Human-induced landscape changes are among the most important processes resulting in biodiversity loss and the deterioration of ecosystem functioning in terrestrial habitats (Fischer and Lindenmayer, 2007). The amount and spatial arrangement of natural habitat fragments have an important role in determining how organisms are distributed in highly modified landscapes (Fahrig, 2017). The majority of habitat fragmentation studies focus on habitat fragments vs. landscape matrix as a black and white dichotomy with habitat fragments being the only suitable area for the focal organisms (Fardila et al., 2017). The matrix is defined as an extensive land cover that has undergone intense anthropogenic perturbation. Matrix quality is the degree to which natural conditions are altered in the matrix. Matrix quality may encompass variation in the availability of resources that supplement those of the habitat patches (Shoffner et al., 2018; Yekwayo et al., 2016). Connectivity and thereby the dispersal of organisms between habitat fragments depend on matrix quality (Antongiovanni and Metzger, 2005; Watling et al., 2011; Freeman et al., 2018). A high quality matrix, which is structurally similar to the habitat fragment, may allow certain rates of dispersal and mitigate the effect of habitat loss and fragmentation by providing additional resources for species (Fischer and Lindenmayer, 2007). Some species are able to exploit a good quality matrix in a way that their populations may remain stable or even increase in abundance throughout the fragmentation process (Antongiovanni and Metzger, 2005; Pfeifer et al., 2017). However, a low quality matrix strongly contrasts to natural habitats and thereby may impede the movement of species across the landscape. Regular immigration due to higher dispersal rates can help to avoid stochastic extinctions of populations with low abundance in small fragments (Öckinger et al., 2012a), thus matrix quality may exert a stronger effect on small fragments than on large fragments.

The landscape scale parameters affect the dissimilarity between local communities, and this may determine landscape-wide biodiversity (Tscharntke et al., 2012). This may also have a key effect on the variation of the species composition of habitat fragments, i.e. beta-diversity (Legendre, 2008). Several frameworks exist to characterize patterns of species distribution by partitioning beta diversity into its two components (Baselga, 2012; Carvalho et al., 2013; Podani and Schmera, 2011). According to Baselga (2012), the two components of beta diversity are, (1) species turnover, and (2) nestedness. Turnover is related to the replacement of species between communities, whereas nestedness expresses to what extent one community is a subset of another one. Their relative importance may change according to the processes structuring meta-communities (Tonkin et al., 2016). Random extinctions may contribute to nestedness in moderately connected fragments, however, increasing dispersal rates between fragments reduce differences in species composition resulting in the homogenization of the meta-community (Gianuca et al., 2017; Samu et al., 2018). Differences between the species composition of highly isolated fragments are mainly due to species turnover. Stochastic drift in species composition and the colonization of matrix species are the two main drivers of this phenomenon (Logue et al., 2011; Collins et al., 2017).

Forest steppes offer the opportunity to test the effect of matrix quality on the biota of different natural habitat types within habitat fragments. Forest steppes are structurally and compositionally very complex ecosystems. Habitat types create a gradient of plant and presumably invertebrate species richness (Erdős et al., 2020). In general, forest and steppe patches provide different physical conditions (Anenkhonov et al., 2015; Tölgyesi et al., 2018a) resulting in different plant communities that harbor a high species diversity with numerous endemic plant (e.g. Dianthus diutinus Kit. ex Schult., Colchicum arenarium Waldst. & Kit.), rare insect (e.g. Myrmeleotettix antennatus (Fieber, 1853), Hyponephele lupine Costa, 1936) and bird species (e.g. Coracias garrulus L., 1758, Caprimulgus europaeus L., 1758) of community interest in the European Union. Thus, forest steppes are recognized as important biodiversity hotspots (Dengler et al., 2014) and listed in the Habitats Directive (European Union 1992, codes: 91I0, 91N0, 6260).

Nevertheless, forest steppes are also among the most threatened ecosystems in Central Europe due to habitat loss and fragmentation. Most forest steppe mosaics of the region have been transformed into commercial plantations of exotic (mainly Pinus sylvestris L. and P. nigra J.F. Arnold) and native tree species (Populus alba L.) with extremely low habitat heterogeneity (Erdős et al., 2018). The dense canopy of plantations decreases the light availability on the forest floor resulting in lower understory biomass, moist and sheltered microclimate and homogeneous habitat structure (Balandier et al., 2006). Contrarily, native forest steppes are characterized by small-scale habitat heterogeneity and high microhabitat diversity due to the mosaic structure of the two prevailing physiognomic units, the forest and the steppe.

In the present study we focused on plants, spiders and ants for several reasons. Herbaceous vegetation is an important component of terrestrial food webs and provides the primary structure of habitats for ground-dwelling arthropods. Spiders are among the most abundant arthropod predators in all terrestrial ecosystems (Wise, 1995; Nyffeler and Birkhofer, 2017), and their diversity pattern sensitively indicates small differences in local and landscape scale environmental changes (Ziesche and Roth, 2008). Ants are a diverse and locally very abundant group of omnivorous terrestrial insects. Besides interspecific biotic interactions, particularly competition, habitat and landscape parameters are also important in determining their diversity (Gibb, 2011).

We aimed to understand the community composition and species richness patterns of plants, spiders and ants in highly modified forest steppe regions by focusing on the three main habitat types, steppe and forest parts of the forest steppe fragments (hereafter ‘steppes’ and ‘forests’) and the most abundant element of the landscape matrix, pine plantations (hereafter ‘plantations’). More specifically, we tested the following hypotheses: (1) Plantations resemble forests more than steppes in terms of plant, spider and ant species composition. (2) Steppes are more isolated than forests amidst the plantation-dominated landscape matrix, thus species turnover is larger, and nestedness is smaller between steppe than between forest patches of isolated habitat fragments. (3) Increasing matrix quality (i.e. higher share of semi-natural habitats in the landscape) has a stronger positive effect on species richness of small than large fragments.

We recorded 164 plant species in the 54 sampling sites of the 18 landscapes. The most abundant graminoids were F. vaginata, S. borysthenica and Carex liparocarpos Gaudin. Common forbs included Teucrium chamaedrys L., Potentilla arenaria Borkh. ex G.Gaertn., B.Mey. & Scherb. and Euphorbia seguieriana Neck (Appendix S2, S3). We collected 5595 adult spiders of 111 species. The most abundant ones were Pardosa alacris (C.L. Koch, 1833), Alopecosa sulzeri (Pavesi, 1872), and Zelotes apricorum (L. Koch, 1876), (Appendix S2, S3). Finally, we collected 8773 ants belonging to 32 species, with Lasius psammophilus Seifert, 1992, Plagiolepis taurica Santschi, 1920 and Myrmica sabuleti Meinert, 1861 being the most abundant species (Appendix S2, S3).

Community composition of plants, spiders and ants was influenced by habitat type (Fig. 2). We found significant differences in the species composition of the three habitat types according to the perMANOVA analyses, with the most expressed differences between steppe and the other habitat types (Table 1). We identified several significant indicator species of forest (plants:19; spiders: 7, ants: 2 species), steppes (plants:24, spiders: 9, ants: 5 species), and plantations (plants:0; spiders: 6, ants: 4 species) (Appendix S2).

Fig. 2.

Ordination plots of non-metric multidimensional scaling. (a) plants, (b) spiders, (c) ants. Open circles: steppes, grey dots: forests, black dots: plantations. Stress values for all NMDS ordinations were low (vegetation: 0.15; spiders: 0.09; ants: 0.13).

(0.1MB).

The turnover component of beta diversity was higher for steppes than for forests both in the case of incidence-based and abundance-based approaches, i.e. we found higher patchiness of the steppe habitat compared to forests (Fig. 3, Table 2.). Plant species richness was highest in forests followed by steppes and it was the lowest in plantations according to the GLMM models, furthermore, the difference between plantations and steppes was larger in small than in large fragments (Fig. 4a, Appendix S4). Matrix quality had a negative effect in small fragments, however this effect was positive in large fragments (Fig. 4b, Appendix S4). We found higher specialist plants species richness in steppes followed by forests and plantations (Fig. 5a, Appendix S4). We found similar interacting effect of matrix quality and fragment size for plants and specialist plats, matrix quality had a negative effect in small fragments, however this effect was positive in large fragments (Fig. 5b, Appendix S4). We found higher species richness of spiders in plantations and forests than in steppes and matrix quality had a positive effect on them (Fig. 6a, b, Appendix S4). However, specialist spider species richness was the highest in steppes followed by plantations and forests (Fig. 6c, Appendix S4). We found no significant effects on ant species richness, yet we detected more specialist ant species in plantations than in steppes (Fig. 6d, Appendix S4).

Fig. 3.

The turnover component of beta diversity measured as the ratio of turnover component and total beta diversity (a) abundance based approach, (b) incidence based approach. Grey and white bars represent forests and steppes (mean±sd), respectively.

(0.09MB).

Fig. 4.

Effect of habitat type, matrix quality and fragment size on the species richness of plants and specialist plants. Species richness of plants: (a) interaction of fragment size and habitat type (mean±sd), light grey bars: small fragment, dark grey bars: large fragments, (b) interaction of matrix quality and fragment size. Open circles: steppes, grey triangles: forests, black dots.

(0.23MB).

Fig. 5.

Species richness of specialist plants: (a) habitat type (mean±sd), (b) interaction of matrix quality and fragment size. Small fragments: size between 0.24 and 0.75ha; Large fragments: size between 1.1 and 6.0ha. Values for matrix quality index are ranged between 0 and 1. Open circles: steppes, grey triangles: forests, black dots: plantations. Dotted lines represent the 95% confidence intervals.

(0.12MB).

Fig. 6.

Effect of habitat type and matrix quality on the species richness of spiders and ants. Species richness of spiders: (a) habitat type (mean±sd), (b) matrix quality, open circles: steppes, grey triangles: forests, black dots. Species richness of specialist spiders: (c) habitat type (mean±sd). Species richness of specialist ants: (d) habitat type (mean±sd). Values for matrix quality index are ranged between 0 and 1. Open circles: steppes, grey triangles: forests, black dots: plantations. Dotted lines represent the 95% confidence intervals.

(0.22MB).

Species composition of plantations was more similar to that of the forests than to the steppes. Along this line, we also found higher species turnover ratios for steppes than for forests for all taxa indicating that steppes are more isolated than forests. Matrix quality affected both plant and spider species richness of all habitat types; this effect was positive for spiders, however, matrix quality effect was modified by fragment size for plants. It negatively affected plant species richness of small fragments, but this effect was positive for large fragments.

We found a clear separation of the biota of focal habitat types (Hypothesis 1). The plants of the three habitat types had no overlap according to the NMDS ordination. Numerous studies showed significant differences between the vegetation structure and plant species composition of forests and steppes in forest steppe ecosystems (e.g. Bátori et al., 2018; Erdős et al., 2014, 2018). However, there is virtually no information on the role of the main matrix habitat, i.e. plantations, as a potential secondary habitat for forest steppe vegetation. We found that the species richness of plantations was less than half of the species richness of the forests and this was also true for the specialist species richness, presumably due to the thick and allelopathic litter layer of pine needles, and the low water and light availability in the forest floor of the plantations (Selvi et al., 2017). The dominant plant species of these plantations were mainly generalist, disturbance tolerant (e.g. Calamagrostis epigejos and Poa angustifolia) or even invasive species (e.g. Asclepias syriaca) indicating that pine plantations are very low quality secondary habitats for natural vegetation.

Arthropod species composition of the three habitat types was also significantly different. However, plantations and forests overlapped according to the NMDS ordination plots for both spiders and ants. Interestingly, we found higher spider species richness in plantations than in steppes. Steppes formed on sandy soils are characterized by harsh microclimatic conditions with high and fluctuating ground temperature (maximum summer temperature of bare ground can exceed 60°C) and low air humidity (Erdős et al., 2014). The hot and dry conditions favor invertebrate species with certain trait state combinations (e.g. xerotolerant and thermophilous species), and serve as a strong environmental filter (Gallé et al., 2018). A relatively small number of specialist invertebrate species are able to colonize such habitats (Gallé and Torma, 2009). In line with Gallé et al. (2018) and Ingle et al. (2019) the present study also suggests that forest generalist species constitute a large part of the fauna of plantations, and these species are unable to colonize forest steppes.

We found that the turnover component had a high contribution (>0.6) to beta diversity of forests, and even higher (>0.9) for steppes for both incidence-based and abundance-based multiple-site dissimilarity approaches, confirming our Hypothesis (2). Such remarkably high contribution of species replacement was observed for island butterflies (Yong et al., 2012) and spiders (Wu et al., 2017) suggesting a higher level of isolation for steppes than for forests. A low-quality matrix with high habitat-matrix contrast may decrease functional connectivity by preventing movement of organisms between fragments (Reider et al., 2018; Tölgyesi et al., 2018a,b). Community composition of plantations was more similar to forests than to steppes according to the NMDS ordinations, suggesting a higher habitat-matrix contrast and a less permeable landscape for the biota of grasslands than that of the forests. Communities in less permeable landscapes are more exposed to random shifts in community composition due to higher extinction rates and random colonization of species (Watling et al., 2011). Therefore, these isolated communities have more divergent species compositions than communities for which the matrix is more permeable (Watling et al., 2011). Another possible explanation for the high turnover ratio between steppes could be the micro-habitat heterogeneity across the sampled grasslands. Small-scale differences in soil properties are known to affect vegetation in dry grasslands (Maestre and Cortina, 2002), and microhabitat structure affects spider (Gallé et al., 2010; Heneberg and Řezáč, 2014) and ants (Castracani et al., 2010) of arid grasslands. Presumably, low matrix permeability together with high microhabitat diversity resulted in the high turnover component of the beta diversity of steppes.

Matrix quality may have an interacting effect with fragment size, as increasing matrix quality may have a weaker effect on large fragments (Öckinger et al., 2012b). Increasing matrix quality had a positive effect on both the total and specialist plant species richness of all habitat types in case of landscapes with large habitat fragments, however, this effect was negative for landscapes with small fragment size (Hypothesis 3). High quality matrix may exert its positive effect via improved immigration rates of forest steppe specialists to the habitat fragments (Rand et al., 2006 Low quality matrix diminishes dispersal and regular immigration of specialist species into habitat fragments (Driscoll et al., 2013). The low dispersal between fragments prevents the displacement of poor competitors by locally strong competitors and promotes the coexistence of specialist species (Levine and Rees, 2002; Büchi and Vuilleumier, 2014). However, this phenomenon is not pronounced in large fragments where stronger competitors are present in higher numbers. This suggests the type of matrix in which natural habitat fragments are situated can explain a substantial amount of the widely reported variability in biodiversity responses to fragmentation (Hatfield et al., 2020).

We found that increasing matrix quality had a positive effect on the species richness of spiders. Several studies also concluded that increasing matrix quality has a positive effect on the species richness of a habitat fragment, and this effect is the most expressed if the amount of suitable habitats and high quality matrix elements are low (Ruffell et al., 2017; Reider et al., 2018). This emphasizes the positive effect of increasing matrix quality for spiders, which are generally moderately good dispersers, even in high quality landscapes.

According to our findings, ant species richness was unaffected by habitat type, matrix quality and fragment size. Matrix may also reduce fragment size effects, i.e. small and large fragments will harbor a similar number of species. This could be because high landscape heterogeneity results in high resilience and stability of ecological processes (Freeman et al., 2018). This is in line with Dauber et al. (2003), who found that matrix quality explains relatively low variance in ant species richness, suggesting the importance of other factors, such as competition, in shaping ant community structure (Hölldobler and Wilson, 1990). Interference competition has an increasing importance in advanced successional stages (Gallé et al., 1998; Gibb, 2011), such as in natural forest steppe patches. However, habitat type had an effect on the species richness of specialist ants. In contrast to plants, we found higher specialist ant species richness in the plantations than in steppe habitats, since numerous forest species colonized plantations presumably due to the thick litter layer and closed canopy.

Several landscape management strategies based on a binary partitioning of land cover types have been suggested to enhance connectivity between habitat patches in fragmented landscapes, such as stepping stones (Saura et al., 2014) and habitat corridors (Gilbert-Norton et al., 2010). Our study emphasizes that matrix quality metrics may provide more insight into the distribution of organisms than binary metrics (Fischer and Lindenmayer, 2007). Landscape perception is species specific and depends on numerous trait states (Rieder et al., 2018), however, detailed databases do not exist for all species groups of conservation importance. We argue that matrix quality assessment and evaluation for several taxonomic groups could help in developing landscape-wide conservation strategies.

The main implication of our study is that enhancing matrix quality by changing land use (i.e. land cover) type and increasing the share of the fragmented natural habitat in the landscape would help to maintain habitat specialist species. Although, it may not be economically feasible to restore large forest steppe areas in Central and Eastern Europe due to the growing demand for timber production, substituting the relatively poor preforming pine plantations with forest plantations of native tree species, such as silver poplar and establishing extensive pastures would support the native biota of forest steppes. Furthermore, isolation of steppes could be mitigated by substituting the neighboring high contrast plantations with low contrast pastures.

The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.