = Rorippa sylvestris, RubsC = Rubus caesius, RumxH = Rumex hydrolapathum, SalxA = Salix alba, StacP = Stachys palustris, TarxO = Taraxacum officinale, TrifR = Trifolium repens, VerbO = Verbena officinalis, VernC = Veronica catenata, VitiR = Vitis riparia, XantS = Xanthium spinosum

Overall, species composition was separated by the first axis, which represents the grazing gradient (intraset correlation, R = − 0.865) and inundation gradient (R = − 0.675), whereas the second axis represents the elevational gradient (R = − 0.364).

According to CCA analysis, 18% of variance in plant composition was explained by the first axis (Table 3). The CCA generated the longest arrows for grazing intensity (GI), total nitrogen (TN), soil magnesium (Mg), available phosphorus (AP) and inundation frequency (Ind) (Fig. 4), which appeared to be the most important in governing plant species composition along CCA axes 1 and 2 and as demonstrated by the intraset correlations (Table 4). This percentage of variance explained by our variables is high considering the number of species in our dataset. Combined, the first two axes explain 27% of the variability in the dataset, meaning that the species are mainly separated along the first axis. The main variation along the second axis related to pH, E, C/N, TN and K. Forward analysis showed grazing intensity as the most prominent variable, accounting for 25.9% of the total variability. Apart from GI, the following variables were significant at P < 0.2: total nitrogen (TN), available phosphorus (AP) and soil calcium content (Ca).
Table 3

Summary of the CCA ordination analysis






Total inertia







Pseudo-canonical correlations





Cumulative percentage variance of response data





Cumulative percentage variance of fitted response data





Sum of all eigenvalues


Sum of all canonical eigenvalues


Table 4

The intraset correlations and cannonical coefficients for axis 1 and 2 for environmental variables provided by CCA

Environmental variable

Intra-set correlation

Cannonical coefficients

Axis 1

Axis 2

Axis 1

Axis 2

Available phosphate (AP)

− 0.698

− 0.288

− 0.289


Elevation (E)


− 0.364

− 0.162

− 0.324

Inundation frequency (Ind)

− 0.675


− 0.065

− 0.183

Grazing intensity (GI)

− 0.865



− 0.067



− 0.596



Calcium (Ca)


− 0.113

− 0.028

− 0.743

Magnesium (Mg)


− 0.078


− 0.842

Potassium (K)




− 0.117


− 0.006

− 0.567

− 0.046


Total nitrogen (TN)






Persistence of the intermediate phase between open grassland and closed-canopy forest, known as wood-pasture, is management-dependent. Tracking structural changes within wood-pasture and linking different stages of the wood-pasture cycle to the most influential environmental variables can provide a useful tool for wood-pasture restoration and conservation measures. Our results demonstrated the complexity of the relationship between herbivory, soil resources and vegetation within different stages of wood pasture.

The available phosphorus was the only soil variable distinctively higher on GR plots (Table 5). Such high values of soil available phosphorus in pasture plots suit the establishment of perennial grasses and weeds and are commonly associated with more frequent flooding (Mathews et al. 1993; Mubyana et al. 2003) and higher grazing pressure (Baron et al. 2001; Wei et al. 2011).
Table 5

Mean values and standard deviation (SD) of soil variables found statistically significant to the overall species variation within different vegetation types. Krčedinska ada, Serbia

Plot variables

Woody cover (WC) ± SD

Grassland (GR) ± SD

Transitional vegetation with thorny shrub (TR) ± SD

Available phosphorus (mg 100 g−1)

10.11 (± 5.17)

24.25 (± 7.72)

8.07 (± 3.62)

Carbon to nitrogen ratio (%)

23.97 (± 2.30)

21.36 (± 6.76)

25.67 (± 1.84)

Calcium (g kg−1)

38.67 (± 1.43)

34.23 (± 3.35)

37.89 (± 2.15)

Magnesium (g kg−1)

15.60 (± 0.29)

13.32 (± 1.84)

15.47 (± 0.80)

Potassium (g kg−1)

4.06 (± 0.60)

3.51 (± 1.81)

4.05 (± 1.72)


8.13 (± 0.08)

8.07 (± 0.23)

8.19 (± 0.07)

Total nitrogen (%)

0.19 (± 0.02)

0.18 (± 0.07)

0.15 (± 0.00)

The distribution of WC plots appears to be influenced by soil nutrients namely total nitrogen, potassium, calcium and magnesium whose higher levels support the requirements of woody vegetation. High soil nitrogen content in WC plots is in accordance with the findings of Tilman (1987) who showed that there is a tendency for herbaceous plants to dominate in low nitrogen soils unlike woody plants which increase their biomass in nitrogen rich soils. Higher content of soil nutrients in WC plots may be related to water retention from the flooding events and soil texture. These plots, being further away from the water line, are located on loamy soil which tends to accumulate more nutrients in comparison to sandy soils (Yajing et al. 2017). Soil nitrogen and phosphorus exhibited different spatial pattern (Figs. 4, 5) which has already been reported for flooded ecosystems by Heinscha et al. (2004) and Xiaolong et al. (2014).

In our study, pH values in the soil varied in the range of 7.74–8.29, having the highest values in TR plots and lowest in GR plots (Table 5). Although variations in soil pH may be a plant induced change as demonstrated in Isermann (2005), the converse effect has been reported as well. Soil pH can affect plant species distribution through altering the macro and micronutrient availability in soils (Sims 1986; Robert et al. 2007) and is linked to vegetation patterning in wetland sites (Li et al. 2008). An experimental study of Tilman and Olff (1991) conducted on grasslands demonstrated that soil pH had significant effect on plant species composition. In their study on vegetation succession, Prach et al. (2007) tested 13 abiotic site factors including soil variables, to determine the most important traits which influence vegetation succession in European anthropogenic habitats and showed that pH and macroclimate were the most important factors. Our results suggest that variation in soil pH, although not very large, may have influenced the distribution of vegetation types and plant species composition in Krčedinska ada.

High C/N ratio values in TR plots (Table 5) indicate that this factor could be an important determinant in the establishment of transitional shrubby plants. Variations in soil C/N ratio, due to anthropogenic N deposition, may induce changes in plant species composition as shown by Bowman et al. (2006). However our results differ from those described by Christine and McCarthy (2005) who found that C/N ratio is a predominant factor in the establishment of herbaceous layer.

Although higher terrain with less flooding is geomorphologically predisposed to the development of woody species within floodplains, interpretations of satellite and aerial images from 1969 showed that both TR and WC plots belonged to wood-pasture type II, exhibiting a greater degree of openness compared to the current, woody encroachment scenario (Fig. 3). Since we noted the absence of coppice-like features in TR plots from the 1969 images, the drastic decline in the number of grazers over the years is a likely explanation for currently greater abundance of thorny shrub species. At high herbivore densities thorny shrubs are regularly eaten by grazers (Smit et al. 2007) especially young plants when thorns are not developed yet (Good et al. 1990). When grazing gets significantly reduced thorny shrubs start to overgrow in the pasture with a tendency to reach closed forest stage (Olff et al. 1999; Gillet 2008; Manning et al. 2009). Long periods without grazing have been shown to result in abundant tree regeneration and cessation of the wood-pasture cycle, with large expanses of herbaceous land cover being closed by trees and shrubs that act as regeneration centers for woody vegetation establishment (Manning et al. 2009; Van Uytvanck and Verheyen 2014). We can assume that when the island had one order of magnitude more animals the plant community was very different to today having higher populations of grassland species and less woodland species.

Although Krčdinska ada has been subjected to grazing over a century, flooding duration as a function of elevation is expected to demonstrate its dominance in determining plant species composition. The high explanatory power of grazing intensity variable in terms of species composition reflects a long history of grazing and possibly inadequate grazing regime.

Studies conducted on floodplains demonstrate the highest nutrient content in high flood zones (Wassen et al. 2002; Tsheboeng et al. 2014). In Krčedinska ada, the highest nutrient content was found in WC plots distanced from the river coast i.e. low flood zone. Since different grazing distribution patterns and preference toward certain locations can result in soil nutrient gradients (Borrow 1967; West et al. 1989; Mathews et al. 1994; Bardgett and Wardle 2003), we may conclude that grazing largely affected the spatial heterogeneity of soil nutrients in Krčedinska ada. Our results indicate that herbivory and soil nutrient heterogeneity act together in structuring and altering plant species composition causing shifts between wood pasture stages. If grazing intensity continues to decline we expect the role of topography to become more influential in soil nutrient status and plant species distribution with woody vegetation predominating on the island.

Main conclusions

Our study demonstrates that the distribution of plant communities on Krčedinska ada is closely related to small-scale abiotic drivers whose variability might be linked to shifts in vegetation physiognomy (grassland—shrubland—woody cover). Since grazing is the most important instrument for conserving wood pastures, conservation management should target specific site conditions while considering the livestock number and the heterogeneity of soil properties. Additional field studies are necessary to improve the assessment of vegetation responses to grazing decline and soil resources. Future research should also examine the influence of vegetation patterns on the ecosystem services this area provides.



This study was funded by the Ministry of Education, Science and Technological Development of the Republic of Serbia within the research project 43002: Biosensing Technologies and Global System for Long-Term Research and Integrated Management of Ecosystems and the project “Geotransformation of Vojvodina in the function of regional development (project number 114-451-2080/2016) of the Provincial Secretariat for Higher Education and Scientific Research, Vojvodina Autonomous Province.


  1. Adams WM, Perrow M (1999) Scientific and institutional constraints on the restoration of European floodplains. In: Marriott SB, Alexander J (ed) Floodplains: Interdisciplinary Approaches. Geological Society, London, Special Publications 163: 89-97Google Scholar
  2. Adler PB, Raff DA, Lauenroth WK (2001) The effect of grazing on the spatial heterogeneity of vegetation. Oecologia 128:465–479CrossRefPubMedGoogle Scholar
  3. AOAC (2000) Official methods of analysis, 17th edn. Gaithersburg, MarylandGoogle Scholar
  4. Augustine DJ, McNaughton SJ (1998) Ungulate effects on the functional species composition of plant communities: herbivore selectivity and plant tolerance. J Wildl Manag 62:1165–1183CrossRefGoogle Scholar
  5. Bardgett RD, Wardle DA (2003) Herbivoremediated linkages between aboveground and belowground communities. Ecology 84:2258–2268CrossRefGoogle Scholar
  6. Baron VS, Dick AC, Mapfumo E, Malhi SS, Naeth MA, Chanasyuk DS (2001) Grazing impacts on soil nitrogen and phosphorus under Parkland pastures. J Range Manag 54:704–710CrossRefGoogle Scholar
  7. Beltman B, Willems JH, Güsewell S (2007) Flood events overrule fertiliser effects on biomass production and species richness in riverine grasslands. J Veg Sci 18:625–634CrossRefGoogle Scholar
  8. Bergmeier E, Roellig M (2014) Diversity, threats and conservation of European wood pastures. In: Hartel T, Plieninger T (eds) European wood pastures in transition. Routledge, Oxon, pp 19–38Google Scholar
  9. Bergmeier E, Petermann J, Schröder E (2010) Geobotanical survey of wood-pasture habitats in Europe: diversity, threats and conservation. Biodivers Conserv 19:2995–3014CrossRefGoogle Scholar
  10. Borrow NJ (1967) Some aspects on the effects of grazing on the nutrition of pastures. J Aust Inst Agric Sci 33:254–262Google Scholar
  11. Bowman WD, Gartner JR, Holland K, Wiedermann M (2006) Nitrogen critical loads for alpine vegetation and terrestrial ecosystem response: are we there yet? Ecol Appl 16:1183–1193CrossRefPubMedGoogle Scholar
  12. Braakhekke WG, Hooftman DAP (1999) The resource balance hypothesis of plant species diversity in grassland. J Veg Sci 10:187–200CrossRefGoogle Scholar
  13. Braun-Blanquet J (1964) Pflanzensoziologie: Grundzuge der Vegetationskunde. Springer, WienCrossRefGoogle Scholar
  14. Briske DD, Fuhlendorf SD, Smeins FE (2003) Vegetation dynamics on rangelands: a critique of the current paradigms. J Appl Ecol 40:601–614CrossRefGoogle Scholar
  15. Carboni M, Dengler J, Mantilla-Contreras J, Venn S, Torok P (2015) Conservation value, management and restoration of Europe’s semi-natural open landscapes. Hacquetia 14:5–17CrossRefGoogle Scholar
  16. Christine JS, McCarthy BC (2005) Relationship of understory diversity to soil nitrogen, topographic variation and stand age in an eastern Oak forest, USA. For Ecol Manag 217:229–243CrossRefGoogle Scholar
  17. Cízková H, Kvet J, Comín F, Laiho R, Pokorný J, Pithart D (2013) Actual state of European wetlands and their possible future in the context of global climate change. Aquat Sci 75:3–26CrossRefGoogle Scholar
  18. Dangol DR, Shivakoti GP (2001) Species composition and dominance of plant communities in western Chitwan, Nepal. Nepal J Sci Technol 3:69–78Google Scholar
  19. Falk S (2014) Wood-pastures as reservoirs for invertebrates. In: Hartel T, Plieninger T (eds) European wood-pastures in transition: a social-ecological approach. Earthscan, London, pp 132–148Google Scholar
  20. Gillet F (2008) Modelling vegetation dynamics in heterogeneous pasture-woodland landscapes. Ecol Model 217:1–18CrossRefGoogle Scholar
  21. Good JEG, Bryant R, Carlill P (1990) Distribution, longevity and survival of upland hawthorn (Crataegus monogyna) scrub in North Wales in relation to sheep grazing. J Appl Ecol 27:272–283CrossRefGoogle Scholar
  22. Grime JP (1974) Vegetation classification by reference to strategies. Nature 250:26–31CrossRefGoogle Scholar
  23. Gulan LJ, Milenković B, Stajić J, Vucković B, Krstić D, Zeremski T, Ninkov J (2013) Correlation between radioactivity levels and heavy metal content in the soils of the North Kosovska Mitrovica environment. Environ Sci Proc Impacts 15:1735–1742CrossRefGoogle Scholar
  24. Güsewell S (2004) N:P ratios in terrestrial plants:variation and functional significance. New Phytol 164:243–266CrossRefGoogle Scholar
  25. Güsewell S, Koerselman W, Verhoeven JTA (2003) Biomass N:P ratios as indicators of nutrient limitation for plant populations in wetlands. Ecol Appl 13:372–384CrossRefGoogle Scholar
  26. Hampicke U (2006) Efficient conservation in Europe’s agricultural countryside—rationale, methods and policy reorientation. Outlook Agric 35:97–105CrossRefGoogle Scholar
  27. Hartel T, Plieninger T, Varga A (2014) Wood pastures in Europe. In: Hartel T, Plieninger T (eds) European wood pastures in transition. Routledge, Oxon, pp 61–64Google Scholar
  28. Heinscha FA, Heilman JL, McInnesb BK, Cobosc DR, Zubererb DA, Roelked DL (2004) Carbon dioxide exchange in a high marsh on the Texas Gulf Coast: effects of freshwater availability. Agric For Meteorol 125:159–172CrossRefGoogle Scholar
  29. Hendricks HH, Bond WJ, Midgley JJ, Novellie PA (2005) Plant species richness and composition along livestock grazing intensity gradients in a Namaqualand (South Africa) protected area. Plant Ecol 176:19–33CrossRefGoogle Scholar
  30. Isermann M (2005) Soil pH and species diversity in coastal dunes. Plant Ecol 178:111–120CrossRefGoogle Scholar
  31. Joyce C (2001) The sensitivity of a species-rich flood-meadow plant community to fertilizer nitrogen: the Lužnice river floodplain, Czech Republic. Plant Ecol 155:47–60CrossRefGoogle Scholar
  32. Kaim D, Kozak J, Kolecka N, Ziółkowska E, Ostafin K, Ostapowicz K, Gimmi U, Munteanu C, Radeloff VC (2016) Broad scale forest cover reconstruction from historical topographic maps. Appl Geogr 67:39–48CrossRefGoogle Scholar
  33. Kalusova V, Le Duc MG, Gilbert JC, Lawson CS, Gowing DJG, Marrs RH (2009) Determining the important environmental variables controlling plant species community composition in mesotrophic grasslands in Great Britain. Appl Veg Sci 12:459–471CrossRefGoogle Scholar
  34. Lepš J, Šmilauer P (2003) Multivariate analysis of ecological data using CANOCO. Cambridge University Press, CambridgeGoogle Scholar
  35. Li W, Xiao-Jing L, Khan MA, Gul B (2008) Relationship between soil characteristics and halophytic vegetation in coastal region of North China. Pak J Bot 40:1081–1090Google Scholar
  36. Lorenzo M, Michelea S, Sebastian K, Johannes I, Angelod P (2007) Effects of local factors on plant species richness and composition of Alpine meadows. Agric Ecosyst Environ 119:281–288CrossRefGoogle Scholar
  37. Manning AD, Fischer J, Lindenmayer DB (2009) Scattered trees are keystone structures—implications for conservation. Biol Conserv 132:311–321CrossRefGoogle Scholar
  38. Mathews BW, Sollenberger LE, Nair VD, Staples CR (1993) Impact of grazing management on soil nitrogen, phosphorus, potassium, and sulfur distribution. J Environ Qual 23:1006–1013CrossRefGoogle Scholar
  39. Mathews BW, Sollenberger LE, Nkedi-Kizza P, Gaston LA, Hornsby HD (1994) Soil sampling procedures for monitoring potassium distribution in grazed pastures. Agron J 86:121–126CrossRefGoogle Scholar
  40. McNaughton SJ, Banyikwa FF, McNaughton MM (1997) Promotion of the cycling of diet-enhancing nutrients by African grazers. Science 278:1798–1800CrossRefPubMedGoogle Scholar
  41. Milchunas DG, Vandever MW (2014) Grazing effects on plant community succession of early- and mid-seral seeded grassland compared to shortgrass steppe. J Veg Sci 25:22–35CrossRefGoogle Scholar
  42. Milchunas DG, Sala OE, Lauenroth WK (1988) A generalized model of the effects of grazing by large herbivores on grassland community structure. Am Nat 132:87–106CrossRefGoogle Scholar
  43. Miller KR (1996) Conserving biodiversity in managed landscapes. In: Szaro RC, Johnston DW (eds) Biodiversity in managed landscapes: theory and practice. Oxford University Press, Oxford, pp 425–441Google Scholar
  44. Montgomery DC, Peck EA, Vining GG (2001) Introduction to linear regression analysis, 3rd edn. Wiley, New YorkGoogle Scholar
  45. Mubyana T, Krah M, Totolo O, Bonyongo M (2003) Influence of seasonal flooding on soil total nitrogen, organic phosphorus and microbial populations in the Okavango Delta, Botswana. J Arid Environ 54:359–369CrossRefGoogle Scholar
  46. Neter J, Kutner MH, Nachtsheim CJ, Wasserman W (1996) Applied linear regression models, 3rd edn. Times Mirror Higher Education Group, ChicagoGoogle Scholar
  47. Ogden RW, Thoms MC, Levings PL (2002) Nutrient limitation of plant growth on the floodplain of the Narran River Australia: growth experiments and a pilot survey. Hydrobiologia 489:277–285CrossRefGoogle Scholar
  48. Olde Venterink HGM, van der Vliet RE, Wassen MJ (2001) Nutrient limitation along a productivity gradient in wet meadows. Plant Soil 234:171–179CrossRefGoogle Scholar
  49. Olff H, Vera FWM, Bokdam J, Bakker ES, Gleichman JM, de Maeyer K et al (1999) Shifting mosaics in grazed woodland driven by the alteration of plant facilitation and competition. Plant Biol 1:127–137CrossRefGoogle Scholar
  50. Otypková Z, Chytry M (2006) Effects of plot size on the ordination of vegetation samples. J Veg Sci 17:465–472CrossRefGoogle Scholar
  51. Peterken GF, Hughes FMR (1998) Limitations and opportunities for restoring floodplain forest in Britain. In: Bailey RG, Jose PV, Sherwood BR (eds) United Kingdom floodplains. Westbury, Otley, pp 428–436Google Scholar
  52. Plieninger T, Hochtl F, Spek T (2006) Traditional land-use and nature conservation in European rural landscapes. Environ Sci Policy 9:317–321CrossRefGoogle Scholar
  53. Plieninger T, Hartel T, Martin-Lopez B, Beaufoy G, Bergmeier E, Kirby K, Jesus Montero M, Moreno G, Oteros-Rozas E, Van Uytvanck J (2015) Wood-pastures of Europe: geographic coverage, social–ecological values, conservation management, and policy implications. Biol Conserv 190:70–79CrossRefGoogle Scholar
  54. Pokorný P, Chytrý M, Juřičková L, Sádlo J, Novák J, Ložek V (2015) Mid-Holocene bottleneck for central European dry grasslands: did steppe survive the forest optimum in northern Bohemia, Czech Republic? Holocene 25:716–726CrossRefGoogle Scholar
  55. Prach K, Pyšek P, Jarošík V (2007) Climate and pH as determinants of vegetation succession in Central-European human-made habitats. J Veg Sci 18:701–710CrossRefGoogle Scholar
  56. Rejmánek M, Richardson DM, Higgins SI, Pitcairn MJ, Grotkopp E (2005) Ecology of invasive plants: state of the art. In: Mooney HA, Mack RN, McNeely JA, Neville LE, Schei PJ, Waage JK (eds) Invasive alien species: a new synthesis. Island Press, Washington DC, pp 104–161Google Scholar
  57. Ren GH, Shang ZH, Long RJ, Hou Y, Deng B (2013) The relationship of vegetation and soil differentiation during the formation of black-soil-type degraded meadows in the headwater of the Qinghai-Tibetan Plateau, China. Environ Earth Sci 69:235–245CrossRefGoogle Scholar
  58. Robert J, Dalling JW, Harms KE, Yavitt JB, Stallard RF, Mirabello M, Hubbell SP, Valencia R, Navarrete H, Vallejo M, Foster RB (2007) Soil nutrients influence spatial distributions of tropical tree species. P Natl Acad Sci Biol 104:864–869CrossRefGoogle Scholar
  59. Roos L, Kuijpers L, Peters Roy CJH, Lamers Leon PM, Roelofs Jan GM (2009) Nutrient limitation along eutrophic rivers? Roles of N, P and K input in a species-rich floodplain hay meadow. Appl Veg Sci 12:362–379CrossRefGoogle Scholar
  60. Semmartin M, Di Bella C, de Salamone IG (2010) Grazing-induced changes in plant species composition affect plant and soil properties of grassland mesocosms. Plant Soil 328:471–481CrossRefGoogle Scholar
  61. Sims JT (1986) Soil pH effects on the distribution and plant availability of manganese, copper, and zinc. Soil Sci Soc Am J 50:367–373CrossRefGoogle Scholar
  62. Smit C, Vandenberghe C, Den Ouden J, Mueller-Schaerer H (2007) Nurse plants, tree saplings and grazing pressure: changes in facilitation along a biotic environmental gradient. Oecologia 152:265–273CrossRefPubMedGoogle Scholar
  63. Sollins P (1998) Factors influencing species composition in tropical lowland rain forest: does soil matter? Ecology 79:23–30CrossRefGoogle Scholar
  64. Stewart RE, Kantrud HA (1972) Vegetation of prairie potholes, North Dakota in relation to quality of water and other environmental factors. US Geological Survey Professional Paper 585-D, Washington DC, USAGoogle Scholar
  65. Teague WR, Dowhower SL, Baker SA, Haile N, DeLaune PB, Conover DM (2011) Grazing management impacts on vegetation, soil biota and soil chemical, physical and hydrological properties in tall grass prairie. Agric Ecosyst Environ 141:310–322CrossRefGoogle Scholar
  66. Ter Braak CJF, Šmilauer P (2002) CANOCO reference manual and CanoDraw for Windows user’s guide: software for canonical community ordination (version 4.5). Microcomputer Power, Ithaca, New York, USAGoogle Scholar
  67. Tilman D (1987) Secondary succession and the pattern of plant dominance along experimental nitrogen gradients. Ecol Monogr 57:189–214CrossRefGoogle Scholar
  68. Tilman D, Olff H (1991) An experimental study of the effects of pH and nitrogen on grassland vegetation. Acta Oecol 12:427–441Google Scholar
  69. Tsheboeng G, Bonyongo M, Murray-Hudson M (2014) Flood variation and soil nutrient content in floodplain vegetation communities in the Okavango Delta. S Afr J Sci 110:3–4CrossRefGoogle Scholar
  70. Tzialla CE, Veresoglou DS, Papakosta D, Mamolos AP (2005) Changes in soil characteristics and plant species composition along a moisture gradient in a Mediterranean pasture. J Environ Manag 80:90–98CrossRefGoogle Scholar
  71. Van Oorschot M, Hayes C, Van Strien I (1998) The influence of soil desiccation on plant production, nutrient uptake and plant nutrient availability in two French floodplain grasslands. River Res Appl 14:313–327CrossRefGoogle Scholar
  72. Van Uytvanck J, Verheyen K (2014) Grazing as a tool for wood pasture restoration and management. In: Hartel T, Plieninger T (eds) European wood pastures in transition. Routledge, Oxon, pp 149–167Google Scholar
  73. Veen GF, Blair J, Smith M, Collins S (2008) Influence of grazing and fire frequency on small scale plant community structure and resource variability in native tallgrass prairie. Oikos 117:859–866CrossRefGoogle Scholar
  74. Vera WM (2000) Grazing ecology and forest history. CAB International, WallingfordCrossRefGoogle Scholar
  75. Vos W, Meekes H (1999) Trends in European cultural landscape development: perspectives for a sustainable future. Landsc Urban Plan 46:3–14CrossRefGoogle Scholar
  76. Wassen MJ, Peeters WHM, Olde Venterink H (2002) Patterns in vegetation, hydrology, and nutrient availability in an undisturbed river floodplain in Poland. Plant Ecol 165:27–43CrossRefGoogle Scholar
  77. Wei L, Hai-Zhou H, Zhi-Nan Z, Gao-Lin W (2011) Effects of grazing on the soil properties and C and N storage in relation to biomass allocation in an alpine meadow. J Soil Sci Plant Nut 11:27–39CrossRefGoogle Scholar
  78. West CP, Mallarino AP, Wedin WF, Marx DB (1989) Spatial variability of soil chemical properties in grazed pastures. Soil Sci Soc Am J 53:784–789CrossRefGoogle Scholar
  79. Westoby M (1989) Selective forces exerted by vertebrate herbivores on plants. Trends Ecol Evol 4:115–117CrossRefPubMedGoogle Scholar
  80. Xiaolong W, Jingyi H, Ligang X, Rongrong W, Yuwei C (2014) Soil characteristics in relation to vegetation communities in the Wetlands of Poyang Lake, China. Wetlands 34:829CrossRefGoogle Scholar
  81. Yajing W, Fangqing C, Miao Z, Shaohua C, Xiangqian T, Meng L, Zhihua H (2017) The effects of the reverse seasonal flooding on soil texture within the hydro-fluctuation belt in the Three Gorges reservoir. J Soils Sediments, China. Google Scholar
  82. Yoshihara Y, Ohkuro T, Buuveibaatar B, Jamsran U, Takeuchi K (2010) Spatial pattern of grazing affects influence of herbivores on spatial heterogeneity of plants and soils. Oecologia 162:427–434CrossRefPubMedGoogle Scholar

Copyright information

© The Ecological Society of Japan 2017

Authors and Affiliations

  • Dušanka Krašić
    • 1
  • Elli Groner
    • 2
  • Minucsér Mészáros
    • 3
  • Tijana Nikolić
    • 1
  • Dimitrije Radišić
    • 4
  • Stanko Milić
    • 5
  • Marko Kebert
    • 6
  • Dubravka Milić
    • 4
  • Ante Vujić
    • 4
  • Zoran Galić
    • 6
  1. 1.BioSense Institute, University of Novi SadNovi SadSerbia
  2. 2.Dead Sea and Arava Science CenterJerusalemIsrael
  3. 3.Department of Geography Tourism and Hotel Management, Faculty of SciencesUniversity of Novi SadNovi SadSerbia
  4. 4.Department of Biology and Ecology, Faculty of SciencesUniversity of Novi SadNovi SadSerbia
  5. 5.Laboratory for Soil and AgroecologyInstitute of Field and Vegetable CropsNovi SadSerbia
  6. 6.Institute of Lowland Forestry and EnvironmentNovi SadSerbia

Personalised recommendations