Agroforestry Systems

, Volume 92, Issue 4, pp 909–919 | Cite as

Effects of grazing exclusion and environmental conditions on the soil seed bank of a Mediterranean grazed oak wood pasture

  • Antonello Franca
  • Giovanni Antonio Re
  • Federico Sanna


Large seed banks in the soils of Mediterranean wood pastures can allow the composition of the understorey vegetation to adapt to changing conditions such as under-grazing, grazing exclusion and climate change. This three year study investigated the effect of grazing exclusion on the transient and persistent seed banks of 23 areas of a Mediterranean wood pasture of Quercus suber L., Q. ilex L. and Q. pubescens Willd. A canonical correspondence analysis was used to determine the effect of topo-climatic (elevation, aspect, slope, rainfall, temperature, tree coverage), soil (pH, soil texture, and soil nitrogen, phosphorus, lime and organic carbon content) and biodiversity (Shannon index, species richness index, and Pastoral Value) variables on the soil seed bank under grazed and ungrazed conditions. The size of the persistent seed bank increased with rainfall, grazing, and the available phosphorus content of the soil. Specific site by site grazing regimes could increase the abundance of legumes in the soil seed bank and the species richness and diversity of the understorey vegetation. These results can help guide the conservation management of this silvopastoral area.


Silvopastoral system Grazed gaps Grassland communities Topography Soil properties Biodiversity 


The structure and ecosystem functions of Mediterranean grazed wood pastures are a result of factors including annual and seasonal variations in climate, the impact of grazing animals, disturbances due to natural events, and the consequences of human use of these landscapes during thousands of years (Rees et al. 2001; Madrigal et al. 2008). Original forests and woodlands of Quercus suber L., Q. ilex L. and Q. pubescens Willd., widespread in the Mediterranean Basin, have reduced in area due to the combination of human management and climate change (Blondel 2010). The resilience of many of the remaining oak woodlands is achieved through silvopastoral systems where trees, shrubs and herbaceous species are integrated with livestock (Moreno et al. 2014). Grazed oak woodlands are widespread in Sardinia and they have an important socio-economic role providing rural employment and ecosystem services (Seddaiu et al. 2013; Rossetti et al. 2015). The livestock (e.g., sheep, goats and beef cattle, and to a lesser extent pigs) typically graze for the complete year using different feed resources (grasses, shrubs and trees) sometimes using common land. Grazing pressure may vary depending on the pasture availability of the different grazed patches. Low tree density areas or open areas (vegetation gaps, cleared areas) can be intensively grazed by sheep and beef cattle due to the high productivity of pasture. By contrast, in many pure stands of cork oak, grazing is excluded and consequently shrubs must be cleared mechanically when encroachment occurs. Recent studies have highlighted the benefits of grazing in reducing wildfires by decreasing fuel biomass and reducing the probability of fire propagation, rate of spread and fireline intensity (Franca et al. 2012). In some cases, silvopastoral areas on common land can be under pressure of overgrazing due to year-round grazing, large herd/flock sizes, the agronomic and physical marginality of the areas, and incorrect management. Moreover, recurring wild fires, climate change, human settlement and abandonment of agricultural land can increase the risk of environmental degradation.

An understanding of plant community dynamics is fundamental for the design and implementation of successful programs for silvopastoral management under Mediterranean conditions, where the seasonal and interannual variations of rainfall and, consequently, the soil water content affect the ratio of annual to perennial plants and the community assemblage. Few studies have focused on these issues (Carneiro et al. 2008; Ramos et al. 2010). Topography is also considered to be important in determining the spatial pattern of Mediterranean vegetation (Davis and Goetz 1990; Pueyo and Alados, 2007; Gusmeroli et al. 2013; Franca et al. 2016). A strong influence of microtopography on the vegetation patterns has been observed through its effects on the spatial organization of surface water flows and infiltration (Vico and Porporato 2009; McGrath et al. 2012; Spasojevic et al. 2013; Harman et al. 2014) and by affecting seed dispersal and distribution (Boudell et al. 2002).

The seed bank ecology of Mediterranean grazed woodlands (which reflects both the history of the vegetation and its composition) creates substantial flexibility for potential changes in the dynamics and persistence of the plant community. In changing conditions, such as undergrazing or grazing exclusion, plasticity of pasture composition species is achieved by a large “persistent seed bank” which can remain dormant in the soil for several years (Leck et al. 1989). Managers can use an understanding of the seed bank ecology to improve the sustainability of the pastures and the persistent fraction of the seed bank can have important implications for the conservation of plant communities and species (Auld and Denham 2006).

In this paper, we studied the impact of several environmental variables (related to topography and climate, plant biodiversity and soil properties), and grazing management on the seed bank composition in the grazed gaps of a Mediterranean silvopastoral system. The two hypotheses tested were that: (i) micro-site environmental variations and ii) grazing exclusion can affect the size and composition of the transient and the persistent seed bank.

Materials and methods

Site description

The trial was carried out during three years (2009–2011) in the Forest of Monte Pisanu (Central Sardinia, Italy), identified as a Site of Community Importance (SIC). The climate of this area is bi-seasonal, with rain in the autumn–winter period and an almost completely dry spring and summer season, which can be aggravated by intense winds. Annual mean temperature is 13 °C (± 5.8 °C SD) and annual mean precipitation is 908 mm (± 40 mm SD). The frequency of snowfall in this mountainous area of central Sardinia does not exceed 5–10 days per year. Frost can occur in the autumn, winter and spring. The geology comprises volcanic soils and clastic deposits which can be permeable and fractured near the land surface, but less permeable at depth. The soil is derived from crystalline schists, which have high levels of potassium, moderate contents of phosphoric anhydride, and low contents of calcium and clay. The landscape vegetation at Monte Pisanu is dominated by downy oak (Quercus pubescens L.) and holm oak (Quercus ilex L.) that extend up to more than 1000 m above sea level. The downy oaks often have a bushy habit and they occur alongside small mountain meadows. There are no trees around the summit of Monte Rasu (1259 m), the highest peak in the area, where the vegetation is restricted to shrub-like perennials such as thyme, helichrysum and rockrose. Areas of oak and maple trees, with some holly, coexist on the east side of Monte Rasu which has been less affected by fires and logging.

Experiment description

Since 2009, 23 grazed sites were identified. Within each site a representative and homogeneous sample area of 50 × 50 m was identified for vegetation measurements. Inside each sample area, a fenced area of 4 × 4 m was identified to allow the comparison of the seed bank size and composition under ungrazed and grazed conditions. This dimension of the fenced area was necessary in order to limit the visual impact of the fences within an area that forms part of a landscape protection programme. The centre of each fenced area was considered as the geographical point to be used for the determination of all topoclimatic variables.

Field surveys to characterize the plant communities and the Pastoral Values of the species were carried out between May 2009 and May 2011. Vegetation data were collected in late spring each year using a point intercept method (Daget-Poissonet 1969) on 2 × 50 m line intersect transects in the 50 × 50 m grazed sample areas (total counts per transect = 200), and on 2 × 4 m line intersect transects into the fenced ungrazed areas (total counts per transect = 40). The recorded list of species was used for the calculation of Pastoral Value (Cavallero et al. 2007) and to estimate the ecological value using the Shannon Index (Shannon, 1948) and species richness index. The evaluation of the Pastoral Value (PV) was based on the specific indexes (Si) of forage plants reported by Roggero et al. (2002) and the original specific indexes assessed during the field surveys of the present research. The value of PV was determined with the formula: PV = 0.2 × (Σ SCi ×  Si), where SCi is the specific contribution (%) of each single species and Si is the specific index, the latter ranging from 0 to 5 (Delpech 1960; Roggero et al. 2002; Cavallero et al. 2007).

Soil core sampling

Given that the experimental area has never been ploughed or tilled, it is supposed that the depth of 5 cm is an approximate limit to which most seeds are distributed and are active in determining the vegetation structure (Traba et al. 2004). In the autumn of each year, in the first days after the first rains, four random soil cores were collected outside (grazed condition, Gr) and two inside (ungrazed conditions, Ung), using a methodology which sought to enable the collection of intact soil cores with minimal disturbance. After removing the vegetation and vegetation debris at the soil surface, a PVC cylinder of 8 cm diameter was inserted into the soil to a depth of 5 cm, was undermined and then placed on a sheet of filter paper. The sheet of filter paper was then wrapped around the sides of the cylinder by securing it with a robust rubber band, so that the cylinder was closed below and open above.

Seed bank recordings

The collected soil samples were then maintained under rainfed conditions and the transient seed bank (TS) was determined by counting the seedlings that emerged from germinating seeds within each intact soil core for a period between one and two months. Once the transient seed bank was removed, the soil samples were then prepared to force germination of the remaining seeds. In each core, the mineral component (e.g. small stones, gravel, and sand) and the visible plant dry residues (e.g. leaves, stems and roots) or other organic residues were separated (Ball and Miller 1989). The mineral component of soil samples were then washed in distilled water, filtered through two sieves with a porosity of 500 and 200 µm and then placed in petri dishes on filter paper, transferred in a germination chamber and submitted to three successive treatments of dormancy breakdown, according to the methodology introduced by Perez et al. (1998):
  1. (a)

    21 days of germination test, diurnal cycle of 8/16 h of darkness/light, and a temperature regime of 20/30 °C.

  2. (b)

    21 days of dormancy assessment test, 3–5 °C.

  3. (c)

    35 days of chemical breakdown test, 600 mg/kg-solution of Gibberellic Acid, diurnal cycle of 8/16 h darkness/light and a temperature regime of 15/35 °C.


Seedlings from germinated seeds during each of these phases were counted, identified and removed. Among the three tests, in order to assess the end of the dormancy breakdown effect, soil samples were wet on filter paper in Petri dishes with distilled water for two weeks in the germination chamber at 18 °C. The number of seeds germinated after the application of a), b) and c) treatments was considered as an estimate of the persistent seed bank (PS). Seed bank components were grouped according to the main families composing the herbaceous vegetation, namely grasses (G), legumes (L), compositae (C) and other (O) species.

Site environmental variables

The environmental characteristics of each site were defined in terms of topographic, climate, tree coverage, and soil variables. The topographic variables of elevation, aspect and slope were extracted from a 10 m resolution Digital Elevation Model of the experimental areas. Mean annual rainfall and mean temperatures (over 30 years) for each site, were obtained from ARPAS, the Regional Office for Environmental Protection (SardegnaARPA 2017). Tree coverage was derived from Ikonos high-resolution (1 m) imageries by calculation of NDVI (Normalized Differential Vegetation Index). A specific NDVI threshold was established to highlight broad-leaf forests from the typical background of the summer season, when the Ikonos images were collected (Goetz et al. 2003). At the centre of each site, soil samples were collected (soil layer 0–40 cm). Soil samples were air-dried and the analysis were made on the < 2 mm fraction after sieving. The pH was determined in 1:2.5 water suspension by potentiometric method using a pH meter. Total N content, available P (P2O5), organic matter and organic carbon were determined by Kjeldahl, Olsen and Walkley–Black method respectively.

Statistical analysis

The effect of topoclimatic (including tree cover), soil and biodiversity variables on the seed bank size and composition was studied performing multivariate analyses with the package CANOCO ver. 4.5 (ter Braak 1989). We have built three data sets of environmental variables (topoclimatic, soil and biodiversity, n = 23) and each one of them was compared with three sets of species variables relative to (i) transient and persistent seed bank size, (ii) transient seed bank composition and (iii) persistent seed bank composition.

A detrended correspondence analysis (DCA) was run a priori to examine the length of the gradients in order to determine the more appropriate type of canonical analysis i.e., either a linear (length < 4) or a unimodal model (length > 4) (Lepš and Šmilauer 2003). Gradient lengths were always lower than 2 for the first axis in the dataset for each of the nine analyses, indicating that the linear method was more appropriate. Consequently, a redundancy analysis (RDA) on both matrices of all environmental variables and seed bank variables was performed. A direct analysis was used to test the null hypothesis that the seed bank composition was unrelated to measured environmental variables. A Monte Carlo test with data randomization (499 unrestricted permutations) was performed to assess the significance of RDA for each annual data set. Lastly a generalised linear model (GLM) was created, using SAS software version 8.2, with the aim of identifying those factors significantly affecting the transient and persistent seed banks, under both grazed and ungrazed conditions.


Relationships between grazing regime and seed bank traits

The seed bank size was affected by the grazing regime. In the ungrazed sample areas, the average number of seeds m−2 was halved compared to that of the grazed surface, both for the transient and the persistent seed bank. On average, the number of transient seeds in the grazed and ungrazed areas were 42,795 and 27,036 m−2, respectively. The number of persistent seeds were 30,272 m−2 and 14,367 m−2 respectively for grazed and ungrazed areas.

In the transient seed banks, the overall composition was dominated by grass seeds representing on average of 63% and 70% in grazed and ungrazed conditions respectively. Other seed types were more abundant in the persistent seed bank. Legume seeds had twice the abundance in the persistent seed bank (12.5%) compared to that (6%) in the transient seed bank. Overall legumes, compositae and other species together comprised 64% of the persistent seed bank composition. The abundance of grasses in the transient seed bank (66.5%) was 46.5% higher than the abundance of 20% in the persistent seed bank.

Relationships between microsite environmental variables and seed bank size

Topoclimatic variables explained 38% of the total variance in the size of the transient seed bank size. The first axis explained the 26.7% of the variance, while the second, third and fourth axes explained 8.9, 1.3 and 1.0%, respectively. There was a significant positive impact of rainfall (P ≤ 0.018) on the size of ungrazed transient and persistent seed bank (Fig. 1a). The site’s soil traits explained about 45% of the variance of seed bank size (Fig. 1b), with significant effect of organic carbon (P ≤ 0.02) and P2O5 (P ≤ 0.046) content. The transient and persistent seed bank size in the ungrazed areas was positively correlated with the organic carbon content, and the transient and persistent seed bank under grazed conditions were positively correlated with P2O5 content. Lastly biodiversity variables could explain 18% of the total variance of seed bank size (Fig. 1c). None of the individual selected biodiversity variables showed a significant correlation with seed bank size, both under grazed and ungrazed conditions. However the average Pastoral Value was significantly higher under ungrazed conditions, and the average Shannon index and Richness Index were significantly higher under grazing (Table 1).
Fig. 1

Biplot from redundancy analysis between the transient (TS) and persistent (PS) seed bank sizes and a the topoclimatic variables elevation (Elev), aspect (Aspect), slope (Slope), average annual rainfall (Rainfall), average annual temperature (Temp) and tree coverage (Tree_cov); b the soil variables pH (pH), nitrogen (Ntot), available phosphorus (P2O5), organic carbon (OC) and organic matter (OM) content, soil traits (lime, sand, clay, loam) and c the biodiversity variables Shannon Index (SH), Species Richness Index (RI) and Pastoral Value (PV). TS_Ung, transient seed bank under ungrazed conditions. TS_Gr transient seed bank under grazed conditions. PS_Ung, persistent seed bank under ungrazed conditions. PS_Gr, persistent seed bank under grazed conditions

Table 1

Results of Fisher’s test: least significant differences (LSD) between Shannon Index (SH), Species richness Index (RI) and Pastoral value (PV) determined for the 23 sites under grazed and ungrazed conditions

Biodiversity index

Grazing conditions




St. dev.


St. dev.

Shannon index (SH)


± 0.31


± 0.63

Species richness index (RI)


± 3.45


± 3.54

Pastoral value (PV)


± 5.52


± 7.75

Different letters indicate a significant difference for P ≤ 0.05. St. dev. Standard deviation

Relationships between microsite environmental variables and seed bank composition

About 45% of the variance of transient seed bank composition can be explained by the site’s topoclimatic traits (Fig. 2a). A significant effect of elevation (P ≤ 0.028) and aspect (P ≤ 0.048) was observed. In particular, the transient seed bank of legumes was negatively correlated with elevation and south facing slopes were detrimental to grasses. Soil variables could explain 32% of the total variance of transient seed bank composition. In particular the sand fraction had a significant effect on the transient seed bank composition (P ≤ 0.0029), especially the abundance of grasses (Fig. 2b).
Fig. 2

Biplot from redundancy analysis between the transient (TS) seed bank composition and a the topoclimatic variables elevation (Elev), aspect (Aspect), slope (Slope), average annual rainfall (Rainfall), average annual temperature (Temp) and tree coverage (Tree_cov); b the soil variables pH (pH), nitrogen (Ntot), available phosphorus (P2O5), organic carbon (OC) and organic matter (OM) content, soil traits (lime, sand, clay, loam); and c the biodiversity variables Shannon Index (SH), Species Richness Index (RI) and Pastoral Value (PV). TSG_Ung, transient seed bank of grasses under ungrazed conditions. TSG_Gr, transient seed bank of grasses under grazed conditions. TSL_Ung, transient seed bank of legumes under ungrazed conditions. TSL_Gr, transient seed bank of legumes under grazed conditions. TSC_Ung, transient seed bank of Compositeae under ungrazed conditions. TSC_Gr, transient seed bank of Compositeae under grazed conditions. TSO_Ung, transient seed bank of Other spp. under ungrazed conditions. TSO_Gr, transient seed bank of other spp. under grazed conditions

Figure 3a shows the effect of site’s topoclimatic traits on the composition of the persistent seed bank under grazed and ungrazed conditions. Topoclimatic variables could explain 29% of the variance, but no variable had a significant individual effect on the persistent seed bank composition. Nevertheless, NW aspect (compass direction between 315 and 360°, data not shown) positively influenced the persistent seed bank of legumes (P ≤ 0.106). The biodiversity variables could explain 42% of the variance in the persistent seed bank composition, with the first and second axis explaining 22.7% and 15.4% respectively. Available P content (P2O5, P ≤ 0.05) had a significant positive effect on the abundance of legumes in the persistent seed bank (Fig. 3b).
Fig. 3

Biplot from redundancy analysis between the persistent (PS) seed bank composition and a the topoclimatic variables elevation (Elev), aspect (Aspect), slope (Slope), average annual rainfall (Rainfall), average annual temperature (Temp) and tree coverage (Tree_cov); b the soil variables pH (pH), nitrogen (Ntot), available phosphorus (P2O5), organic carbon (OC) and organic matter (OM) content, soil traits (lime, sand, clay, loam); and c the biodiversity variables Shannon Index (SH), Species Richness Index (RI) and Pastoral Value (PV). PSG_Ung, persistent seed bank of grasses under ungrazed conditions. PSG_Gr, persistent seed bank of grasses under grazed conditions. PSL_Ung, persistent seed bank of legumes under ungrazed conditions. PSL_Gr, persistent seed bank of legumes under grazed conditions. PSC_Ung, persistent seed bank of Compositeae under ungrazed conditions. PSC_Gr, persistent seed bank of Compositeae under grazed conditions. PSO_Ung, persistent seed bank of Other spp. under ungrazed conditions. PSO_Gr, persistent seed bank of other spp. under grazed conditions

Under grazed conditions, a significant positive correlation was observed between Pastoral Value and the abundance of legume and compositae transient seeds (Fig. 2c). By contrast, low Pastoral Values were correlated with a high abundance of “other” species in the transient seed bank. Within the persistent seed bank, a high Pastoral Value was also positively correlated with a high abundance of legumes seeds under ungrazed and grazed conditions (P ≤ 0.01 and 0.03, respectively) (Fig. 3c). Also, high values of Richness Index were significantly (P ≤ 0.024) and positively correlated with the abundance of “other” and compositae species in both grazing conditions.

Considering the effect of all the environmental factors, the model could explain 66% and 57% of the variance in seed bank size for the transient and persistent seed bank respectively. The proportion of sand in the soil, the elevation, the slope, and the proportion of clay in the soil had a positive significant effect on the transient seed bank (Table 2). The size of the permanent seed bank was significantly negatively affected by the clay percentage in the soil. Grazing regime did not affect significantly the size of both two seed bank components.
Table 2

Significance (P) of all the environmental factors for the two dependent variables transient and persistent seed banks, under both grazed and ungrazed conditions


Transient seed bank

Persistent seed bank
















Tree coverage









Shannon index



Richness index



Pastoral value






N tot






Organic carbon






Grazed vs. ungrazed



Levels of P > 0.05 are not reported and indicated as not significant (n.s.)


Effect of tree cover

In Mediterranean oak grazed woodlands, light limitation precludes seedling establishment of most grassland species in the understorey and facilitates the competitiveness of some pasture species (Marañon and Bartolome 1993). By contrast, high light levels in the gaps between trees can result in greater forage availability than under the tree canopy, increasing the feed resource for grazing animals (Moreno et al. 2014). In this study, the tree coverage did not significantly affect the size of the transient and persistent seed banks. Across the 23 grazed sites, the mean level of tree coverage was 32% and in only six sites did the tree coverage exceed 50%. It is proposed that these levels of tree density were insufficient to create a significant effect on the seed dissemination of pasture species or a measurable indirect effect on the other factors.

Effect of environmental variables on seed bank size and composition

The results of the Redundancy Analysis (RDA) allow us to partially confirm the hypothesis that micro-site environmental (microtopography and microclimate, soil traits and biodiversity indexes) variations affect the seed bank size and composition. The transient seed bank size was greatest at sites with high rainfall, with a particularly positive effect on the abundance of grasses. This result is in accordance with Ooi (2012), who stated that rainfall is the most important climate variable facilitating seed germination and emergence from the seed bank. The germinative response of grass seed to the increase of soil moisture after rainfall is rapid. By contrast the seeds of many legume species have water-impermeable coats inducing a form of physical dormancy (hardseededness) (Baskin et al. 2000; Sulas et al. 2000; D’hondt et al. 2010), that can maintain the persistence of the seeds in this stage over many years. The positive response of grass seed to high levels of rainfall, as found on Monte Pisanu, can explain the predominance of grasses in the transient seed banks at the studied sites. On the contrary, a negative relationship between rainfall and the transient seed bank of legumes would suggest an adaptive response of legumes to dry conditions. These contrasting responses of grasses and legumes can be useful to understand how plant populations will potentially respond to inter-annual variations in weather and longer-term climate change.

Combining the effect of topoclimatic and soil variables, legumes seeds were favoured at low elevation sites with low rainfall and high soil levels of available P. In particular, legume seeds dominated the persistent seed bank of sites whose soils were rich in available P. This extends previous studies of the effect of the soil chemical environment on seed bank composition and density that only considered the transient seed bank (López-Mariño et al. 2009). Arévalo and Chinea (2009) found that P and pH were significant predictors of the number of legume seeds in the transient seed bank of pastures on Tenerife in the Canary Islands. Our results also support the conclusions of Iannucci (2014), who stated that P addition favoured leguminous species and in particular the development of nitrogen-fixing bacterial nodules and thereby root development. The use of phosphorus fertilizers is a well-known technique for improving the productivity, seasonal distribution and floristic composition of Mediterranean pastures and particularly the abundance of legumes. This study indicates that in Mediterranean oak woodlands, soils rich in available phosphorus also favour the constitution of a legume-rich persistent seed bank and, as stated by Kigel (1995), act as a long-term refuge for annual plant populations in the highly variable environments typical of the Mediterranean region.

Effect of grazing exclusion on seed bank size and composition

This study shows a positive effect of grazing on the seed bank size. This contrasts with Kinloch and Friedel (2005) and Solomon et al. (2006) who reported negative effects of grazing on the soil seed bank. Sternberg et al. (2003) argue that grazing might negatively affect the seed bank of annual species when grazing animals hinder seed production. Van Langevelde et al. (2016) also report that large soil seed banks are found in ungrazed semi-arid rangelands and the size of the soil seed bank declines with increasing herbivore density. On the other hand, many other studies emphasized the role of grazing in increasing seed banks of forbs and annual species, and in decreasing the seed bank contribution of perennial grasses (Kinucan and Smeins, 1992; O’Connor and Pickett, 1992). Annual species are an important component of the wood pastures on Monte Pisanu and their persistence is linked with the development of a large soil seed bank. In the area, sheep and cattle grazing is performed all year, including the grazing of dry pasture in summer with an estimated 650 Livestock Units (LU) year−1 over the total area of 2200 ha. These totals hide differences in the grazing management of dairy sheep and cattle for beef. The grazing by sheep is typically on a rotational basis and intermediate to heavy, whilst the grazing by cattle tends to be continuous and less intensive. Dairy sheep typically graze where the tree canopy is less dense or open (gaps, cleared areas), while cattle will graze even under more dense tree canopies. As a consequence, cattle will typically graze a larger part of the territory than the sheep who tend to focus on the most productive areas. It is possible that in this study, the impact of grazing on the seed bank traits is mainly a result of sheep rather than cattle grazing. Grazing by dairy sheep is most intensive between September and early November before lambing, and less intensive during late spring and early summer months when the dairy sheep are dry and their feed requirement is less. O’Connor and Pickett (1992) argue that heavy grazing during late spring and early summer could cause the removal of reproductive structures and result in smaller seed banks which could limit the recovery and persistence of palatable annual species. Instead, in our study, grazing pressure is low during spring months and the subsequent flowering, maturation and dispersion could enable the development of a large seed bank, even under subsequent heavy grazing. This strategy may have important practical implications for the sustainable management of Mediterranean grazed forests. The application of the grazing calendars aimed at minimizing the impact on seed production and distribution, can be an important management tool to ensure a large persistent seed bank, with particular regard to the presence of hard seeded legumes. The production of dormant seeds is an escape strategy that allows an adaptive response of annuals to the unpredictability and the variability of the Mediterranean environments, during periods of limited resource supply (Perevolotsky and Seligman 1998). As concluded by van Langenvelde et al. (2016), a correct intensity of grazing and the availability of seeds in the soil are essential conditions for achieving the production potential of grasslands (Russi et al. 1992; Meissner and Facelli 1999) and for the recovery of degraded semi-arid rangelands (Westoby et al. 1989; Smith et al. 2007; Leck et al. 1989), with positive effects also on the biological diversity.

Grazing also had a significantly positive effect on Shannon and species richness indices. Studies on Mediterranean grazing systems by Landsberg et al. (2002) and Brooks et al. (2006) have already demonstrated that grazing can promote the abundance of annuals and forbs. Perevolotsky (2005) also concluded that grazing in a Mediterranean woodland ecosystem may help to maintain the genetic diversity within plant populations, particularly enabling small herbaceous species, such as geophytes, to compete for light against tall plants. As a consequence, grazing not only increases species richness and diversity but, moreover, may generate significant changes in species composition. Hanke et al. (2014) suggested that species diversity alone might not adequately reflect the shifts in vegetation structure that occur in response to increased grazing intensity. Whilst it may be useful as a taxonomic approach, there is also value in a functional approach relating indicators to ecosystem services. For example, it is interesting to observe the opposite effects of grazing on canonical biodiversity indices (Shannon index and Species Richness index) compared to that on the Pastoral Value. The Pastoral Value is an indicator that synthetizes information on the contribution of species to the plant community with functional traits such as productivity, palatability, digestibility, and the absence of lignified tissue (Roggero et al. 2002). It increased significantly under ungrazed conditions (Table 1), but neither the transient nor persistence seed bank size had a significant impact on Pastoral Value under both ungrazed and grazed conditions. The highest values of Pastoral Value in ungrazed conditions might be due to the presence of perennial grasses (such as Lolium perenne L. and Dactylis glomerata L.) that, according to Alrababah et al. (2007), are grazing-sensitive species and can accumulate in fenced areas. Moreover, Fig. 3c highlights that legume and grasses seeds abundance in the persistent seed bank may explain the high Pastoral Value both in grazed and ungrazed conditions.


The sustainable management of Mediterranean oak wood pastures can benefit from an improved understanding of the complex negative and positive interactions between the abiotic (e.g., topography, climate and soil traits) and biotic (e.g., oak trees, shrubs, herbaceous plants, and grazing animals) components of the ecosystem. This paper focused on the wood pastures of Monte Pisanu demonstrates that the resilience of the understorey of grazed oak wood pastures expressed by the size of the persistent seed bank in the soil, increases with rainfall, grazing, and available phosphorus in the soil. There was not a significant effect of tree cover. The positive effect of grazing on the seed bank size could be a result of the highest intensity grazing occurring in the autumn, rather than the summer and spring. The abundance of legumes in the soil seed bank and the overall quality of the pasture could be increased through specific site-by-site grazing regimes and phosphorus fertilization. Grazing also increased the species richness and diversity of the understorey vegetation, but it decreased the Pastoral Value of the understorey. Understanding such relationships can help managers maintain the integrity of Mediterranean wood pastures in the context of climate change and risks of land abandonment.



The AGFORWARD project (Grant Agreement No. 613520) is co-funded by the European Commission, Directorate General for Research & Innovation, within the 7th Framework Programme of RTD, Theme 2 - Biotechnologies, Agriculture & Food. We thank Giovanni Piras (Forestas Regional Agency) for providing topographic data, Bachisio Arca (CNR IBIMET) for the elaboration of tree coverage data and Giovanna Seddaiu (NRD UNISS) for the support to the statistical analysis. Also, we are grateful to Maddalena Sassu, Salvatore Nieddu, Piero Saba, Daniele Dettori and Daniele Nieddu (CNR ISPAAM) for their technical support.

Supplementary material

10457_2018_203_MOESM1_ESM.docx (16 kb)
Supplementary material 1 (DOCX 15 kb)
10457_2018_203_MOESM2_ESM.xlsx (25 kb)
Supplementary material 2 (XLSX 24 kb)


  1. Alrababah MA, Alhamad MA, Suwaileh A, Al-Gharaibeh M (2007) Biodiversity of semi-arid Mediterranean grasslands: impact of grazing and afforestation. Appl Veg Sci 10(2):257–264CrossRefGoogle Scholar
  2. Arévalo JR, Chinea E (2009) Pastures seedbank composition in relation to soil nutrient content in areas under goat grazing management (Tenerife). J Food Agric Environ 7(3&4):710–716Google Scholar
  3. Auld TD, Denham AJ (2006) How much seed remains in the soil after a fire? Plant Ecol 187(1):15–24CrossRefGoogle Scholar
  4. Ball DA, Miller SD (1989) A comparison of techniques for estimation of arable soil seedbanks and their relationship to weed flora. Weed Res 29(5):365–373CrossRefGoogle Scholar
  5. Baskin JM, Baskin CC, Li X (2000) Taxonomy, anatomy and evolution of physical dormancy in seeds. Plant Species Biology 15:139–152CrossRefGoogle Scholar
  6. Blondel J (2010) The Mediterranean region: biological diversity in space and time. Oxford University Press, New yorkGoogle Scholar
  7. Boudell JA, Link SO, Johansen JR (2002) Effect of soil microtopography on seed bank distribution in the shrub-steppe. West North Am Natural 62(1):14–24Google Scholar
  8. Brooks TM, Mittermeier RA, da Fonseca GA et al (2006) Global biodiversity conservation priorities. Science 313(5783):58–61CrossRefPubMedGoogle Scholar
  9. Carneiro JP, Simoes N, Macas ID, Tavares-de-Sousa M (2008) Pasture improvement in montado extensive farming system. In: Porqueddu C, Tavares-de-Sousa MM (eds) Sustainable Mediterranean grasslands and their multi-functions. options méditerranéennes, serie A: 79, centre international de hautes etudes agronomiques méditerranéennes (CIHEAM), Zaragoza, pp 193–197Google Scholar
  10. Cavallero A, Aceto P, Gorlier A, Lombardi G, Lonati M, Martinasso B, Tagliatori C (2007) I tipi pastorali delle alpi piemontesi. Alberto Perdisa Editore, BolognaGoogle Scholar
  11. Daget Ph, Poissonet T (1969) Analyse phytologique des praisies. INRA, Montpellier Document 48:66Google Scholar
  12. Davis FW, Goetz S (1990) Modeling vegetation pattern using digital terrain data. Landscape Ecol 4(1):69–80CrossRefGoogle Scholar
  13. Delpech R (1960) Critères de jugement de la valeur agronomique des praires. Fourrages 4:83–98Google Scholar
  14. Dhondt B, Brys R, Hoffmann M (2010) The incidence, field performance and heritability of non-dormant seeds in white clover (Trifolium repens L.). Seed Sci Res 20(3):169–177CrossRefGoogle Scholar
  15. Franca A, Sanna F, Nieddu S, Re GA et al (2012) Effects of grazing on the traits of a potential fire in a sardinian wooded pasture. In: Acar Z, Lòpez-Francos A, Porqueddu C (eds) New approaches for grassaland research in a context of climate and socioeconomic changes. Zaragoza: CIHEAM/Ondokuz Mayis University 541: 307–311Google Scholar
  16. Franca A, Caredda S, Sanna F, Fava F, Seddaiu G (2016) Early plant community dynamics following overseeding for the rehabilitation of a Mediterranean silvopastoral system. Grassl Sci 2(62):81–91CrossRefGoogle Scholar
  17. Goetz SJ, Wright RK, Smith AJ, Zinecker E, Schaub E (2003) IKONOS imagery for resource management: tree cover, impervious surfaces, and riparian buffer analyses in the mid-Atlantic region. Remote Sens Environ 88(2003):195–208CrossRefGoogle Scholar
  18. Gusmeroli F, Della Marianna G, Fava F, Monteiro A, Bocchi S, Parolo G (2013) Effects of ecological, landscape and management factors on plant species composition, biodiversity and forage value in alpine meadows. Grass Forage Sci 68:437–447CrossRefGoogle Scholar
  19. Hanke W, Böhner J, Dreber N, Jürgens N, Schmiedel U, Wesuls D, Dengler J (2014) The impact of livestock grazing on plant diversity: an analysis across dryland ecosystems and scales in southern Africa. Ecol Appl 24(5):1188–1203CrossRefPubMedGoogle Scholar
  20. Harman CJ, Lohse KA, Troch PA, Sivapalan M (2014) Spatial patterns of vegetation, soils, and microtopography from terrestrial laser scanning on two semiarid hillslopes of contrasting lithology. J Geophys Res Biogeosci 119:163–180CrossRefGoogle Scholar
  21. Iannucci A (2014) Soil seed-bank germination patterns in natural pastures under different mineral fertilizer treatments. Span J Agric Res 12(4):1018–1028CrossRefGoogle Scholar
  22. Kigel J (1995) Seed germination in arid and semiarid regions. Seed dev germination 1995:645–699Google Scholar
  23. Kinloch JE, Friedel MH (2005) Soil seed reserves in arid grazing lands of central Australia. part 1: seed bank and vegetation dynamics. J Arid Environ 60(1):133–161CrossRefGoogle Scholar
  24. Kinucan RJ, Smeins FE (1992) Soil seed bank of semi-arid Texas grassland under three long-term (36 years) grazing regimes. Ame Middle Nat 128:11–21CrossRefGoogle Scholar
  25. Landsberg J, James CD, Maconochie J, Nicholls AO, Stol J, Tynan R (2002) Scale-related effects of grazing on native plant communities in an arid rangeland region of South Australia. J Appl Ecol 39(3):427–444CrossRefGoogle Scholar
  26. Leck MA (1989). Wetland seed banks. Ecology of soil seed banks, p 283–305Google Scholar
  27. Lepš J, Šmilauer P (2003) Multivariate Analysis of Ecological Data using CANOCO. Cambridge University Press, Cambridge, pp 1–269Google Scholar
  28. López-Mariño A, Luis-Calabuig Fillat E, Bermúdez FF (2009) Floristic composition of established vegetation and the soil seed bank in pasture communities under different traditional management regimes. Agric, Ecosyst Environ 78:273–282CrossRefGoogle Scholar
  29. Madrigal J, García-Rodriguez JA, Roberto J, Puerto A, Fernández-Santos B (2008) Exploring the influence of shrubs on herbaceous communities in a Mediterranean climatic context of two spatial scales. Plant Ecol 195:225–234CrossRefGoogle Scholar
  30. Marañon T, Bartolome JW (1993) Reciprocal transplants of herbaceous communities between Quercus agrifolia woodland and adjacent grassland. J Ecol 81(4):673CrossRefGoogle Scholar
  31. McGrath GS, Paik K, Hinz C (2012) Microtopography alters self-organized vegetation patterns in water-limited ecosystems. J Geophys Res 117:G3CrossRefGoogle Scholar
  32. Meissner RA, Facelli JM (1999) Effects of sheep exclusion on the soil seed bank and annual vegetation in chenopod shrublands of South Australia. J Arid Environ 42(2):117–128CrossRefGoogle Scholar
  33. Moreno G, Franca A, Pinto-Correia T, Godinho S (2014) Multifunctionality and dynamics of silvopastoral systems. Options Méditerranéennes, Series A 109:421–436Google Scholar
  34. O’Connor TG, Pickett GA (1992) The influence of grazing on seed production and seed bank of some African savanna grassland. J Appl Ecol 29:247–260CrossRefGoogle Scholar
  35. Ooi KJ (2012) Seed bank persistence and climate change. Seed Science Research 22:S53–S60CrossRefGoogle Scholar
  36. Perevolotsky A (2005) Integrating landscape ecology in the conservation of Mediterranean ecosystems: the Israeli experience. Israel J Plant Sci 53(3–4):203–213CrossRefGoogle Scholar
  37. Perevolotsky A, Seligman NAG (1998) Role of grazing in Mediterranean rangeland ecosystems. Bioscience 48(12):1007–1017CrossRefGoogle Scholar
  38. Perez CJ, Waller SS, Moser LE, Stubbendieck JL, Steuter AA (1998) Seedbank characteristics of a Nebraska sandhills praire. J Range Manag 51:55–62CrossRefGoogle Scholar
  39. Pueyo Y, Alados CL (2007) Effects of fragmentation, abiotic factors and land use on vegetation recovery in a semi-arid Mediterranean area. Basic Appl Ecol 8:158–170CrossRefGoogle Scholar
  40. Ramos ME, Robles AB, Gonzalez-Rebollar JL (2010) Ley-farming and seed dispersal by sheep: two methods for improving fallow pastures in semiarid Mediterranean environments? Agric Ecosyst Environ 137:124–132CrossRefGoogle Scholar
  41. Rees M, Condit R, Crawley M, Pacala S, Tilman D (2001) Long-term studies of vegetation dynamics. Science 293:650–655CrossRefPubMedGoogle Scholar
  42. Roggero PP, Bagella S, Farina R (2002) Un archivio dati di Indici Specifici per la valutazione integrata del valore pastorale. Rivista di Agronomia 36(2):149–156Google Scholar
  43. Rossetti I, Bagella S, Cappai C, Caria MC, Lai R et al (2015) Isolated cork oak trees affect soil properties and biodiversity in a Mediterranean wooded grassland. Agr Ecosyst Environ 202:203–216CrossRefGoogle Scholar
  44. Russi L, Cocks PS, Roberts EH (1992) Seed bank dynamics in a Mediterranean grassland. J Appl Ecol 1992:763–771CrossRefGoogle Scholar
  45. SardegnaARPA (2017), Accessed 16 Feb 2017
  46. Seddaiu G, Porcu G, Ledda L, Roggero PP, Agnelli A, Corti G (2013) Soil organic matter content and composition as influenced by soil management in a semi-arid Mediterranean agro-silvo-pastoral system. Agr Ecosyst Environ 167:1–11CrossRefGoogle Scholar
  47. Shannon CE (1948) A mathematical theory of communication. The Bell Syst Tech J 27:379–423CrossRefGoogle Scholar
  48. Smith DMS, McKeon GM, Watson IW, Henry BK, Stone GS, Hall WB, Howden SM (2007) Learning from episodes of degradation and recovery in variable Australian rangelands. Proc Natl Acad Sci 104(52):20690–20695CrossRefGoogle Scholar
  49. Solomon TB, Snyman HA, Smit GN (2006) Soil seed bank characteristics in relation to land use systems and distance from water in a semi-arid rangeland of southern Ethiopia. S Afr J Bot 72(2):263–271CrossRefGoogle Scholar
  50. Spasojevic MJ, Bowman WD, Humphries HC, Seastedt TR, Suding KN (2013) Changes in alpine vegetation over 21 years: are patterns across a heterogeneous landscape consistent with predictions? Ecosphere 4(9):1–18CrossRefGoogle Scholar
  51. Sternberg M, Gutman M, Perevolotsky A, Kigel J (2003) Effects of grazing on soil seed bank dynamics: an approach with functional groups. J Veg Sci 14(3):375–386CrossRefGoogle Scholar
  52. Sulas L, Franca A (2000) Caredda S (2000) Persistence and regeneration mechanisms in forage legumes. Cahiers Options Méditerranéennes 45:331–342Google Scholar
  53. ter Braak CJF (1989) CANOCO—an extension of DECORANA to analyze species–environment relationships. Hydrobiologia 184:169–170CrossRefGoogle Scholar
  54. Traba J, Azcárate FM, Peco B (2004) From what depth do seeds emerge? A soil seed bank experiment with Mediterranean grassland species. Seed Sci Res 14:297–303CrossRefGoogle Scholar
  55. van Langevelde F, Tessema ZK, de Boer WF, Prins HH (2016) Soil seed bank dynamics under the influence of grazing as alternative explanation for herbaceous vegetation transitions in semi-arid rangelands. Ecol Model 337:253–261CrossRefGoogle Scholar
  56. Vico G, Porporato A (2009) Probabilistic description of topographic slope and aspect. J Geophys Res. CrossRefGoogle Scholar
  57. Westoby M, Walker B, Noy-Meir I (1989) Opportunistic management for rangelands not at equilibrium. J Range Manag 42(4):266–274CrossRefGoogle Scholar

Copyright information

© Springer Science+Business Media B.V., part of Springer Nature 2018

Authors and Affiliations

  1. 1.CNR – ISPAAM, Italian National Research Council, Istituto per il Sistema Produzione Animale in Ambiente Mediterraneo, Institute for Animal Production System in Mediterranean EnvironmentSassariItaly

Personalised recommendations