1 Introduction

In Europe populations of the wild boar (Sus scrofa) substantially increased in the last decades. Indeed, the wild boar is included in the list of the 100 most invasive species at the global level in terms of impact on biological diversity and/or human activities (Lowe et al. 2000).

The wild boar can live in a wide range of habitats as it feeds opportunistically on plant and animal species. Consequently, the wild boar affects several taxa both directly (through predation) and indirectly (through the alteration of different ecosystems parameters). The wild boar activity that has the strongest effects on ecosystems is rooting, as this may alter soil and vegetation in areas up to hundreds of hectares (Welander 2000).

In hardwood forests, understorey species represent the great majority of wild boar diet and several authors agree that wild boar negatively affects the understorey composition (Ickes et al. 2005; Siemann et al. 2009) and abundance (Howe and Bratton 1976; Hone 2002; Pinna et al. 2007).

Nevertheless, results of previous research focused on the density of understorey species are contrasting (Massei and Genov 2004), and, in particular, there is no information on which species in particular are strongly affected by wild boar rooting. The aim of our study is to investigate the impact of wild boar rooting activity on understorey vascular plants in two Mediterranean lowland forests in central Italy. We conducted our research in forests of high conservation concern (i.e. Mediterranean lowland forests dominated by the deciduous oaks Quercus cerris and Q. frainetto). Indeed, no data on the conservation status of the subtype Balkanic thermophilous oak woods are reported for Italy (http://www.bd.eionet.europa.eu/article17/chapter4) although these ecosystems are recognised by the Habitat Directive (Habitat 91M0 see Biondi et al. 2012).

2 Materials and methods

2.1 Study area

Our work was carried out in the Castelporziano Estate, which is included amongst the State Reserves since 1999. This area is located 25 km southwest from Rome along the Tyrrhenian coast of Lazio (central Italy), and it covers approximately 5,900 ha at an altitude of 0–85 m a.s.l. The climate is typically Mediterranean (Blasi et al. 1999): during the period 2004–2013 the average annual temperature was 17.1 °C, with minimum and maximum temperatures of, respectively, 9.8 and 20.4 °C; annual rainfall for the considered period was 784 mm. The substratum is mostly made up of pleistocenic and holocenic eolian sands, but is locally characterised by alluvial sediments and pyroclastic materials from Latium volcanic system (Badalì et al. 1999).

The Castelporziano Estate is a site of outstanding importance for the conservation of vascular plant species. Several species of conservation concern (regionally endangered or vulnerable according to the IUCN categories) have a distribution limited to the estate. The area is particularly important for the preservation of mesophilous woods, Mediterranean maquis and forests, as well as for wetlands that once covered extensive areas along the mid-Tyrrhenian coast. Furthermore, the area includes remnants of dunal complexes and temporary ponds of high conservation concern (Capotorti et al. 2013; Celesti-Grapow et al. 2013). Such habitats have now become very fragmented due to urban and agricultural pressure and land reclamations dating from the early twentieth century (Bertacchi and Lombardi 2014; Farris et al. 2013; Ricotta et al. 2001).

We focused on the most widespread forest type, which is dominated by the deciduous oaks Q. cerris L. and Q. frainetto L. (Mespilo germanicae-Quercetum frainetto Biondi, Gigante, Pignattelli et Venanzoni 2001).

In the area that is now occupied by Castelporziano Estate, the occurrence of ungulates (including the wild boar) is recorded since Roman times; in the last century, the wild boar population was relatively stable and increased after the Second World War reaching around 800 individuals in the late twentieth century (Mecella et al. 1999).

It was not possible to identify stands with low-rooting levels in Castelporziano Estate. For this reason, we compared data collected in the State Reserve of Castelporziano with data collected in the Circeo National Park forest (Fig. 1), where areas with different levels of rooting activity (low and high) occurred. The two study sites are very similar for climate, lithology and natural potential vegetation (Mespilo germanicae-Quercetum frainetto). Climatic parameters for the lowland forest in Circeo National Park are close to those of Castelporziano Estate: for the period 2004–2013, average annual temperature was 15.9 °C with minimum and maximum temperatures of, respectively, 9.6 and 22.9 °C; annual rainfall for the considered period is on average 808 mm (source of data: Pontinia meteorological station which is located 6 km from the sampled forest; source: http://www.arsial.it/portalearsial/agrometeo).

Fig. 1
figure 1

Study areas location in Italy and sampled stands

Full size image

In the Circeo forest, the density of wild boar has fluctuated considerably since the 1930s with a steady increase in the last 50 years, in 2004 the number of individuals was estimated to be between 600 and 700.

2.2 Data collection

Rooting intensity, forest structure and understorey species composition were sampled between May and July 2010. Sampling was conducted within forests co-dominated by the deciduous oaks Q. cerris and Q. frainetto. (site A: Figurone—Castelporziano Estate; site B: lowland forest in Circeo National Park).

To assess rooting intensity, we sampled 21 transects (100 m long and 5 m wide) in the site A, and 12 equal transects in the site B. The intensity of rooting was evaluated through counts of single rooting signs, defined as an approx. 15-cm-wide sign of rooting activity on the ground. Wider signs were evaluated as multiples of a single sign (i.e. a 30-cm-wide sign = 2 rooting signs), as described in Fagiani et al. (2014). We found a high-rooting intensity for the site A, while in the site B we found areas with contrasting rooting levels (high and low).

We conducted the understorey vegetation sampling in six square plots (400 m2 each) in site A, and in 12 identical plots (six for each level of rooting) in site B; these were randomly selected from a dataset of 24 total plots (Burrascano et al. 2014).

The occurrence and the percentage cover value of each vascular plant taxon were recorded. To assess species cover in more detail, each sampling unit was divided into four subunits (10 × 10 m each one). Taxa were identified following the Flora d’Italia (Pignatti 1982) and the Flora Europaea (Tutin et al. 1968–1980, 1993). Nomenclature follows Conti et al. (2005) and Pignatti (1982).

As overstorey structure substantially influences understorey composition and diversity (Burrascano et al. 2011; Sabatini et al. 2014), we investigated forest structural attributes to verify the occurrence of similar conditions across all the sampling units in the two sites. Specifically, overstorey structure was sampled through a variable radius plot, using a wedge prism, centred on the square plots used for the vegetation sampling. The diameter at breast height (DBH) and the vitality class of all the standing living and dead trees included in the variable radius plot were recorded. We also recorded the mid-diameter, length and decay class (Hunter 1990) of every piece of lying deadwood with a diameter >10 cm in each plot. We calculated and included in the analysis seven structural variables that could be used as indicators of structural heterogeneity in mature and over-mature forests (Corona et al. 2010; Višnjić et al. 2013): number of diameter classes; number of standing trees with DBH >40 cm; diameter range; the proportion of standing dead trees; coarse woody debris volume; number of decay classes; and basal area.

2.3 Statistical analysis

We analysed differences in understorey species composition through permutational multivariate analysis of variance (Permanova—Anderson 2001). This analysis is an extension of the traditional analysis of variance that tests the simultaneous response of dependent variables (species) to one or more factors (site and rooting) on the basis of a resemblance matrix. Statistical significance is then assessed through permutations. We used Bray–Curtis similarity coefficient, and we tested the result for significance through 9,999 permutations under a reduced model. As the groups of plots belong to different sites, we included the site as a covariate in the model design. If covariates are included in the analysis, their effect is tested for significance in the first place, then the effect of the factors is analysed excluding the effect of the covariates. To visualise and compare the results yielded by Permanova, we performed a Principal Coordinates Analysis (PCoA). PCoA is a data ordination method that places the sampling units onto Euclidean axes on the basis of a resemblance matrix (the same used for Permanova) using the information provided by the variables (i.e. species abundances). Permanova and PCoA were performed with the software Primer-E (Clarke and Gorley 2006).

To further analyse the differences in species composition, the indicator values of the understorey vascular plant species of the three groups of plots were calculated by means of the Indicator Species Analysis (ISA—Dufrene and Legendre 1997), with 4,999 permutations using PC-Ord 5 (McCune and Mefford 2006). Species with an indicator value higher than 50 % and a p value lower than 0.05 were considered to be indicative of one out of the three groups of plots.

Finally, differences in species density, Shannon–Weaver diversity and evenness (diversity indices), and overstorey structure between the three groups of plots were tested for significance by means of Analysis of Variance (ANOVA) and the Tukey post hoc test in R 2.12.0 (R Core Team 2009).

3 Results

3.1 Species composition and indicator species

The species composition of the investigated plots was significantly different in areas with contrasting rooting levels. The first axis of the PCoA explained 44 % of the variance and highlighted differences related to the site. Nevertheless, part of the variance explained by axis 1 should be related to rooting levels (Fig. 2), since the Permanova returned highly significant differences between rooting levels, using the area as a covariate (Table 1).

Fig. 2
figure 2

Principal coordinates analysis scatterplot along the first two axes. Sampling units belonging to different groups are represented by different symbols

Full size image
Table 1 Permanova results
Full size table

The comparison of overstorey structure yielded non-significant results for most variables. Indeed, the ANOVA returned significant differences only relatively to the density of large living trees and the DBH range (Table 2), with a significantly higher density of large living trees resulting also in higher values of DBH range in site A (Table 3).

Table 2 ANOVA based on forest structural attributes
Full size table
Table 3 Post hoc tests for the forest structural variables that resulted significantly different in the ANOVA
Full size table

These results confirm that the overstorey structure is similar in the three groups of plots, with limited differences between the two study areas, and support our decision not to use forest structure as a further determinant of compositional differences in the Permanova analysis.

The indicator species analysis allowed us to characterise the differences in composition amongst the groups of plots (Table 4). In the low-rooting plots of site B, six species were found to have a significant indicator value with Ruscus aculeatus being the species with the highest indicator value; while in the high-rooting plots of site B, the ISA returned three indicator species (Rubia peregrina, Rosa sempervirens, Melica uniflora). The species that show a significant degree of association to the high-rooting plots of site A are more numerous and include also Conyza albida (=Erigeron sumatrensis), an invasive species from South America (Celesti-Grapow et al. 2009).

Table 4 Species with an indicator value higher than 50 % and a p value lower than 0.05 for each of the three groups of plots
Full size table

3.2 Diversity indices

Despite the marked differences in understorey composition, the contrasting levels of rooting do not determine significantly different levels of species density and of Shannon diversity indices (Table 5). Through the ANOVA, we found marginally significant differences only for the Shannon evenness index; however, such differences were not confirmed by the post hoc tests (Table 6).

Table 5 ANOVA based on understorey diversity indices
Full size table
Table 6 Post hoc tests for the understorey diversity index that resulted significantly different in the ANOVA
Full size table

4 Discussion

4.1 Wild boar determines strong differences in understorey species composition

The sampling areas of this study are characterised by the same vegetation type (Mespilo germanicae-Quercetum frainetto) and have similar overstorey structure; nevertheless, we found marked differences in the understorey composition. Therefore, rooting strongly affected the understorey of deciduous oak woodlands in our sampled areas, determining significant changes in the occurrence and abundance of several species.

The analysis of the species that resulted as potential indicators gives relevant insights on the processes that occur in the understorey in relation to rooting intensity, and on the mechanisms that favour or phase out species with different ecological requirements and adaptations.

The highest indicator value in the low-rooting areas of the Circeo forest (site B) is for R. aculeatus. Based on its life-history traits, this species could be favoured by wild boar rooting as it is by grazing. Indeed, it was demonstrated that R. aculeatus frequently occurs in grazed forests (Onaindia et al. 2004), where it can persist for at least 10 years after the ceasement of grazing disturbance (Debussche et al. 2001). The shoots of this species are less likely to be affected by rooting than those of non-spinescent species. Moreover, R. aculeatus is a clonal species that rapidly sprouts after predation or mechanical damage.

The low-rooting areas are characterised also by the occurrence of Asphodelus ramosus, a Mediterranean species that is typically found in areas frequently disturbed either by grazing or fire. The presence of these two indicator species (R. aculeatus and A. ramosus) suggests that areas with low-rooting levels may have been intensely affected by wild boar feeding activities in the past. These previous activities may have determined major compositional and structural modifications resulting in the dominance of clonal, thorny, and unpalatable species. The current high cover values of these species, and especially of R. aculeatus, are likely to further inhibit feeding activities by wild boar, therefore, the areas where they are particularly abundant display low-rooting values.

The species with a significant indicator value in the Circeo groups of plots, both with low- and high-rooting levels, are highly consistent with the studied communities that are characterised by a combination of Mediterranean climbing species (Lonicera etrusca, Rosa sempervirens, Rubia peregrina), and of perennial tussocks typically occurring in deciduous forests (Luzula forsteri, Melica uniflora). These species are likely to result as indicators due to the fact that they do not occur at all in the Castelporziano plots. Indeed, species such as Piptatherum miliaceum (=Oryzopsis miliacea), Erigeron sumatrensis (=Conyza albida), Juncus effusus and Calamintha nepeta, which are indicators of Castelporziano plots (site A), may be considered not consistent with the vegetation type under study, whereas they are likely to be related to specific characteristics of the Castelporziano estate and to the abundance of wild boars in this area.

The species P. miliaceum was found to be strongly associated to recently urbanised areas under a Mediterranean climate (Dana et al. 2002) as is the area surrounding the Castelporziano estate. The proximity to urban environments may also be the reason of the occurrence of Conyza albida in site A. This species is native of America, but it is currently globally widespread and listed as invasive in Italy (Celesti-Grapow et al. 2009). The occurrence of C. nepeta and J. effusus in Castelporziano (site A) may be related to a higher presence of gaps where high light levels and high temperatures may affect the composition of the understorey, indeed both species have high Ellenberg indicator indices for light and temperature (Pignatti 2005). However, J. effusus may also be substantially favoured by the high density of wild boar in the Castelporziano forest since this species is effectively dispersed by wild boars both through endo- and ecto-zoochory (Schmidt et al. 2004).

The other indicator species of site A point to a great cover of evergreen (Rhamnus alaternus, Phillyrea latifolia) and deciduous (Carpinus orientalis, Cytisus scoparius) shrubs, also in relation to the death and senescence of very large and old trees in this forest (see also Tables 2, 3) together with the scarce recruitment of deciduous oak trees (Fanelli and Tescarollo 2006; Macuz et al. 2006).

Despite the fact that previous studies have demonstrated substantial effects of wild boar on the recruitment of oaks (Gomez and Hodar 2008), our results did not highlight this relationship. This may be a consequence of the fact that, we compared areas characterised by different levels of rooting intensity rather than rooted and unrooted areas and, as acorns represent an important part of wild boar’s diet (Pinna et al. 2007), it is likely that oak recruitment is affected by wild boar also in areas where the intensity of rooting is low. The same explanation applies to Cyclamen repandum and C. hederifolium as both these species are amongst the most actively predated by wild boar; therefore, only experiments based on the total exclusion of rooting disturbance may allow the detection of wild boar effects on them.

4.2 Long-term exclosure experiments are needed to detect variation in species diversity

Our results highlight: (1) the occurrence of significant compositional changes in forest communities in response to wild boar activities and (2) that these changes are derived from present as well as from past disturbance by wild boar. Therefore, our study underscores the need of long-term experimental studies to be conducted through exclosures. We think that Castelporziano Estate represents a unique opportunity to perform such studies in an experimental forest of high biogeographical and conservation concern.