Introduction

Habitat fragmentation and deterioration of habitat quality are two of the major threats to biodiversity. Habitat fragmentation leads to a loss of habitat, a reduction of patch size and an increase in patch isolation (Andrén 1994; Fahrig 2003). In western Europe, where habitat fragmentation has reached a high level, studies reporting a decline in insect biodiversity are numerous (e.g. Thomas et al. 2004; Wenzel et al. 2006; Stefanescu et al. 2011). The effect of habitat fragmentation on local population dynamics has received attention as the main subject of metapopulation studies (e.g. Thomas et al. 1992; Hanski 1994) and, for practical reasons, the conservation of many species (Hanski and Ovaskainen 2000). According to predictions from metapopulation theory (Levins 1969), patch size and its isolation are the key factors influencing species occurrence, abundance and density in a fragmented landscape (e.g. Hill et al. 1996; Hokit et al. 1999). At the community level it has been demonstrated that fragmented, smaller habitat patches have lower species richness than larger and less isolated ones (Steffan-Dewenter and Tscharntke 2000; Tscharntke et al. 2002). In addition, several field studies focused on the resource-based concept of habitat indicate that quality of habitat patches may be as important as patch area and isolation (Thomas et al. 2001; Dennis et al. 2003; Dennis and Sparks 2006; Dennis 2010).

The effects of patch size, isolation and quality may also be modified by several other factors such as human disturbance or permeability of the landscape matrix between patches (Tscharntke and Brandl 2004; Grundel and Pavlovic 2007). For example, in areas of high human density some areas may be trampled, littered or the behaviour of butterflies may be modified (Stankowich and Blumstein 2005). Obviously, the permeability of the matrix as well as habitat patch characteristics are now inevitably the result of human activity (Dover and Settele 2009). Some elements in a landscape may be ecological corridors and others may be barriers that enhance and limit, respectively, the dispersal of individuals (Grundel and Pavlovic 2007; Prevedello and Vieira 2009).

Obviously, species differ in their response to changes in the environment. Some species are more vulnerable to habitat fragmentation and others to habitat deterioration, or the species respond differently to the same environmental characteristics (Dover et al. 2011). From the conservation perspective the question of if, and how, species within a community are similar in their response to the same environmental variables is of primary importance (WallisDeVries 1999). High unpredictability in species response to the changes in their environment may cause difficulty in establishing effective conservation procedures, especially when differently responding species are also those of high conservation value. It is astonishing, to some degree, that this problem has only rarely been studied (Shreeve et al. 2001; Dover et al. 2011). In conservation biology there is a prevailing focus on the conservation of individual species or the conservation of total species richness and diversity. Therefore, for well-designed conservation treatments, it is essential to have detailed knowledge about responses to environmental factors of each butterfly species, other co-occurring species, as well as the whole community (Shreeve and Dennis 2011).

One of the vegetation types considered severely threatened due to several human-related changes (e.g. abandonment of management, subsequent invasion of shrubs, high fragmentation due to agriculture intensification, habitat quality changes related to nutrient flow from surrounding farmland) is semi-natural calcareous grassland (Wenzel et al. 2006). In fact, in central Europe, many calcareous grasslands are limited to small isolated patches, often located on hills or limestone outcrops (Krauss et al. 2003; Wenzel et al. 2006). This vegetation type, however, is of leading conservation interest because of a high richness of plant and insect species (Van Swaay 2002; WallisDeVries et al. 2002). Butterflies are considered a model organism for fragmentation studies (New 1991) since they respond relatively quickly to habitat changes (Bourn and Thomas 2002). Finally, butterflies are known to be indicators of habitat quality and ‘umbrella’ species because conservation of their habitats may also benefit other taxa (New 1991; Thomas et al. 2004).

The aim of this study is to determine the (relative) influence of several environmental factors affecting butterfly species richness, diversity, commonness and individual species in fragmented calcareous grasslands. Based on metapopulation landscape ecology and resource-based concepts, we predict that butterflies should show a higher species richness, probability of presence, and abundance in grasslands that were: (1) larger, (2) less isolated, (3) of higher quality, (4) distant from human settlement, (5) located in a more permeable landscape matrix with higher grassland cover (possibly positively affecting dispersal) and lower forest cover (possible barrier to movements).

Materials and methods

Study area

The study was conducted in 32 calcareous grassland patches in southern Poland (50°01′N; 19°54′E), 4–10 km southwest of Kraków’s city centre. The total area covered by calcareous grasslands was 87.3 ha and the study landscape covered ca. 40 km2. Calcareous grasslands were easily distinguishable from the surrounding landscape as they were located on hills adjacent to the flat valley of the River Vistula which is covered mostly by arable land and fallow. At the time of this study all but three calcareous grasslands had been abandoned for about 10–15 years and isolated shrubs were scattered within them. Mowing was noted in two grassland patches: in one a small part (~20% of the patch) was mown in late July 2008 and in the second grassland 70% of its area was mown (twice: in June and in August 2008). We noted also one grassland where goats grazed.

Butterfly surveys

In all habitat patches we established 5 m-wide transects, on which the presence and abundance of each butterfly species was noted. The length of the transect was proportional to patch area (r = 0.89, P < 0.001) and varied between 30 and 960 m (mean ± SE: 248 ± 46 m). The transects followed a zig-zag pattern within the patch. Three counts were made in each between the 15th July and 20th August 2008 separated by approximately 1 week. Butterflies were surveyed between 09.00 and 16.00 h during favourable weather conditions (maximum wind: 3 on the Beaufort scale, cloud cover up to 25%, minimum temperature 20°C). The average speed of the transect walk was about 200 m per 10 min. However, the minimum time spent for counting butterflies was 5 min in the smallest patches. The order in which the potential habitat patches were recorded was random.

Explanatory and dependent variables

We measured the following eight environmental variables potentially affecting butterfly populations:

  1. 1.

    Patch size (ha);

  2. 2.

    Isolation; distance to the nearest calcareous grassland (m);

  3. 3.

    Grassland; % cover of permanent grassland within 500 m of the patch boundary. Grasslands may act as an ecological corridor or stepping stones enhancing dispersal of the species in calcareous grasslands. For many species in this study, grasslands were also breeding sites;

  4. 4.

    Forest; % cover of forest within 500 m of the patch boundary. Forests may act as a barrier to butterfly movements;

  5. 5.

    Buildings; distance to the nearest human settlement (m) as a measure of human activity;

  6. 6.

    Number of plant species as a measure of diversity. 5–10 circular quadrats of 1 m diameter (0.79 m2) were randomly placed on each transect in July and the cover of each species recorded and averaged per patch (Skórka et al. 2007).

  7. 7.

    Wind shelter. The % of the patch perimeter protected by forest or shrubs;

  8. 8.

    Vegetation height; mean height (cm) from 10 random measurements per quadrat (see 6 above) and averaged per patch (Skórka et al. 2007).

Environmental variables 1–5 and 7 were calculated with ImageJ software or directly in the field with the use of a GPS. Variables 6 and 8 were measured during field surveys. The basic characteristics of investigated patches are summarised in Table 1.

Table 1 Basic characteristics of calcareous meadow patches (n = 32) in the study landscape
Full size table

Total butterfly species numbers (species richness), the total number of individuals (abundance) and Simpson’s Index of Diversity were calculated for each patch. The number of occupied grid squares from the Polish Butterfly Atlas (Buszko 1997) based on field work in 1986–1995 was used to provide a commonness (non-rarity) value for each species. A mean commonness score, weighted by species abundance, was then calculated for each patch. For our recorded species this score could, hypothetically, range from 2 (if only Minois dryas was recorded in a patch) to 784 for Pieris brassicae only. In our patches the mean score varied from 301 to 657.

Statistical analysis

Forward selection stepwise regression was used to relate the measures of butterfly richness, abundance, diversity and commonness to the eight environmental variables. Variables significant at 0.05 < P < 0.1 were retained in these models. A canonical ordination was used to relate the abundance of the individual species to the eight environmental variables using the CANOCO package. Since the length of the longest gradient in DCCA was so short (1.24) we opted to use Redundancy Analysis (a canonical form of PCA) for this ordination. Species data were log(x + 1) transformed before analysis. In addition to Redundancy Analysis, we built presence-absence and abundance models to estimate statistical significance of each environmental variable for each species. Presence-absence data were analysed by generalized linear models with a logit-link function and abundance data were analyzed with stepwise regression. The presence-absence and abundance models were built for species found in at most 27 and at least in five patches, respectively. As before, a forward selection method was used and variables significant at 0.05 < P < 0.1 were retained in these models.

Results

General description of the butterfly community

A total of 2,685 individuals belonging to 36 species were observed during transect surveys (Table 2). The most abundant species were Maniola jurtina (15.5%), Polyommatus icarus (12.2%), Melanargia galathea (11.5%), Aphanthopus hyperanthus (10.2%), Pieris rapae (7.4%), Thymelicus lineola (6.4%), Polyommatus coridon (6.3%), Coenonympha pamphilus (5.3%) and Aglais io (5.0%).

Table 2 Factors affecting the occupancy and abundance of butterfly species in calcareous grasslands (species ranked by abundance)
Full size table

Responses of richness, abundance, diversity and commonness to environmental variables

A summary of the regression models is presented in Table 3. Species richness, abundance and diversity were all positively related to shelter. Butterfly species richness was also positively associated with patch size and plant species richness. Abundance was also positively associated with patch size. The commonness score was positively associated with distance to buildings and negatively with vegetation height suggesting that rarity was greater close to buildings and in the presence of short swards.

Table 3 Significant predictor variables of butterfly species richness, abundance, diversity and commonness from stepwise regression models
Full size table

Community responses to environmental factors

The first two axes of the RDA ordination explained 19.2% of the variation in butterfly species, of which the environmental variables explained 59.5%. The ordination of the species is shown in Fig. 1, where labels for species present in less than 5 patches have been omitted. One group of species appears to be spread in a positive direction along axis 1 and another in a positive direction along axis 2. The ordination of the eight environmental variables is shown in Fig. 2. This suggests that positive values on axis 1 were associated with larger patches (and to a lesser extent wind shelter and isolation) and positive values on axis 2 with short vegetation.

Fig. 1
figure 1

RDA ordination of 36 butterfly species in 32 habitat patches. Species are identified by abbreviated scientific names. Labels for species occurring in less than five patches have been omitted. The two axes explained 19.2% of the variation in butterfly numbers

Full size image
Fig. 2
figure 2

RDA ordination of eight environmental variables in relation to the butterflies shown in Fig. 1. The environmental variables explained 59.5% of the variation displayed in Fig. 1

Full size image

Presence-absence and abundance models for individual species

Models predicting presence/absence of each species explained on average less variation (29%, n = 17 models) than regression models of the abundance of each species (45%, n = 13) (Table 2). In presence/absence models, variables that were significant most often included patch size (six cases), isolation (six), plant species richness (five) and grassland cover within a 500 m radius (four cases) (Table 2). Moreover, for six species none of the variables significantly explained the presence/absence on transects (Table 2). For 13 species, presence/absence models could not be built due to small sample size (Table 2).

In models predicting butterfly abundance the most important factors appeared to be patch size (six cases) and plant species richness (five). In ten species none of the variables was significant (Table 2). For 13 species, models could not be built due to small sample size (Table 2).

The most consistent influence on species was patch size which positively affected the probability of presence and the abundance of species in most models (Table 3). The species responses to shelter and vegetation height were also consistent in both types of models, however only a few species were affected by these variables (Table 2). For other variables the response of species was highly inconsistent. For example, increasing isolation positively affected the probability of presence of two species but negatively of four other species. Plant species richness positively affected the abundance of three species but negatively for two other species (Table 2).

Discussion

These results show the importance of habitat patch size in explaining butterfly species richness and abundance in fragmented calcareous grasslands in southern Poland. The models for individual species also showed that patch size was the most frequent variable explaining variation in butterfly presence-absence and abundance in calcareous meadows. These findings are consistent with most studies in north-western Europe on butterfly-patch area relationships (e.g. Steffan-Dewenter and Tscharntke 2000; Krauss et al. 2003; Polus et al. 2007), however some studies also identified that patch size was not a good predictor of population persistence (Pellet et al. 2007). The benefit of large habitat patches for butterfly communities may result not only from its greater size but also from higher colonisation rates and often increased microhabitat diversity (Nowicki et al. 2007).

One of the most interesting results in our study was that shelter played an important positive role in butterfly species richness, abundance and diversity. At the individual species level this variable also consistently positively affected occupancy and abundance. Wind is a phenomenon that may affect butterfly behaviour (Dover et al. 1997; Brattström et al. 2008), however its direct influence on the foraging behaviour of individuals and community structure is hardly known. Calcareous grasslands in our study area are exposed slopes of hills in a rather flat rural landscape, thus they may be especially exposed to wind. In another study (authors’ unpublished data) we noted that, in meadows located in forested areas, various butterfly species were active (frequent flights) on windy days, when butterfly activity in open landscapes was low. Moreover, the edges of calcareous grasslands with trees or shrubs may provide special habitat conditions (e.g. perches) or food resources attractive to some species.

Patch quality, as measured by the number of plant species, was also important for butterfly richness in our study landscape. These results are in line with those from other studies (e.g. Steffan-Dewenter and Tscharntke 2000; Thomas et al. 2001). Moilanen and Hanski (1998) found that immigration increased and emigration reduced in patches containing high densities of flowers. This indicates that additional resources may support larger populations and/or enhance population persistence (Sutcliffe et al. 1997). Thus, the quality of a patch can, to some extent, act as a substitute for patch area (Hanski and Ovaskainen 2000) and enhance the predictive power of metapopulation models (Thomas et al. 2001). Although occupancy and abundance of several species was positively affected by plant species richness there were a few butterflies (e.g. Aphantopus hyperanthus, Melanargia galathea) which responded negatively to this variable. Larvae of these species are associated with grasses and adult butterflies may prefer areas with a higher cover of grass rather than other plant species.

Another component of habitat quality—vegetation height—had no significant effect on total species richness, abundance or diversity but was important in the ordination of butterfly communities and in explaining differences in commonness and rarity. This suggests that, to successfully protect species inhabiting calcareous grassland, vegetation should be kept at diverse heights within the patch to meet microhabitat requirements of different species, but that shorter swards are critical for some rarer species. This may be achieved by grazing and/or mowing at variable intensity. Such conservation interventions are necessary because currently most of the calcareous grasslands in the study area are abandoned.

The presence of dispersal barriers (forest cover) and potential dispersal corridors (other grassland types) in the surrounding of the habitat patch seemed less important for species richness, total butterfly abundance, diversity and the commonness score. Only at the individual species level were these variables significant for a few species but in an inconsistent way. For example, the abundance of Maniola jurtina was positively affected by both forest and grassland cover in the landscape surrounding the patch whereas the abundances of other species e.g. Aphanthopus hyperanthus, Pieris rapae were negatively affected by forest and grassland cover, respectively. Landscape heterogeneity in the surroundings of habitat patches is known to influence patch connectivity and, in turn, movement patterns (Roland et al. 2000; Tischendorf and Fahrig 2000). Moreover, surrounding grasslands may provide additional food resources for some species and enhance population density and survival rates (Debinski and Holt 2000; Norton et al. 2000). It should also be noted that the proximity of forest might be favourable for butterflies when acting as protection from the wind, enhancing butterfly activity in adverse, windy weather (Dover et al. 1997). Our study indicates that these variables affect individual species rather than total species richness and abundance.

Curiously we found that distance to buildings appeared to be influential on commonness-rarity with higher rarity closer to buildings. Several individual species also had higher occupancy or abundance near human settlements (e.g. Polyommatus coridon, Thymelicus lineola, Erynnis tages). The grasslands close to farms could be those abandoned last and, therefore, probably the most favourable for many rare species associated with short vegetation. It is also possible that people living in the neighbourhood and visiting the patches (most are located in a scenic landscape) may prevent shrub succession to some degree due to trampling and therefore they may create microhabitats for butterflies. Another possibility is that rarer butterflies may benefit from the proximity of human settlement because most of them are single-family houses often possessing flower-rich gardens. Thus, gardens might be a collateral source of nectar for butterflies (Dunning et al. 1992; Ouin et al. 2004). Results from other studies are equivocal; some showed a significant negative effect of human activity on butterfly populations (Kitahara and Fujii 1994; Clark et al. 2007) while others did not find an effect (e.g., Collinge et al. 2003; Nowicki et al. 2007).

Management recommendations and final remarks

We have shown that species richness, abundance, diversity and commonness are often linked with contrasting environmental variables. This requires conservation actions to balance and manipulate several environmental variables. Moreover, conservation actions may need to be complex because different butterfly species respond differently to the same environmental variables (Dover et al. 2011; Shreeve and Dennis 2011; Williams 2011) and, in effect, these species may show opposing responses to different management treatments (WallisDeVries et al. 2002). Our results strongly support these findings. Therefore, appropriate habitat management depends on the specific aim of conservation of all species since conservation of particular species may not necessarily coincide with conservation at the community level and vice versa (WallisDeVries et al. 2002). It has to be borne in mind that conservation action for butterflies in calcareous grassland which is focused on sustaining or improving a specific component of biodiversity (e.g. a high number of species) may not guarantee that other components of biodiversity (e.g. high population sizes or species diversity) will be achieved. Weibull et al. (2000) suggested that to ensure high butterfly species richness and diversity, habitat patches as well as the surrounding landscape should be heterogeneous (Shreeve and Dennis 2011). The results of our study also allow to us to suggest several important management recommendations.

Firstly, our results show that habitat patch size is one of the more important variables influencing species number and abundance. This variable was also important for several individual species. Therefore, maintaining large calcareous grasslands should be the key issue in conservation of butterflies in this habitat. Shelter (low wind) appeared the second most important variable and may be achieved by diversification of grassland edges by planting trees and shrubs and maintaining hedgerows acting as a shelter (Sparks and Parish 1995; Dover et al. 1997; Dover and Sparks 2000). To maintain high plant species richness and protect habitat patches from invasion of shrubs, extensive grazing or mowing should be applied as it is known to increase plant species number (Morris 2000). We propose that maintaining diverse vegetation heights in these grasslands is also important, therefore it may be good practice to apply rotational grazing or mowing (Cremene et al. 2005).

In this study we focused on factors driving different components of butterfly diversity in calcareous grasslands and the response of individual species to habitat and landscape variables. We have shown that the conservation of butterflies involves several variables, often acting in opposing ways, especially at the level of individual species. We think the most important challenge for future studies is to work out a model of management of calcareous grasslands that guarantees conservation of the entire community of insects and plants in this important vegetation type.