Remediation of lands devastated by industry includes various forms of restoration, such as technical reclamation and spontaneous succession. These management approaches are debated regarding conservation strategies for postindustrial landscapes. Mining areas consisting of early- to late-developmental stages of both reclaimed and unreclaimed sites offer an opportunity to examine the roles of restoration strategies in a complete successional series for biodiversity and disentangle the contributions of particular biotopes available at postindustrial sites.
Using linear models and multivariate analysis, I tested the effects of (a) developmental stages from early successional sites to mature forests, combined with (b) the initiation process at the sites, which was either technical reclamation or spontaneous succession, and (c) vegetation cover on (i) species richness, (ii) rarity and (iii) species composition of bird communities on 60 plots (100 × 100 m) within opencast mining areas in the lignite basin of north-western Czech Republic.
Bird communities were consistently more species rich on spontaneously developed sites compared with reclaimed sites throughout all stages of the succession series. Species richness increased with site age due to increasing habitat heterogeneity.
The conservation value of bird communities was generally lower on reclaimed sites than on spontaneously developed sites and decreased with site age. The most valuable communities developed on early successional sites and native shrublands, because these were inhabited by specialists that were scarce in the surrounding landscape. By contrast, technically reclaimed sites resulted in impoverished communities, usually with narrow spectra of common species.
Synthesis and applications. The results highlight the importance of spontaneously established sites and complete succession series for developing valuable bird communities in postindustrial areas such as opencast mining sites. In particular, early successional sites and shrublands create refuges for early successional specialists disappearing from the common landscape and these should be promoted at the expense of reclaimed sites wherever possible. My results support an effort for systematic implementation of early successional sites into conservation practice.
World-wide spread of postindustrial sites such as spoil heaps after coal mining, sandpits and stone quarries creates an urgent need to solve problems relating to ecological restoration of affected regions. Mining areas represent about 1% of the world's land area, and in particular, surface lignite mining and associated extensive disturbances influence biodiversity over large areas (Cardoso da Silva & Vickery 2002; Hüttl & Gerwin 2005). A process of technical reclamation has been commonly applied, which includes enriching substrates with nutrients and surface planing, followed by over-grassing or afforestation (Holl 2002; Hendrychová 2008). By contrast, some postmining sites have been left without intervention, thus providing opportunity for spontaneous natural succession. Studies on plants and invertebrates have revealed that these spontaneously developed sites often promote more diverse communities and/or provide better habitats for rare species than do sites that had undergone reclamation processes (Bröring & Wiegleb 2005; Prach & Řehounková 2006; Kirmer et al. 2008; Mudrák, Frouz & Velichová 2010; Tropek et al. 2010, 2012; Hendrychová et al. 2012). On the other hand, similar studies focused on vertebrates such as birds are rare (Karr 1968; Bejček & Šťastný 1984) and the differences between the communities of these animals on spontaneously developed and technically reclaimed sites are poorly documented.
Natural succession has occasionally been used as a tool in postmining rehabilitation programmes (e.g. in Germany; Schulz & Wiegleb 2000). Promotion of this alternative to technical reclamation has stimulated debate on the interactions between restoration method and successional age on the character of developing communities (Prach et al. 2011). Primary succession in temperate Europe consists of a stage sequence initiated from bare ground with sparse annuals and biennials persisting up to 10 years, followed by predominant perennial forbs. Shrublands expand after about 15 years and are replaced by deciduous forest after a few decades (Prach & Pyšek 2001; Wiegleb & Felinks 2001). This pattern is also inherent to secondary succession on reclaimed sites where the earliest postdisturbance stages are skipped (Ferguson 2001) and the community development is accelerated and directed by imported nutrient-rich soil and plants (Prach et al. 2011). Spatial and structural habitat diversification that emerges during succession is accompanied by increasing numbers of species (Walker & Moral 2003), a widely recognized indicator of species richness patterns. Studies on plants and invertebrates in postindustrial areas have revealed, however, that early succession sites are inhabited by scarce species, which are early successional specialists (Beneš, Kepka & Konvička 2003; Řehounková & Prach 2008; Tropek & Konvicka 2008; Tropek et al. 2010). The presence of such species enhances the conservation value of these sites, even though they may contain species-poor communities. Therefore, the aspect of species rarity, contributing to community patterns, should also be taken into account (Samu, Csontos & Szinetar 2008; Filippi-Codaccioni et al. 2010).
To add to our current knowledge, differences in community patterns in colonizing species with a high dispersal ability, such as birds, should be examined between technically reclaimed and spontaneously developed sites throughout the complete successional series from early successional sites to mature forests. Birds comprise one of the best indicator-animal groups for evaluating the success of large-scale ecosystem restoration (Cardoso da Silva & Vickery 2002). It is important that newly established suitable sites may be readily occupied by active colonizers and long-distance migrants (Bröring & Wiegleb 2005; Kirmer et al. 2008). However, only a few previous studies have compared the effects of spontaneous succession versus technical reclamation on biodiversity and nature conservation value. In addition, most of the available studies deal with plants or invertebrates. To my knowledge, moreover, none of these few studies took into account the entire successional series from early postdisturbance sites to mature forests.
Postmining areas are large landscape segments consisting of broad habitat spectra comprising of all successional series stages with different management histories; therefore, they provide a unique opportunity to (a) test the combined effect of successional series and management intervention on the community, (b) assess the refugial role of early successional sites in postmining areas for habitat specialists and (c) reveal the importance of postmining sites for biodiversity in human-altered landscapes. In this study, I examined bird community patterns through species richness, rarity and species composition in lignite mining areas ranging from bare ground to mature forests. I tested how the communities in particular stages differ between spontaneous successions and reclaimed sites. I intend for these findings to be useful both in implementing reclamation practice within postindustrial areas as well as in formulating general concepts for nature conservation strategies in human-altered landscapes.
Materials and methods
Research area and study sites
The study was conducted in the lignite basin of Most, north-western Bohemia, Czech Republic, between the towns of Most, Litvínov, Bílina and Chomutov (44 km west–east and 16 km north–south, 50°28′–50°33′N, 13°30′–13°43′E). The basin is hemmed by afforested slopes of the Krušné hory (Ore Mountains) to the north and by semi-steppe mounds of the České středohoří (Central Bohemian Uplands) on the south. Actively working opencast lignite mines combined with overburden dumps after brown-coal extractions dominate the landscape. These are interspersed with agricultural fields, settlements, forests, shrub patches, roads and water bodies. Many spoil heaps and margins of mining pits have been technically reclaimed but some were temporarily left to spontaneous succession. Hereafter, I refer to those sites after technical reclamation as ‘reclamations’ while spontaneously developed sites I term ‘successions’.
In accordance with the internal guidelines of regional reclamation companies, most agricultural reclamations include establishment of permanent grasslands with initial sowing of a species-poor grasses mixture (species of genus Festuca, Dactylis, Phleum, Poa, Cynosurus, Agrostis) mixed with about 10% of legumes (Trifolium, Coronilla, Lotus, Medicago) in amounts of 40–50 kg ha−1 applied yearly for 2–3 years. Neither high-diversity seed mixtures containing several dozen plant species of local provenance to accelerate succession and increase site ecological value (Kirmer, Baasch & Tischew 2012) nor any other kind of assisted site recovery (Baasch, Kirmer & Tischew 2012) is applied in the study area. Afforestation includes predominantly homogenous plantations of even-aged growth with mixes of both autochtonous and allochthonous trees. Monocultures of poplars (Populus) or mixed growths with other deciduous trees (Acer, Quercus, Fraxinus, Tilia, Carpinus) or conifers (Larix) have been supplemented by exotic shrubs (Symphoricarpos, Spiraea, Lonicera, Philadelphus). Early successional sites are characterized by bare ground with sparse annuals and biennials, followed by perennials (Tanacetum, Artemisia, Cirsium) and grasses (especially Calamagrostis epigejos and Arrhenatherum elatius) scattered with shrubs (Sambucus, Rosa, Betula, Crataegus). Mature forests have been birch-dominated (Betula pendula), mixed with other deciduous trees (Salix, Populus) (Prach & Pyšek 2001; own data).
With support of technical documentation (unpublished sources maintained at the Brown Coal Research Institute, Most), I demarcated 60 sites with dimensions 100 × 100 m, evenly distributed between reclamations and successions (the two categories of management history) and with reference to years since initiation (site age) as follows: early successional (2–10 years old) herb-dominated sites, medium-aged (11–17 years) shrub-dominated patches and late (18–45 years) forest-dominated growths. Early successional tree plantations (with sapling monocultures above 1 m in height and up to 10 years since planting) were avoided, as they had no adequate spontaneous counterparts. Medium- and late-aged agricultural reclamations that do not differ from their early-aged parallels were also avoided. The study sites were dispersed randomly across the region but with limitations given by the availability of suitable habitats (Fig. S1 in Supporting Information). To minimize pseudoreplication arising from the proximity of identical habitat, sites with different habitats were preferentially selected in close proximity (but never <400 m apart) within one spoil heap or around one mining pit. Given that many mining pits and adjacent spoil heaps pass indistinctly from one to another; however, precise assignment was usually difficult. The margins of all sites were further than 30 m from the habitat's edge to reduce edge effect.
In addition to management history and site age, the following attributes were estimated using aerial orthophoto maps combined with field inspection: the approximate magnitude of terrain depressions (absent, moderate, distinct, huge), herb cover (up to 1 m of height; % of the area), shrub cover (1–3 m; %) and tree cover (>3 m; %). I mapped the birds during the breeding season from mid-April to early June, excluding days with strong wind and rain. The field survey was carried out between 2007 and 2011, and 12 sites were examined each year. To minimize the year effect, each year I combined sites with various management histories and age. Each site was surveyed twice within a single year by walking at a slow pace for approximately 25 min during the peak hours of bird song activity between 05:00 and 09:30. All birds associated with a selected site (breeding, displaying, feeding, resting) were recorded along a tortuous route chosen to cover the complete study site (up to 500 m in length on sites with dense vegetation or with huge terrain depressions) but never recrossing segments already recorded (Fig. S2 in Supporting Information).
I used bird species richness (i.e. numbers of bird species recorded at each site) as a measure of species diversity. Species richness highly correlated with total bird numbers (Pearson's correlation coefficient, r =0·77, N =60, P <0·003, R2 = 59%), suggesting that species presences could be adopted as an underlying attribute of bird quantity in further analyses. Furthermore, I quantified a species rarity index reflecting the scarcity of each species throughout the Czech Republic (Table S1 in Supporting Information). The index results from large-scale quadrat mapping of birds during 2001–2003 (Šťastný, Bejček & Hudec 2006) and was calculated using the formula 1 − N/628 where N represents the number of quadrats occupied by the species from 628 in total. The sum of indices for all species recorded on a site then gave the site its (bird) rarity value. The indices reflected averages of population ranges for particular species in Europe during the 1990s (BirdLife International 2004; Spearman's rank correlation coefficient rs = −0·54, P <0·05) and thus documented the validity of the obtained indices on a wider-than-regional scale.
Species richness and rarity at the sites were used as response variables in testing the effects of management history, site age and additional habitat attributes on bird community patterns. Prior to analyses, I assessed general relationships among these predictors. Shrub cover positively correlated with tree cover (Pearson's correlation coefficient, r =0·72) as well as with successional age (r =0·74), and tree cover highly correlated with successional age (r =0·92). Moreover, terrain depressions were much more extensive on successions than on reclamations (Mann–Whitney U-test, Z =44·16, N1= 30, N2= 30, P <0·0001). Therefore, shrub and tree cover as well as terrain depressions was interpretable by either age or management history and were thus omitted from statistical analyses; instead, the management history, age, herb cover and their interactions were included as predictors in the models. Species richness as well as square-root transformed values of rarity were normally distributed (Shapiro–Wilk normality test, both W <0·99, P >0·35).
I applied ArcGis (ESRI® ArcMap 9.2, ESRI, Redlands, California, USA) to analyse the spatial relationships among six habitat categories (early, medium and late stages of both successions and reclamations) using Moran's I (Fortin & Dale 2009). The distance between nearest neighbouring plots varied greatly between 400 and 2100 m, with median 780 m, so the spatial relationships were conceptualized following the inverse distance method with critical distance equal to this median. The arrangement of habitats showed not a spatially autocorrelated but a random pattern on this scale [Moran's I =0·43, Z score (SD) = 1·93, P >0·05), justifying the use of nonspatial regression methods for data analysis. I employed general linear models using r 2.13.1 (R Development Core Team 2011). Estimates with standard errors (SE) are presented for significant predictors as a measure of effect size.
I explored variation in species composition of bird communities using direct gradient (redundancy) analysis (RDA) summarizing the relationships between occurrences of bird species and habitat variables (same predictors as in the univariate analyses, that is, management history, age, herb cover and their interactions). Prior to this, I checked, using detrended correspondence analysis (DCA), that species data respond linearly to the gradients and that use of a linear method (RDA) is appropriate (ter Braak & Šmilauer 2002). A Monte Carlo permutation test with 499 unrestricted permutations and significance level P = 0·05 was applied to test the importance of predictors and canonical ordination axes. I utilized a manual forward selection procedure from an empty to more complex model with stepwise ranking of variables by their importance computed by CANOCO for Windows 4.5 (ter Braak & Šmilauer 2002).
Herb cover on early successions was lower (maximum 70%) compared with reclamations of the same age (cover between 70% and 100%; Mann–Whitney U-test, Z = 3·34, N1 = 9, N2 = 8, P =0·0002; Fig. 1), but in later stages, it varied widely without significant differences between successions and reclamations (Z =0·11, N1 = 21, N2 = 22, P =0·913). Shrub cover, not detected before the 10th year after site initiation, varied widely with no significant effect of management history (Z =0·67, N1 = 30, N2 = 30, P =0·497). However, tree cover, detected from the 18th year of site development, was lower on successions (between 70% and 95%) than on reclamations (usually between 90% and 100%; Mann–Whitney U-test for stages between 18 and 45 years: Z =2·87, N1 = 14, N2 = 14, P =0·003).
Bird species richness and rarity
In total, 72 bird species (listed in Table S1) entered into the analysis. Site age and management history significantly affected both species richness and rarity (Table 1). Species richness increased with age (estimate = 0·17 ± 0·030) for both reclamations and successions, while successions were richer in all particular stages (−4·19 ± 0·90; Fig. 2a). Rarity, by contrast, decreased with site age (−0·01 ± 0·003) on both successions and reclamations, even though the regression slope for the former was moderately steeper (with nonsignificant difference between the correlation coefficients, P =0·485). Again, the value was higher on successions than reclamations (−0·55 ± 0·100) throughout all stages of the successional series (Fig. 2b; Table 2). Neither herb cover nor any interaction contributed significantly to variation in species richness and rarity.
Table 1. Results of the general linear model documenting contributions of treated predictors to species richness (A) and regional rarity (B) of birds in opencast mining areas
Management history × herb cover
Management history × site age
Management history × site age
Management history × herb cover
Table 2. Means and standard errors (±SE) of bird species richness and rarity for all habitat categories including early, medium and late stages of spontaneously developed sites (S) and technically reclaimed sites (R) and summary results for S and R irrespective of site age
Number of plots
Bird community composition
Bird species composition was most strongly influenced by site age followed by herb cover and management history (Table 3). In general, species composition on the sites was more sensitive to habitat alternatives, with regard to significant interaction of management history and site age, than were species richness and rarity. The fixed effect of site age was strongly correlated with canonical axis 1 (r =0·99), of herb cover with axis 2 (r =0·66), and of management history with axis 3 (r = −0·61; Table 3). If any variable other than site age was prioritized and tested as a single predictor, its explanatory power would not achieve that of site age (herb cover: F =2·80, P =0·01; management history: F =1·78, P =0·048; management history × site age: F =4·37, P =0·002; management history × herb cover: F =2·69, P =0·014), suggesting that site age was the prime predictor of bird community structure.
Table 3. Summary results of forward selection procedure in redundancy analysis of the effect of predictors on species composition of bird communities in opencast mining areas
The ordination diagram (Fig. 3) presenting the two principal gradients (site age and herb cover) and projecting the directions of increasing occurrences in particular bird species provides global insight into variation in the composition of bird communities. Early sites were represented by either bare ground (successions) inhabited by species such as Anthus campestris, Charadrius dubius and Oenanthe oenanthe, or contrariwise, with denser herb cover typically inhabited by Locustella naevia, Sylvia communis, Miliaria calandra, Saxicola rubetra and S. torquata. Frequent bird species of moderately overgrown early sites were Luscinia svecica, Alauda arvensis, Motacilla flava and Carduelis cannabina. Medium-aged stages were represented by shrub species such as Lanius collurio, Emberiza citrinella and Sylvia curruca. Also, these species tended to appear on successions rather than reclamations. Late stages were inhabited by a number of common forest species. Positions of envelopes for forest sites indicate, similar to the previous shrub stage, an association of many species with successions but only few species (e.g. Phylloscopus sibilatrix or Erithacus rubecula) with forest plantations. Finally, the shapes of envelopes refer to relationships between variation in habitat structure and bird communities. Vertically elongated envelopes for herb and scrub successions indicate higher variation in herb cover and associated diversification of bird communities compared with reclamations, the latter of which, by contrast, exhibit more uniform and intermediate positions that reflect lower variations in both habitat structure and bird communities.
Exclusiveness of bird communities
I tested differences between the numbers of species that exclusively appeared in particular habitats. An uneven pattern was found (Fig. 4), revealing that successions held significantly more exclusive species than did reclamations (17 vs. 7 species; homogeneity test, = 4·17, P =0·041). However, there were no significant differences between herb, shrub and forest stages (= 3·25, P =0·197).
Habitat deprivation on reclamations
Birds established generally richer communities with higher species rarity on spontaneously developed sites than on reclaimed sites throughout all stages of the successional series from early bare ground to mature forests. Several habitat disparities between reclamations and successions can be responsible for these differences.
First, the initially uneven soil surface has been regularly reduced by exhaustive planning during reclamation. This led to habitat deprivation for birds such as Luscinia svecica, Charadrius dubius, Locustella naevia and Acrocephalus arundinaceus, which inhabit wet patches as well as associated vegetation (sedges, reeds, willows) that emerge in the small depressions which occur widely across successions above impermeable clay-type subsoil. Diverse vertical habitat architecture, including ragged soil surface, important for community diversification in plants (Prach & Pyšek 2001; Mudrák, Frouz & Velichová 2010), invertebrates (Hendrychová 2008; Tropek et al. 2010) and amphibians (Doležalová et al. 2012), has also been found to be influential in birds (Karr 1968; this study).
Second, introduction of species-poor but continuous grasslands and uniform wood plantations has been found to hinder diversification of vegetation cover (Mudrák, Frouz & Velichová 2010; Kirmer, Baasch & Tischew 2012) that is required, for example, for rich communities of surface-dwelling beetles (Brändle, Durka & Altmoos 2000). Also, uniform and continuous herb cover on reclaimed sites leads to an absence of specialists such as Anthus campestris, Oenanthe oenanthe or Charadrius dubius, which require patches of bare ground for breeding and foraging (Šťastný, Bejček & Hudec 2006). In mature forests, this adjustment results in contrasts between reclamations characterized by species-poor, age and spatially uniform spinneys with impoverished bird assemblages and successions with patchy clumps of unevenly aged trees interspersed with glades inhabited by diverse bird communities (Bejček & Šťastný 1984; Hendrychová, Šálek & Řehoř 2009). High internal heterogeneity is inherent to natural forests in central Europe (Fuller 2003), and most temperate woodland birds are attracted by diverse tree mosaics rather than by uniform and dense secondary forests (Šálek, Svobodová & Zasadil 2010). Excessive use of allochthonous woody plants in reclamation practice may add to this structural simplification on reclaimed sites and reduce bird diversity (Marzluff & Ewing 2001) with negative effects particularly on more specialized species that are adapted to island-like forest patches with complex structure (Magura, Baldi & Horvath 2008).
Bird species richness and rarity
Species richness and rarity represent two different indicators of community patterns. Whereas species richness directly reflects general habitat complexity across taxa and habitats (Karr 1968; Walker & Moral 2003), rarity and specialization rate are less trivial as they are derived from regional abundances and habitat use of individual species (Filippi-Codaccioni et al. 2010). Consistent with the above prediction, species richness increased with successional time, but, in contrast, rarity decreased with site development. While forests were dominated by common birds, shrub communities included scarce species (Luscinia megarhynchos, Lanius collurio, Miliaria calandra, Sylvia nisoria), many of which are regarded as shrub specialists (Reif, Jiguet & Šťastný 2010). In large patches of medium-aged successions, namely those with predominant Rosa canina, these birds created unique assemblages that were absent in forests. Similarly, the earliest postdisturbance successions with sparse vegetation formed very specific communities with few highly specialized and/or threatened species (M. calandra, Motacilla flava, Anthus campestris, Oenanthe oenanthe, Saxicola sp.).
The discrepancy in trends of species richness and rarity throughout succession might have a basic ecological explanation that common species are more likely occupy all available sites and contribute to diversity patterns more than rare species, which as a result, dominate in poor-community sites (Lennon et al. 2004; Šizling et al. 2009). However, rarity in this study was expressed so as to emphasize the contribution of rare species (instead of using number proportions only) while reducing the effect of the abundant ones. Moreover, none of the common species would appear everywhere, owing to the huge habitat variation throughout the complete successional series. Therefore, instead of this basic ecological explanation, I suggest that the decreasing rarity during succession is due to the different status of particular successional stages in the context of the surrounding landscape.
Species rarity in landscape context
The forests of the Czech Republic cover large areas and are broadly populated by common forest birds. In contrast with the stable or moderately increasing populations in many of these species (Reif et al. 2008), numerous farmland species sharply declined during the second half of the 20th century across European countries (Gregory et al. 2005), which is similar to the situations observed in several other groups of plants and invertebrates known as early successional specialists (Beneš, Kepka & Konvička 2003; DeGraaf & Yamasaki 2003; Kadlec et al. 2008; Řehounková & Prach 2008; Tropek et al. 2010). A major reason for this negative trend across taxa seems to be reduced availability of suitable habitats throughout the landscape, as the prevailing open landscape is agricultural and has been stressed by permanent cultivation and crop harvesting. This form of ‘disclimax’ (Timoney, Peterson & Wein 1997) constitutes an interminable impediment to creating a dynamic mosaic of early successional sites, each of which would persist for at least several years. If small fragments remain uncultivated but adjacent to managed fields, they undergo a process of secondary succession and are quickly replaced by herb generalists (Prach & Řehounková 2006). In such a landscape, survival of many plant and animal species demanding early successional sites is greatly limited. Among birds, the strongest early successional specialists found on the studied plots, Anthus campestris and Oenanthe oenanthe, had been common in the past on extensive pastures and idle fields but have disappeared from the intensely cultivated farmland across Europe (Hagemeijer & Blair 1997). Reif & Marhoul (2010) have reported that these species had disappeared from abandoned military areas once the early successional denuded patches had become overgrown with denser herbs. This implies that early successional sites were more common in the past and that at least some primary successional specialists tolerate secondary successional sites.
Conclusions and conservation implications
Postmining areas comprise valuable surrogate habitats for sustaining biodiversity, including among birds. Indeed, more than 40% of bird species found in this study were observed to be declining in Europe at the end of the 1990s (BirdLife International 2004). Technical reclamation leads to habitat impoverishment and reduces the natural potential of mining sites with respect to biodiversity and nature conservation value over several decades. Therefore, technical reclamations should be avoided wherever possible in favour of a dynamic mosaic of different successional stages. Leaving freshly disturbed sites completely unreclaimed can be adopted as the lowest-cost and proper management tool, except in the most extreme parts, such as where toxic soil needs to be buffered (Hendrychová 2008) or stabilization against erosion is necessary (Hodačová & Prach 2003). In the latter situations, a combination of spontaneous and assisted site recovery using near-natural restoration methods with respect to site potentials (Baasch, Kirmer & Tischew 2012) might be a proper alternative. Support for spontaneously developed forests is justified particularly where forests constituted the major natural ecosystem before initiation of industrial activities (Hüttl & Weber 2001). Ongoing mining and heaping of spoil dumps create newly denuded sites and provide the complete successional series from early successional sites up to mature forest and supports high β-diversity of the region (Harabiš & Dolný 2012).
Lignite mining areas are similar to such other postindustrial sites as stone quarries, extracted peatlands, sand and gravel pits, or suburban landfills, as these produce highly specific habitats with limited nutrients and/or experiencing occasional disturbances combined with temporary abandonment, all of which prevent rapid vegetation development and continuous cover (Brändle, Durka & Altmoos 2000; Šálek et al. 2004; Kadlec et al. 2008; Dulias 2010; Tropek et al. 2010; Hendrychová et al. 2012). This gives early successional specialists recruited from vascular plants, various groups of arthropods as well as birds the opportunity to appear and to use these sites as habitat refuges within human-altered landscapes. Recognizing the positive effects of early successional sites for maintaining biodiversity provides motivation for their greater support in other agricultural landscapes within Europe, as well as in North America, where these habitats have become scarce (Dettmers 2003; Oehler 2003). It is not easy, however, to perform such maintenance in conservation practice. Various management tools might be potentially available, as documented in the case of wildfires supporting the formation of open-land Mediterranean bird communities (Clavero, Brotons & Herrando 2010) or human activities recommended as a replacement use for military training areas after abandonment by soldiers (Reif & Marhoul 2010; Reif et al. 2011). In addition, diverse agri-environmental schemes including extensive grazing and cutting regimes (Vickery, Carter & Fuller 2002; Kleijn & Sutherland 2003; Woodcock et al. 2009) could offer suitable alternatives applicable to intensively cultivated open landscape. However, one must discriminate among various types of disturbance and landscape contexts because disturbances might also lead to unfavourable homogenization of communities (Clavero, Brotons & Herrando 2010) or spread of alien species (Botham et al. 2009). Therefore, to specify the best approach in particular sites and circumstances, we urgently need more detailed studies analysing effects of different types of disturbances on the survival and reproduction of disappearing specialists, which form the communities on early successional sites.
I thank M. Hendrychová for providing facilities and habitat descriptions, V. Kubelka, M. Řehoř and S. Chocholoušková for field assistance, as well as J. Douda, K. Gdulová and F. Harabiš for constructive suggestions. Anonymous referees provided valuable comments on earlier versions of this manuscript. The author owes special thanks to G.A. Kirking and C. Kimbrell for their useful linguistic advice. This study was funded by the Czech Science Foundation, grant no. 105/09/1675.