Diversity of Butterflies in the Agricultural

ECOGRAPHY 23: 743–750. Copenhagen 2000

Diversity of butterﬂies in the agricultural landscape: the role of farming system and landscape heterogeneity

Ann-Christin Weibull, Jan Bengtsson and Eva Nohlgren

Weibull, A.-C., Bengtsson, J. and Nohlgren, E. 2000. Diversity of butterﬂies in the agricultural landscape: the role of farming system and landscape heterogeneity. – Ecography 23: 743–750.

To examine the importance of management practices and landscape structure on diversity of butterflies 16 farms with organic or conventional management were censused during 1997 and 1998. On each farm a transect route was walked during July and the beginning of August, six times in 1997 and five times in 1998. The farms were located in the central part of Sweden in two adjacent regions with the same pool of species. The organic and conventional farms were paired with help of the Bray-Curtis dissimilarity index according to land use to control for landscape structure on the farm level. On each farm calculations were made of large- and small-scale landscape heterogeneity with the help of GIS. A grid with a mesh size of 400 m was placed over each farm and the small-scale heterogeneity was calculated as the mean habitat diversity of four squares. The large-scale landscape heterogeneity described the landscape in which the farms were imbedded, and covered an area of 5×5 km. No differences in butterfly diversity, number of species or number of observations were noted between organic and conventional farms. Butterfly diversity was positively correlated with small-scale landscape heterogeneity while butterfly abundance was positively correlated with large-scale heterogeneity. Both large-scale and small- scale heterogeneity were important for the composition of species. The landscape structure seemed to be more important for butterfly diversity and species composition than the farming system in itself.

A.-C. Weibull ([email protected]), J. Bengtsson and E. Nohlgren, Dept of Ecology and Crop Production Science, Swedish Uni6. of Agricultural Sciences, Box 7043, SE-750 07 Uppsala, Sweden.

In present agriculture, the use of pesticides and chemi- of agrochemicals are prohibited, and since several cal fertilizers is a common practice, and the use of groups of organisms, for example butterﬂies (Rands modern machinery has created incitement to take mar- and Sotherton 1986), weeds (Hald and Reddersen ginal zones away. As a consequence the heterogeneity 1990), carabids (Kromp 1990, Moreby et al. 1994) and of the agricultural landscape has decreased as many birds (Braae et al. 1988), have been shown to be non-cultivated habitats, such as ditches, hedgerows and sensitive to pesticides, it has been suggested that the habitat islands have disappeared. It has been suggested absence of agrochemicals on organic farms should be of that the removal of these features, and the use of great importance for species richness. agrochemicals, has caused a reduction in the species However, the effect of organic farming on species richness in the agricultural landscape (Svensson and richness is poorly investigated. Few studies have com- Wigren 1982, 1983, Robertson et al. 1990, Clements et pared diversity under different farming systems and al. 1994, Fuller et al. 1995). On the contrary, organic most of these studies have focused on small parts of the farming has been suggested to enhance diversity (Pao- farm, i.e. local ﬁelds, as the unit for comparison letti 1995, Stopes et al. 1995). On organic farms the use (Dritschilo and Erwin 1982, Hokkanen and Holopainen

ECOGRAPHY 23:6 (2000) 743 1986, Kauppila 1990, Kromp 1990, Hald and Red- relative importance of farming systems and other fac- dersen 1990, Berry and Karlen 1993, Moreby et al. tors, such as landscape composition for conservation of 1994, Drinkwater et al. 1995, Pfiffner and Niggli 1996), species. not on the farm as a whole (Braae et al. 1988). Since Since both farming system and landscape heterogene- recent studies have shown that local species richness ity may influence diversity, it is appropriate to try to often depends on the surrounding landscape (Robert- tease apart the effects of these factors on species rich- son et al. 1990, Marino and Landis 1996, Jonsen and ness in the agricultural landscape. A unique and com- Fahrig 1997), studies of single fields may not capture monly occurring feature of the Swedish agricultural the whole effect of farming system, or landscape hetero- landscape is its mosaic structure. Habitat islands con- geneity, on diversity. sisting of surface bedrock or glacial till, form patches Butterflies are likely to respond to organic farming within the fields, and the proportion of woodland in the (Rands and Sotherton 1986, Feber et al. 1997), and as agricultural areas is relatively high (for illustration see the adults are more or less mobile they are likely to be Bommarco 1998). These features lead to a landscape sensitive to landscape structure. Further, butterflies with varying heterogeneity, which may enhance survival have many qualities that make them suitable as indica- of species (Lo¨rtscher et al. 1997) and promote diversity tor or umbrella species in nature conservation (New (Ryszkowski 1995, Jonsen and Fahrig 1997). This vary- 1997). For example, they are easy to monitor and their ing landscape heterogeneity makes it possible to study biology is rather well understood. Oostermeijer and the effect of landscape structure and farming practices Swaay (1998) suggested that butterflies could be useful at the same time. as indicators of biodiversity. In addition, butterflies In this study the confounding factors of landscape have an important function as pollinators of wild plant heterogeneity at two spatial scales are included in the species (Jennersten 1988), which makes them interesting analysis of the effect of farming system on butterfly from this conservation perspective as well. diversity and abundance. Also, we analyze if butterfly The Swedish government has decided that 20% of the species assemblages differ between organic and conven- arable land should be managed organically by the year tional farms, and whether the characteristics of the 2005, but it is not clear what will be achieved in landscape are more important than the farming system implementing this goal. If diversity is not enhanced by for the species composition and diversity. organic farming practices, we risk losing species that should have been saved with appropriate management applications. Therefore it is of interest to investigate the effects of organic farming on diversity, and evaluate the Materials and methods Study area and pairing of farms The farms in this study were either managed organically or conventionally. The difference between farming systems is that organic farmers do not use pesticides and chemical fertilizers while conventional farmers might use agrochemical more or less frequently. Each organic farm was paired with one conventional farm so that the farms within pairs were as similar as possible in terms of land-use, or habitats, on the farm. All farms in this study were located in two provinces, Uppland and So¨dermanland, in the central part of Sweden (Fig. 1). There are ca 3000 farms in those two provinces. To find matching pairs of organic and conventional farms we calculated Bray-Curtis (BC) dissimilarity indices based on land-use. The BC index is a measure of dissimilarity between two sets of samples (in this case farms); it ranges from 0 (similar) to 1 (dissimilar) (Krebs 1989);

% Xio −Xic BC=% (Xio +Xic)

Fig. 1. The locations of farms in the provinces of Uppland and So¨dermanland, Sweden. Organic farms are deﬁned with where Xio and Xic are the areas of habitat i on organic circles, conventional farms with triangles. (o) and conventional (c) farms respectively. The habi-

744 ECOGRAPHY 23:6 (2000) Table 1. Landscape heterogeneity measurements, total number of species (S) and total number of observations (c) recorded during 1997 and 1998 on organically and conventionally managed farms in the province of Uppland (U) and So¨dermanland (S) in the central part of Sweden.

Pair Farming Region OrganicallyLarge scale Small scaleSize (ha) S 97 c 97 S 98 c 98 system managed since heterogeneity heterogeneity

1O U 1993 0.99 0.70 241.5 18 354 18 225 1 C U 1.01 0.75 214 16 211 16 124 2 O U 1987 1.090.36 106.6 12 368 15 241 2C U 1.21 0.83 137.3 19 310 17 205 3 O U 1989 1.07 1.00 68 17 161 15 187 3 C U 0.730.93 79.5 16 235 17 140 4 O U 1993 0.98 0.93 116.1 16 291 17 174 4 C U 0.93 1.03 132.9 16 206 19 201 5 O U 1992 1.100.84 568.1 15 231 16 131 5 C U 0.98 0.42 664.6 15656 14 269 6 O S 1994 1.32 0.90 312 18 418 20 400 6 C S 1.130.60 318.1 20 449 15 308 7 O S always 1.21 0.99 87.2 23 320 16 263 7 C S 1.27 0.51 80.4 23 576 20 438 8 O S 1982 1.330.95 80.6 23 561 17 234 8 C S 1.03 0.96 77.9 21 361 18 255 tats in this procedure were cereals, peas and beans, closer identification was necessary. When catching a potatoes and beets, oil plants, other crops, lay, forest, butterfly the transect walk was stopped and the count- used pastures and non-used pastures. We used data ing was resumed after restarting the walk. When a from 1995 for this calculation (Anon. 1995). butterfly could not be identified to species level, it was From the 3000 farms in the central part of Sweden, classified to the family level. Since Lycaeidas idas and all organic farms (i.e. organically managed since at least Plebejus argus are difficult to distinguish in field, as are 1992, see Table 1) \20 ha in area were included in the Lasiommata maera and L. petropolitana, all individuals BC procedure, as were all conventional farms within of those species were classified as P. argus and L. maera the same parishes and with approximately the same respectively. Systematics followed Gustavsson (1994). total area as the organic farms. In total, 60 organic and All species recorded during this study were expected to 245 conventional farms were included. With a dissimi- be found in both regions (Henriksen and Kreutzer larity threshold of 0.25 there were 40 possible pairs of 1982). farms recognized. Then pairs in which at least one farm Sampling was carried out from early July to the was B30 ha were excluded, as well as pairs with a beginning of August, and each farm was visited 6 times distance between farms shorter than 3 km or longer in 1997 and 5 times in 1998. Recording was done only than 12 km. This was done because farms very close to under more or less sunny conditions with a minimum each other might share individuals and thus not be temperature of 17°C, and between 9:00 and 16:00 Cen- independent. Very distant farms, on the other hand, tral European Time. may be climatically different. Some farms, and hence pairs, were excluded because of their odd location, e.g. on islands in lake Ma¨laren, or because the farmer wished not to be included in the study. In the end, eight Measuring heterogeneity pairs of farms were chosen for this study (Fig. 1). Landscape heterogeneity was measured on two different spatial scales (Anon. 1998). With the large-scale measurement, the landscape that each farm was imbed- Sampling ded in was classified. The purpose of the small-scale measurement was to estimate the heterogeneity within To record butterflies the line transect method was used each farm. Both large- and small-scale heterogeneity (Pollard 1977). On each farm a permanent transect of were measured using the Shannon-Wiener diversity in- ca 2 km was set up. The placement of the transect was dex, − pi ln pi, where pi is the proportion of habitat i chosen to obtain the best representation of the land-use in the area (square) (Forman 1995). For the large-scale (habitats) on the farm. Each transect was divided into heterogeneity a 5×5-km square was placed over each sections corresponding to changes in habitat. farm and the Shannon-Wiener diversity index was cal- The transect was walked at an approximately uni- culated for this large square. The different land-types form pace (0.8 m s−1) and all butterflies within5mon included in the calculation were arable land, other open both sides and in front of the recorder were noted. The land, mixed forest including clear-cuts, deciduous recorder used a hand-net to catch the butterflies when forest, water and built-up areas. To calculate the small-

ECOGRAPHY 23:6 (2000) 745 scale heterogeneity another square, covering just the factor. A Pearson correlation analysis was made to farm, was placed over each farm. This square had a investigate the impact of landscape heterogeneity on grid with a mesh size of 400-m, and hence the small- butterfly diversity and numbers. The effect of landscape squares were 400×400-m in size. The small-squares heterogeneity on species composition of butterflies was that consisted of \50% forest, built-up areas or water tested using Principal Component Analyses (PCA), ex- were excluded. Then, four small-squares were randomly cluding all species with a mean number of total obser- chosen and the Shannon-Wiener index was calculated vations during both years B1. For the Pearson for each of the small squares. The mean of the four correlation analyses and for the PCA the data from values was used as a measure of the small-scale hetero- both years were summed. All analyses were done with geneity. Because the maps used for calculating the two SAS statistical software (Anon. 1996). indices had different resolution, the measure of small- scale heterogeneity used other land type classes, namely arable land, other open land, mixed forests, clear-cuts, habitat islands, water and built-up areas. Results The large-scale heterogeneity differs between the two Diversity and total abundance provinces, although there is an overlap in the measure. The small-scale heterogeneity does not differ between The number of species recorded on each farm varied regions. between 12 and 23 in 1997, and between 14 and 20 in 1998 (Table 1). The total number of species recorded were 36 in 1997 and 38 in 1998, altogether 42 species over the two years (Appendix). Although all species Statistics were expected to be found in both regions, six species were exclusively found in So¨dermanland while four In addition to number of butterfly species (S), three were exclusively found in Uppland (Appendix). Signifi- different diversity measures were calculated for each cantly more butterflies were observed during 1997 than farm and year. The diversity measurements were the in 1998 (year factor in the split-plot Anova, p=0.012, Shannon-Wiener index (H%=−S p ln p ), Simpson’s i i Table 2), probably because of the warmer and dryer inverted index (1/D=[(n (n −1))/(N(N−1))]) and i i summer in 1997 (Pollard 1988, Pollard and Yates 1993). the Evenness index (J=H%/(log S)) (Magurran 1988). Sixty percent of all observations consisted of the ringlet Differences in diversity, number of species and number Aphantopus hyperantus and the green-veined white of observations between organic and conventional Pieris napi. farms within pairs for each year were tested with split- Only one of the diversity indices showed a tendency plot Anovas, with year as the split factor. In these to be positively correlated with the total number of analyses the effect of landscape is included in the ‘‘pair’’ species, namely H% (Table 3, p=0.0595). However, the different diversity indices were strongly correlated with Table 2. Results from the Split-plot Anovas, with farming systems as treatments and year as split factor. Since there is a each other (Table 3). Neither species richness nor any strong autocorrelation between the different diversity indicies, of the diversity indices differed between farming sys- only H% is shown. tems (Table 2). However, the pairs differed significantly DF F p amongst one another. The pair factor captures both variations in the landscape and regional differences Log S between Uppland and So¨dermanland. The interaction Pair 72.96 * between farming system and pair was also significant Farming system 1 0.28 ns Farming system×Pair7 1.14 ns meaning that for some pairs the organic farms were Year1 1.51 ns found to have a higher diversity than conventional Farming system×Year 1 0.07 ns farms, whilst for other pairs, the reverse was true. In Log c this sense the data is heterogeneous. This heterogeneity Pair 7 8.04 *** Farming system 1 0.09 ns was present also for the total number of observations, Farming system×Pair7 5.01 ** where the interaction between farming system and pair Year1 29.96 *** was again significant, as was the pair factor (Table 2). Farming system×Year 1 0.16 ns The landscape heterogeneity, both the large-scale and H% the small-scale, was important for butterflies. The total Pair7 2.88 * Farming system 1 0.44 ns number of species was positively correlated with small- Farming system×Pair 7 3.77 * scale heterogeneity, as were the three diversity indices Year1 0.63 ns (Table 3, Fig. 2). The total number of observations was Farming system×Year1 0.12 ns positively correlated with large-scale heterogeneity, but *** pB0.001, ** pB0.01, * pB0.05. ns=not significant (p\ negatively correlated with small-scale heterogeneity 0.05). (Table 3).

746 ECOGRAPHY 23:6 (2000) Table 3. Correlation between species number (S), total number of observations (c) and the diversity indices: Shannon-Wiener (H%), Simpson’s (1/D) and Evenness (J). The data are summed over the two years. The ﬁrst two axes from the PCA are also included.

small-scale large-scale log S log c H% 1/D pc1 −0.142 0.626** pc2 0.669** 0.1554 large-scale −0.107 log S 0.522* 0.346 log c −0.501* 0.554* 0.159 H% 0.849*** −0.108 0.481 −0.695** 1/D 0.642** −0.138 0.229 −0.721** 0.902*** J 0.791*** −0.268 0.297 −0.842*** 0.972*** 0.915***

*** pB0.001, ** pB0.01, * pB0.05.

Response of single species and species composition tional farms and suggested that a higher frequency of clover leys on organic farms explained this pattern. In The abundance of single species of butterflies did not our study we had taken this variation in account when differ between organic and conventional farms. How- pairing the farms, which might explain why our results ever, some species were affected by the pair factor. are not consistent with those of Feber et al. (1997). Since the large-scale landscape heterogeneity differs The crop rotation and rearing of animals on organic between regions (one-way Anova, p=0.0084) and the farms can promote a higher landscape heterogeneity on pair factor incorporates a regional effect, we interpret such farms. On the other hand, many organic farms this difference in abundance as being attributable to the have been intensively managed conventional farms in landscape heterogeneity. Different species showed dif- the past. Therefore, the farm-level heterogeneity is not ferent reactions to landscape heterogeneity, some were necessarily higher on organic farms (Drinkwater et al. positively correlated with large-scale heterogeneity 1995). In fact, organic farms can vary just as much as whereas a few species were negatively correlated with conventional farms in the degree of heterogeneity and small-scale heterogeneity (Appendix). presence of non-cultivated areas and habitat islands. The PCA clearly showed the importance of landscape Since we paired the farms according to land-use and the structure for species composition (Fig. 3). The first axis heterogeneity did not differ significantly between farms in the PCA explained 30% of the variation in butterfly within pairs, differences in the landscape structure were assemblage composition and was significantly corre- taken into account when testing for farming system. If lated with large-scale heterogeneity (Table 3), while the there had been an effect of farming system on butterfly second axis explained 18% of the variation and was diversity or abundance, we expected to find it with this correlated with small-scale heterogeneity (Table 3). design. Since the pattern was consistent between years, None of these axes were related to the farming system. we conclude that in a mosaic landscape, butterflies are unlikely to respond significantly to organic farming. On the other hand, Younie and Armstrong (1995) argue that the effects of organic farming on diversity take a Discussion Our results suggest that variation in landscape heterogeneity is more important than the farming system for butterfly diversity and abundance. The general opinion that organic farming enhances diversity is not necessarily true. It is an important conclusion that landscape structure, but not always farming practices, has effects on diversity. Therefore farmers of all kinds should be encouraged to increase the small-scale heterogeneity on their properties to enhance diversity. This can be done with rather simple methods, such as leaving small habitat islands or strips within fields, increasing field margins and open up ditches where possible. Organic farms often have animals and therefore also a certain crop rotation and higher frequency of grazed areas which can be important for species richness. For Fig. 2. The effect of small-scale landscape heterogeneity on example, Feber et al. (1997) found a higher butterfly butterfly species diversity (H%). Organic farms are shown as abundance on organic farms compared with conven- circles and conventional farms as triangles. (n=16).

ECOGRAPHY 23:6 (2000) 747 of our study support this idea, since landscape heterogeneity clearly was shown to be an important factor for butterfly species diversity (Table 1). In our case, the small-scale heterogeneity correlates with species diversity. In general on a more heterogeneous farm the fields are smaller and there are more habitat islands within the fields which can provide shelter. Shelter is known to be important for butterflies (Dover 1996, Dover et al. 1997). To examine if shelter could be important for our results we calculated a shelter index modified from Dover (1996). This shelter index was found to be positively significant correlated with small-scale heterogeneity. Therefore we suggest that it might be the degree of shelter that explains the importance of small-scale heterogeneity on species diversity. Different species clearly have specific demands on their surroundings: some species are restricted to par- ticular habitats (island species) while others are common and widespread and use not only the islands in the landscape but also the matrix (Pollard and Yates 1993). This might explain why the total number of observations was negatively correlated with small-scale heterogeneity but positively correlated with large-scale heterogeneity (Table 3). Since 60% of the observations consisted of the two common and widespread species A. hyperantus and P. napi (Pollard and Yates 1993), Fig. 3. a) A principal component analysis of butterfly species we separated the observations of these species from composition on 8 organic (circles) and 8 conventional (trian- the rest of the observations. The combined abundance gles) farms. PC1 and PC2 together explain 48% of the varia- of A. hyperantus and P. napi alone were negatively tion in the data. PC1 is significantly correlated with large-scale heterogeneity (r=0.626**) and PC2 is significantly correlated correlated with small-scale heterogeneity, while the with small-scale heterogeneity (r=0.669**). (n=16). b) The abundance of all other species was positively corre- scores of the species included in the principal component lated with large-scale heterogeneity. Thus, the pattern analysis. of both positive and negative correlation with the two measurements of landscape heterogeneity is likely to long time to manifest. In our study, the present design depend on the characteristics of single species, for with paired farms did not allow us to examine this example habitat preference (Dal 1978). possibility because of the scarcity of farms that have Butterfly species composition showed the same pat- been under organically management for many years. tern as the species diversity: There were no differences Obviously there are other factors than farming sys- between organic and conventional farms, but there tem influencing diversity. For example, the presence was a difference in species composition in relation to and distribution of suitable habitats might be impor- landscape heterogeneity (Fig. 3). Here, both large- and tant for the occurrence of species (Thomas et al. small-scale heterogeneity were of importance. This is 1992), especially for those using the non-arable land in of interest from a conservation point of view. The the agricultural landscape (Mineau and McLaughlin structure of the landscape on a larger scale is an 1996). Since different species are adapted to different important factor for diversity (Robertson et al. 1990) habitats all species will not be more abundant, or even but also uncultivated habitats are of importance for found in, an organic agricultural system (Armstrong enhancing diversity (Clausen et al. 1998, Fry and and McKinley 1995). This was the case in this study Robson 1994). (Appendix). In conclusion, our study shows that other factors in Landscape characteristics have been shown to be the agricultural landscape than the farming system per important for species condition (Bommarco 1998, se seem to be of importance for butterfly diversity. O8stman et al. 2000) as well as for species richness for Butterflies did not respond to organic farming in the different groups of organisms, for example bumble- mosaic landscape of central Sweden, while landscape bees (Svensson et al. 2000), birds (Robertson et al. heterogeneity could explain much of the variation in 1990), parasitoids (Marino and Landis 1996), and in- diversity and species composition. This is important sect herbivores (Jonsen and Fahrig 1997). The results for species conservation in arable land.

748 ECOGRAPHY 23:6 (2000) Acknowledgements – We thank La¨nstyrelsen in Uppsala and Hald, A. B. and Reddersen, J. 1990. Fugleføde i kornmarker – Nyko¨ping and SCB in O8rebro for kindly providing us with insekter og vilde planter. – Miljøprojekt nr. 125, information about the farms. Peter Saetre wrote the program Miljøstyrelsen, København, in Danish. for the BC analyses. Mikael Ekman and Stefan Eriksson Henriksen, H. J. and Kreutzer, I. B. 1982. The butterflies of skillfully helped with the fieldwork. Lotta Hansson, Mats W. Scandinavia in nature. – Skandinavisk Bokforlag, Odense. Pettersson and Torbjo¨rn Ebenhard have provided valuable Hokkanen, H. and Holopainen, J. K. 1986. Carabid species comments and criticism during different stages of this work. and activity densities in biologically and conventionally We thank the farmers for their cooperation and the interest managed cabbage fields. – J. Appl. Entomol. 102: 353– that they have shown in our study. Lammska stiftelsen and 363. SJFR financially supported this study. Jennersten, O. 1988. Pollination in Dianthus deltoides (Caryophyllaceae): effects of habitat fragmentation on visi- tation and seed set. – Conserv. Biol. 2: 359–366. Jonsen, I. D. and Fahrig, L. 1997. Response of generalist and specialist insect herbivores to landscape spatial structure. – Landscape Ecol. 12: 185–197. References Kauppila, R. 1990. Conventional and organic cropping systems at situa. IV: weeds. – J. Agricult. Sci. in Finland 62: Anon. 1995. SCB – Utdrag ur svenskt lantbruksregister. – 331–337. Swedish Bureau of Statistics. Krebs, C. J. 1989. Ecological methodology. – Harper and Anon. 1996. SAS Inst. – Cary, N.C., USA. Row. Anon. 1998. MapInfo Professional 4.5.2. – Troy, New York, Kromp, B. 1990. Carabid beetles (Coleoptera, Carabidae) as USA. bioindicators in biological and conventional farming in Armstrong, G. and McKinley, R. G. 1995. The effect of Austrian potato fields. – Biol. Fertil. Soils 9: 182–187. intercropping cabbages with clover on pitfall catches of Lo¨rtscher, M., Erhardrt, A. and Zettel, J. 1997. Local move- carabid beetles: a study in invertebrate diversity. – In: ment patterns of three common grassland butterflies in a Isart, J. and Llerena, J. J. (ed.), Biodiversity and land use: traditionally managed landscape. – Bull. Soc. Entomol. the role of organic farming. First ENOF-Workshop, Bonn, Suisse 70: 43–55. pp. 23–33. Magurran, A. E. 1988. Ecological diversity and its measure- Berry, E. C. and Karlen, D. L. 1993. Comparison of alternative farming systems. II. Earthworm population density ment. – Cambridge Univ. Press. and species diversity. – Am. J. Alternative Agricult. 8: Marino, P. C. and Landis, D. A. 1996. Effect of landscape 21–26. structure on parasitoid diversity and parasitism in agroe- Bommarco, R. 1998. Reproduction and energy reserves of a cosystems. – Ecol. Appl. 6: 276–284. predatory carabid beetle relative to agroecosystem com- Mineau, P. and McLaughlin, A. 1996. Conservation of biodi- plexity. – Ecol. Appl. 8: 846–853. versity within Canadian agricultural landscapes: integrating Braae, L., Nøhr, H. and Petersen, B. S. 1988. Fuglefaunaen pa˚ habitat for wildlife. – J. Agricult. Environ. Ethics 9: 93– konventionelle og økologiske landbrug. – Miljøprojekt nr. 113. 102, Miljøstyrelsen, København, in Danish. Moreby, S. J. et al. 1994. A comparison of the flora and Clausen, H. D., Holbeck, H. B. and Reddersen, J. 1998. arthropod fauna of organically and conventionally grown Butterflies on organic farmland: associations to uncropped winter wheat in southern England. – Ann. Appl. Biol. 125: small biotopes and their nectar sources. – Entomol. Medd. 13–27. 66: 33–44. New, T. R. 1997. Are Lepidoptera an effective ‘‘umbrella Clements, D. R., Weise, S. F. and Swanton, C. J. 1994. group’’ for biodiversity conservation? – J. Insect Conserv. Integrated weed management and weed species diversity. – 1: 5–12. Phytoprotection 75: 1–18. Oostermeijer, J. G. B. and Swaay, C. A. M. 1998. The Dal, B. 1978. Fja¨rilar i naturen. Europas dagfja¨rilar: 1 relationship between butterflies and environmental indica- Nordeuropa. – Stockholm, in Swedish. tor values: a tool for conservation in a changing landscape. – Biol. Conserv. 86: 271–280. Dover, J. W. 1996. Factors affecting the distribution of satyrid 8 8 butterflies on arable land. – J. Appl. Ecol. 33: 723–734. Ostman, O. et al. 2000. Landscape complexity and farming Dover, J. W., Sparks, T. H. and Greatorex-Davis, J. N. 1997. practice influence the condition of polyphagous carabid The importance of shelter for butterflies in open land- beetles. – Ecol. Appl. in press. scapes. – J. Insect Conserv. 1: 89–97. Paoletti, M. G. 1995. Biodiversity, traditional landscapes and Drinkwater, L. E., Letourneau, D. K. and Shennan, C. 1995. agroecosystem management. – Landscape Urban Plann. Fundamental differences between conventional and organic 31: 117–128. tomato agroecosystems in California. – Ecol. Appl. 5: Pfiffner, L. and Niggli, U. 1996. Effects of biodynamic, or- 1098–1112. ganic and conventional farming on ground beetles (Col. Dritschilo, W. and Erwin, T. L. 1982. Responses in abundance Carabidae) and other epigaeic arthropodes in winter and diversity of cornfield carabid communities to differ- wheat. – Biol. Agricult. Hort. 12: 353–364. ences in farm practices. – Ecology 63: 900–904. Pollard, E. 1977. A method for assessing changes in the Feber, R. E. et al. 1997. The effect of organic farming on pest abundance of butterflies. – Biol. Conserv. 12: 115–134. and non-pest butterfly abundance. – Agricult. Ecosyst. Pollard, E. 1988. Temperature, rainfall and butterfly numbers. Environ. 64: 133–139. – J. Appl. Ecol. 25: 819–828. Forman, R. T. T. 1995. Land mosaics. The ecology of land- Pollard, E. and Yates, T. J. 1993. Monitoring butterflies for scapes and regions. – Cambridge Univ. Press. ecology and conservation. – Chapman and Hall. Fry, G. L. A. and Robson, W. J. 1994. The effects of field Rands, M. R. W. and Sotherton, N. W. 1986. Pesticide use margins on butterfly movement. – In: Boatman, N. D. and cereal crops and the changes in the abundance of (ed.), Field margins: integrating agriculture and conserva- butterflies on arable farmland in England. – Biol. Conserv. tion. British crop protection council (BCPC) Monogr. No. 36: 71–82. 58, BCPC; Farnham, U.K., pp. 111–159. Robertson, G. M., Eknert, B. and Ihse, M. 1990. Habitat Fuller, R. J. et al. 1995. Population declines and range con- analysis from infrared aerial photographs and the conser- tractions among lowland farmland birds in Britain. – vation of birds in Swedish agricultural landscapes. – Am- Conserv. Biol. 9: 1425–1441. bio 19: 195–203. Gustavsson, B. (ed.) 1994. Catalogus Lepidopterorium Sue- Ryszkowski, L. 1995. Managing ecosystem services in agricul- ciae. – Swedish Museum of Natural History, Stockholm. tural landscapes. – Nature and Resources 31: 27–36.

ECOGRAPHY 23:6 (2000) 749 Stopes, C. et al. 1995. Hedgerow management in organic Svensson, B., Lagerlo¨f, J. and Svensson, B. G. 2000. Habitat farming-impact on diversity. – In: Isart, J. and Llerena, preferences of nest seeking bumblebees (Hymenoptera: J. J. (ed.), Biodiversity and land use: the role of orga- Apidae) in an agricultural landscape. – Agricult. Ecosyst. nic farming. First ENOF-Workshop, Bonn, pp. 121– Environ. 77: 247–255. 125. Thomas, C. D., Thomas, J. A. and Warren, M. S. 1992. Svensson, R. and Wigren, M. 1982. Na˚gra ogra¨sarters till- Distribution of occupied and vacant butterﬂy habitats in bakaga˚ng belyst genom konkurrens-, go¨dslings- och her- fragmented landscapes. – Oecologia 92: 563–567. bicidfo¨rso¨k. – Svensk Bot. Tidskr. 76: 241–258, in Younie, D. and Armstrong, G. 1995. Botanical and inverte- Swedish. brate diversity in organic and intensively fertilized grass- Svensson, R. and Wigren, M. 1983. A,kerogra¨sﬂoran 1958 land. – In: Isart, J. and Llerena, J. J. (ed.), Biodiversity och 1980 i na˚gra va¨stska˚nska socknar. – Svensk Bot. and land use: the role of organic farming. First ENOF- Tidskr. 77: 241–257, in Swedish. Workshop, Bonn, pp. 35–44.

Appendix. Species list and numbers of butterlies found on each farm during 1997 and 1998

Pair number 1 12233445566778 8 Treatment O C O C O C O C O C OCOCO C

Pieris brassicae 28197 3 18 214 17 31 4 Pieris rapae 2131671 110 2 Pieris napi* 223 106 149 130 59 23 43 82 111 369 8459 19 242 115 75 O,S Colias palaeno 1 Gonepteryx rhamni 3 132 5251429 12 9 13 21 17 Leptidea sinapis 1 1 C Limenitis populi 1 1 Inachis io 2 12 5 6 18 12 Vanessa atalanta 11 Cynthia cardui 3 8351318958 7 1 2 4 1 Aglais urticae*96171067 27 729 3 21 1 13 6 21 O Polygonia c-album 2 2 Argynnis paphia 221 1 14 2 14 Mesoacidalia aglaja 4 19 6 14 5 2 1244133 O,S Fabriciana adippe 1 2 O,U Issoria lathonia 1 Brenthis ino 14 1 3 10 18 17 8 3 3 45236 Clossiana selene 11362234 1 C,U Clossiana euphrosyne 14 Mellicta athalia 22121218524269 Erebia ligea 18 46 8 26 13 4 11 5 22 1 3 3 O,S Maniola jurtina 5 Aphantopus hyperantus* 101 41 305 146 87 153 198 99 67 388 406429 214 436 287 261 Coenonympha pamphilus 222429 1 18 14 28 4 3 78 Coenonympha arcania* 271825433536344015175625 75 78 31 18 Lasiommata maera/petropolitana*1223241691655321312910 5 U Satyrium w-album 11 C Lycaena phlaeas 1 1 2 Lycaena 6irgaureae 11 216311121 1 C Lycaena hippothoe 11 3 1 C,U Glaucopsyche alexis 1 S Plebejus argus/idas 4 1 O,S Vacciinina optilete 2 Aricia artaxerxes 11211 1 Cyaniris semiargus 3323 9583261 5 11 2 2 Polyommatus amanda 42031113 7 4 11 8 1 2 8 16 15 6 Polyommatus icarus 27 15 9 34 21171811 12 2 21 3 4 1 Thymelicus lineola* 552841152313122346376195 63 52 90 111 Ochlodes 6enatus*76251 6 17 11 2 5 6 8 17 22 18 9 8 Zygaena ﬁlipendulae 3 231 C Zygaena 6iciae 1 1 O Zygaena lonicerae 1 12

* Species occurring in all farms, O or C species occuring on only organic or conventional farms respectively, U or S species occurring only in the region of Uppland or So¨dermanland respectively.

750 ECOGRAPHY 23:6 (2000)