Movement of two grassland butterflies in the same habitat network: the role of adult resources and size of the study area


Christine Schneider, Swedish University of Agricultural Sciences, Department of Landscape Planning Alnarp, PO Box 58, S-23053 Alnarp, Sweden. E-mail: christine.schneider@


Abstract. 1.  Movement patterns of two butterfly species (meadow brown Maniola jurtina L. and scarce copper Lycaenae virgaureae L.) were studied in a 172 ha area within a landscape with a high percentage of suitable habitats for mark–release–recapture experiments.

2.  Adult resource density, but not patch size or larval food plant abundance, influenced the numbers and the fractions of residents, emigrants, and immigrants.

3.  Differences between species were observed in movement frequency and maximum distances moved but not in mean distances moved.

4.  The scarce copper showed much greater movement ability than expected from the results of published studies. This is believed to be a result of the comparatively large size of the study area and the high cover of suitable habitat (>50%).

5.  The mean and maximum distances travelled by butterflies reflected differences in the size of the study area.


The loss and fragmentation of habitat has become common in large parts of Europe and is seen as a major threat to many species (Wilcove et al., 1986; Jongman, 2000), and much research in conservation biology has been focused on understanding processes that enable species to survive in fragmented landscapes (Fry, 1995; Cooper & Power, 1997). The metapopulation concept provides one explanation for how species in fragmented landscapes can survive. Movements of relatively few individuals between remnant, spatially separated populations ensure the survival of the metapopulation as a whole through recolonisation of patches where local populations have gone extinct (Gilpin & Hanski, 1991; Hanski & Gilpin, 1997). Therefore, studying animal movement and understanding the factors affecting movement have become important issues in conservation biology and landscape management.

Many butterfly species are threatened or are declining rapidly due to loss and fragmentation of their habitats (Thomas, 1984, 1991; New et al., 1995), consequently there has been an increasing need to understand their ecology and in particular their dispersal. It has been pointed out, for example by Petit et al. (2001), that it is important to quantify not only movement but also the factors that affect movement. So far, patch area, patch isolation, patch quality, and sex have been identified as factors influencing butterfly dispersal in landscapes with fragmented habitats (Dover et al., 1992; Hanski, 1994; Hill et al., 1996; Kuussaari et al., 1996; Baguette et al., 1998, 2000; Petit et al., 2001).

In this study, the impact of patch characteristics on the movement of two grassland butterflies (meadow brown Maniola jurtina L. and the scarce copper Lycaena virgaureae L.) was tested using mark–release–recapture experiments. Most butterfly movement studies have focused on the abundance of larval food plants when patch quality was considered as a factor affecting movement. In this study, flower density, as a measure of adult resource density, was investigated in addition to patch area, patch distance, and the abundance of larval food plants.

A large study site was chosen because recent work has shown that the classification of a species as sedentary may be related more to the size of the study area than to the species' actual dispersal ability. Some so-called sedentary species have been shown subsequently to be able to disperse several kilometres (Brunzel & Reich, 1996; Nève et al., 1996; Mousson et al., 1999; Petit et al., 2001). The issue of underestimation of butterfly dispersal in mark–release–recapture experiments due to the size of the study area has also been discussed by Wilson and Thomas (2002).

A landscape with a high percentage of good quality habitat was chosen in order to prevent underestimation of movement due to isolation effects of having to cross non-habitat areas rather than a lack of ability to cover larger distances. Knowledge of butterfly movements in a landscape favourable for a butterfly species contributes to understanding the impacts of fragmentation.


Study area

The study area was situated in south-east Sweden (Småland), 20 km west of Oskarhamn and the Baltic Sea (57.3°N, 16.2°E; Fig. 1a). The study area was 172 ha. The principal cover of the study site was agricultural areas with a mean patch size of 0.95 ± 0.08 ha; the remaining parts of the site were buildings with gardens and small forest patches (0.58 ± 0.11 ha). Thus, the different land use types created a fine-grained mosaic (Fig. 1b). The cover of the different land uses is shown in Table 1. In the 1960s, Bråbygden was identified as one of eight areas in Sweden best representing agricultural landscapes of high cultural and natural value (Statens offentliga utredningar, 1971). The study area was largely surrounded by forest.

Figure 1.

(a) Location of the study area in Sweden and (b) land-use map of the study area in Bråbygden (Kalmar län), Sweden in 1999 (thick line = border of study area, thin line = borders of marking areas, dotted line = borders of marking patches within a continuous grassland patch).

Table 1.  Percentage of land use cover in the study area.
Land usePercentage of land use cover
Intensively used grassland23
Houses and gardens9
Roads and lanes1

The study area was mapped on the basis of rectified aerial photographs (orthophotographs, 1:10 000) and field surveys in 1998; changes in land use were adjusted in 1999. The area was then digitised in the geographical information system ArcView 3.1b (ESRI, 1996).

The species

The meadow brown Maniola jurtina L. and the scarce copper Lycaena virgaureae L. were selected for this study because they use the same type of habitat and were abundant enough in the study area for mark–release–recapture studies. The meadow brown is a widespread and common grassland species. The scarce copper is locally common but has declined in several parts of Europe (Ebert, 1993). The larvae of the meadow brown use a wide range of very abundant food plants, feeding on grasses (Poa spp. L. and possibly Milium L.; Henriksen & Kreutzer, 1982). The larvae of the scarce copper feed on Rumex acetosella L. or Rumex acetosa L. (Douwes, 1976).

Field methods

Mark–release–recapture experiments were carried out with both species in the period 5–30 July 1999. Butterflies were marked on 41 patches on extensively used semi-natural grassland, mostly pastures. Patches could be situated within single large grasslands, in adjacent grasslands, or in spatially separated grasslands (Fig. 2).

Figure 2.

Map of the number of captures on the marking patches for Maniola jurtina and Lycaena virgaureae . Maximum column height represents 210 individuals.

Patches ranged from 0.05 to 2.2 ha and covered a total of 24 ha. Patches were selected on the basis of their suitability as habitat for both species at the beginning of the experiment, and represented a range of inter-patch distances.

Butterflies were netted, marked individually by the 1-2-4-7 method of Ehrlich and Davidson (1961) using a fine-tipped permanent ink pen (Faber-Castell Stabilo, Stein, Germany), and released immediately at the place of capture. All 41 patches were visited five times. Due to the large number of patches investigated, it was not possible to survey all patches every day, so a rotation system was used, visiting the same patch every fourth to sixth day (depending on weather conditions). Sampling effect per patch was not the same for all patches but was related to area and butterfly abundance.

Date of capture, location, and sex were recorded for every marked and released butterfly. The flower density of each patch was evaluated in three classes (1: low flower density, 2: medium flower density, 3: high flower density) to estimate adult resource density. This was repeated each time that mark–release–recapture experiments were carried out on a patch (five times). For analysis, the sum of the five flower density evaluations was calculated (Steffan-Dewenter & Tscharntke, 1997). The abundance of the larval food plants of the scarce copper, Rumex acetosa and Rumex acetosella, was surveyed, on a scale from 1 (rare) to 3 (abundant), on all patches in August 2002. The relative abundance of Rumex acetosa and Rumex acetosella was not expected to have changed between 1999 and 2002 within the study patches because grassland management was not modified during this period.


For each individual, the distance between pairs of captures and the total distance moved was estimated using the geographical information system MapInfo 4.5 (MapInfo Corporation, 1998). Inter-patch distances were calculated as straight lines from mid-point to mid-point of the patches. This method was preferred to using edge-to-edge distances, which would have underestimated distances covered by butterflies between adjacent patches.

The distance decay curves were calculated according to the method described by Hill et al. (1996), in which the inverse cumulative proportion of individuals moving certain distances was fitted to a negative exponential and an inverse power function. The fraction of residents of a patch was calculated as the number of residents (R) divided by the sum of R +  E + I, where E was the number of emigrants and I was the number of immigrants. The emigrant fraction was calculated as E/E + R and the immigrant fraction as I/I + R. Exchange rates between pairs of recaptures were calculated according to Sutcliffe and Thomas (1996), in which the exchange rate between a pair of patches is the number of individuals marked in one patch and recaptured in the other patch (movement in both directions is considered) divided by the number of individuals marked in the two patches and recaptured in any other patch, including the considered pair of patches.


Abundance and distribution

Both species were abundant in the study area but the number of meadow browns marked (n = 1243) was almost twice as high as the number of scarce coppers (n = 646). The recapture rate was 24% for the meadow brown and 29% for the scarce copper. The two species showed a different spatial distribution of abundance within the study area (Fig. 2). The number of captures on a patch was correlated with the flower density index (Spearman rank coefficient, meadow brown 0.55, P < 0.001; scarce copper 0.41, P < 0.01). Patch area was not correlated with the number of captures for the scarce copper (Spearman rank coefficient −0.06, P > 0.05) and correlated only weakly for the meadow brown (Spearman rank coefficient 0.38, P < 0.05). There was no correlation between the abundance of larval food plants and the number of scarce copper captures on a patch (Spearman rank coefficient, Rumex acetosa−0.09, P > 0.05; Rumex acetosella−0.20, P > 0.05).


No significant differences could be detected between meadow brown and scarce copper movement in terms of the distance between captures (meadow brown n = 190, mean 322 ± 21 m; scarce copper n = 104, mean 272 ± 24 m; Mann–Whitney z = 1.6, P > 0.05). The frequency of movements for both species is shown in Fig. 3. The maximum recorded distance covered by an individual was 2100 m for the meadow brown and 1460 m for the scarce copper.

Figure 3.

Frequency of movements of (a) Maniola jurtina and (b) Lycaena virgaureae .

The distance decay curves for both the meadow brown and the scarce copper fitted a negative exponential function better than an inverse power function (meadow brown, negative exponential: R2 = 0.87, F1,20 = 130.1, P < 0.001; inverse power function: R2 = 0.68, F1,20 = 42.8, P < 0.001; scarce copper negative exponential: R2 = 0.96, F1,20 = 529.3, P < 0.001; inverse power function: R2 = 0.80, F1,20 = 78.0, P < 0.001).

Significant differences between the species were evident in movement frequencies. While 46% of the meadow brown recaptures took place within the same patch, 59% of the recaptures of the scarce copper were within the same patch (χ2 = 9.59, P < 0.01). Where ≥4 days elapsed between recaptures, 16% of the meadow browns were found on the same patch as before compared with 48% of scarce coppers.

Exchange of individuals between patches

The number and fraction of residents, emigrants, and immigrants of each patch were analysed in relation to flower density and patch area. Only flower density was correlated significantly with the number or fraction of residents, emigrants, and immigrants for either species (Table 2). The abundance of the larval food plants of the scarce copper was not correlated with residency.

Table 2.  Spearman rank correlation between number of captures, residents, emigrants, immigrants, and their fractions with patch area, flower density, and larval food plant abundance.
  Flower FlowerRumexRumex
  1. NS=not significant P>0.05, *P<0.05, **P<0.01, ***P<0.001.

Captures0.38*0.55***−0.06 NS0.41**−0.09 NS−0.20 NS
Number of residents0.14 NS0.36*−0.09 NS0.36*−0.26 NS−0.22 NS
Resident fraction0.08 NS0.10 NS−0.11 NS0.37*−0.16 NS−0.05 NS
Number of emigrants0.21 NS0.41**0.12 NS0.19 NS−0.11 NS−0.21 NS
Emigrant fraction0.01 NS−0.18 NS0.12 NS−0.45**0.18 NS0.05 NS
Number of immigrants0.15 NS0.47**−0.02 NS0.39*−0.16 NS−0.15 NS
Immigrant fraction−0.07 NS−0.05 NS0.17 NS−0.16 NS0.14 NS0.05 NS

Inter-patch distance is an important factor determining exchange rates for butterflies. The exchange rate for both species between pairs of patches was correlated significantly with the distances between the patches (meadow brown, Mantel statistic =−0.36, P= 0.001; scarce copper, Mantel statistic =−0.31, P= 0.001).

The main difference in movement pattern between the two species was that the exchange of individuals was between different patches of the habitat network or at a different frequency between the same patches (Fig. 4).

Figure 4.

Maps of the inter-patch movement of (a) Maniola jurtina and (b) Lycaena virgaureae recorded in Bråbygden in 1999.


Distribution and movement

The two butterfly species used the study area differently, and this is reflected in their spatial distribution, abundance, and movement patterns. The differences in butterfly abundance between patches could be explained by adult resource density and patch area (meadow brown), but not by the abundance of the larval food plants (scarce copper). The large differences in the spatial distribution of the two species could not be explained wholly by the recorded variables, because both species were more abundant on flower-rich patches. The differences are, in addition, believed to be caused by patch management because the scarce copper was particularly abundant on abandoned patches or patches with the lowest intensity of use. This feature of species using the same habitat network showing important differences in patch use has also been shown by, for example, Gutiérrez et al. (2001).

The two butterfly species differed in their frequency of movement between patches but not in terms of distances covered. The higher percentage of scarce coppers caught in the same patch as the previous capture indicates that this species moves between patches less often than the meadow brown. This becomes even more pronounced with the time between recaptures. The fact that > 40% of copper recaptures were on a different patch, however, conflicts with the impression of an immobile species suggested in the literature (Douwes, 1975; Fjellstad, 1998).

The movement behaviour of the two species between patches fitted best a negative-exponential function, which indicates similar movement patterns for both species in the investigated area. The movement frequency and the maximum distance covered by an individual, however, indicate that the meadow brown had a greater movement ability (2.1 km) than the scarce copper (1.5 km). There are two possible explanations for the similar distances moved between the species. The first is that both species have similar movement ability. The second is that the size of the study area (172 ha) was not large enough to expose differences in movement ability between the species.

Scale effect

Several authors have pointed out the influence of the size of the study area on the movement distances recorded (Mousson et al., 1999; Wilson & Thomas, 2002). A comparison with other movement studies of the meadow brown and the scarce copper shows how the size of the study area can influence the recorded mean distances and maximum movement distances (Table 3).

Table 3.  Results of other mark–release–recapture studies of Maniola jurtina and Lycaena virgaureae .
 Size ofNature of theMeanMaximum 
Speciesstudy areasitedistances (m)distances (m)Author (year)
  • Mean distance (m) between recaptures for the two sexes, m=males, f=females.

  • Maximum recorded distance (m) between first and last recapture.

  • §

    Maximum mean distances (m) recorded in one of the three sites.

Maniola jurtina1.4–3ha (3 sites)Mainly meadows66.7 (m), 68.3 (f)§320Brakefield (1982 )
 ≈9ha (1 site)Alpine grassland88.5 (m), 94.2 (f)377Lörtscher etal. (1997 )
 40–110ha (3 sites)Mosaic grassland,
80–180 (m+f)Ouin (2000 )
 300haArable farmland414 (m), 347 (f)1770Dover etal. (1992 )
 172haMosaic of grassland,
arable, and settlement
323 (m), 318 (f)2100This study
Lycaena virgaureae3.4ha (1 site)Mainly grassland55 (m), 59 (f)>1000Douwes (1975 )
 ≈40haAbandoned alpine
58.9 (m), 79.6 (f)1424Fjellstad (1998 )
 172haMosaic of grassland,
arable, and settlement
275 (m), 220 (f)1460This study

In smaller study areas, where mainly within-habitat movements are investigated, few long distance movements are recorded (Douwes, 1975; Brakefield, 1982; Lörtscher et al., 1997; Fjellstad, 1998). In larger study areas, on the other hand, the number of short movements detected is smaller due to longer distances between the mark and release sites (Dover et al., 1992). Both factors increase the recorded mean distances in large study areas. Thus the mean distances recorded for the scarce copper in this study are about four times higher than in the two other studies carried out in smaller study areas. For the meadow brown, the mean distances observed were lower than those found by Dover et al. (1992), who worked in a much larger study area.

Effects of patch characteristics on emigration and immigration

The analysis of the number and fraction of residents, emigrants, and immigrants in relation to patch characteristics revealed the importance of adult resource density, compared with larval food plant abundance and habitat size. Patch size showed no influence on numbers or fractions of residents, emigrants, or immigrants. These results are similar to the findings of Fjellstad (1998), who did not find a relationship between number of immigrants and emigrants of the scarce copper and habitat size. On the other hand, Hill et al. (1996) showed that the emigration rate of Hesperia comma decreased with patch size.

Flower-rich patches appear to be especially important for the scarce copper. The number of residents and immigrants is higher on flower-rich patches, while the fraction of emigrating butterflies is lower. The observation of a decreasing emigration rate with higher flower density on patches has also been made by Kuussaari et al. (1996) for Melitaea cinxia. Higher numbers of meadow browns were caught on flower-rich patches than on patches with fewer nectar sources; however on flower-rich patches, not only is the number of residents and immigrants higher, but also the number of emigrants. The meadow brown seems more inclined than the scarce copper to leave a good patch.

The abundance of the larval food plant of the scarce copper could not be identified as a factor influencing numbers or fractions of residents, emigrants, or immigrants in this study. This is probably related to the fact that the food plants of the scarce copper were abundant and widespread in the study area, so that this factor had little influence on butterfly movement.

Inter-patch distance is a significant factor determining the exchange rates of butterflies between patches in this study. Sutcliffe and Thomas (1996) have also shown patch distance to be a significant explanatory factor for exchange rates between patches. Maps of the inter-patch movements (Fig. 4) show that the two species exchange individuals between different patches of the habitat network. A comparison with the map of butterfly abundance (Fig. 2) indicates that butterfly abundance influences exchange of individuals between patches, because movements occur predominantly between patches with high population density.


Adult resource density was identified as a factor influencing movement of both butterfly species studied, while other patch characteristics such as patch size and larval food plant abundance (scarce copper) showed no influence on movement.

The results of this study also emphasise the need for large-scale mark–release–recapture studies in order to be able to assess a species' movement ability, and highlight the importance of movement studies in comparatively favourable environments as a reference for the impacts of habitat fragmentation. In such study areas, the assumption that sedentary behaviour is linked to the low number and small size of available habitats can be avoided. Movement data gained in larger-scale studies in areas with a high percentage of suitable habitat can contribute important information for conservation management by providing knowledge about the scale at which the spatial ecology of butterflies may operate under optimal conditions.


We thank Chris Thomas and Gabriel Nève for their valuable comments on earlier versions of the paper. We want to thank Carl Dahlberg, Louise Hansare, and Jenny Hedh for their help with the field work and data collection. We are also thankful to Lars G. B. Andersson, who digitised the map of the study area and Vegar Bakkestuen and Jan-Erik Englund for help with the statistics. Christine Schneider was financed by the inter-disciplinary research program ‘The pastoral landscapes’ of the Swedish University of Agricultural Sciences, John Dover and Gary Fry were supported by the Norwegian Institute for Nature Research and the Norwegian Research Council (Grant 121261).