The effect of organic farming on butterfly diversity depends on landscape context


Maj Rundlöf, Department of Ecology, Animal Ecology, Lund University, SE-223 62 Lund, Sweden (fax + 46 46 222 47 16; e-mail


  • 1The recent dramatic decline in farmland biodiversity is often attributed to agricultural intensification and structural changes in the agricultural landscape. One suggested farm practice seen to benefit biodiversity and reverse declines is organic farming. Because organic farming is viewed as a more sustainable form of agriculture it is currently subsidized by European agri-environment schemes. However, the efficiency of agri-environment schemes to preserve biodiversity has recently been questioned, partly because their uptake has been highest in extensively farmed more heterogeneous landscapes.
  • 2We investigated the effect of farming practice on butterfly species richness and abundance along cereal field headlands and margins on 12 matched pairs of organic and conventional farms in contrasting landscapes (homogeneous and heterogeneous landscape diversity).
  • 3Both organic farming and landscape heterogeneity significantly increased butterfly species richness and abundance. There was also a significant interaction between farming practice and landscape heterogeneity, because organic farming only significantly increased butterfly species richness and abundance in homogeneous rather than heterogeneous landscapes.
  • 4An analysis of the distribution of organic farming in Sweden in relation to productivity of the arable land (yield of spring barley, kg ha−1) indicated that the distribution of organic farms was skewed towards extensively farmed agricultural areas.
  • 5Synthesis and applications. The species richness and abundance of butterflies can be enhanced by actions aimed at both promoting organic farming and increasing landscape heterogeneity. However, the beneficial effect of organic farming was only evident in intensively farmed homogeneous landscapes. Currently, the majority of organic arable land in Sweden is located in heterogeneous landscapes where changing the type of farming practice adds little to the existing biodiversity. We therefore propose that the interaction between landscape heterogeneity and farming practice must be considered when promoting farmland biodiversity, for example in Europe by developing context-based agri-environment schemes to increase the amount of organic farming in intensively farmed landscapes. We also propose that in homogeneous agricultural landscapes, organic farming could be used as a more efficient tool to restore landscape heterogeneity if the creation of semi-natural landscape elements was mandatory in the regulations associated with organic agri-environment schemes.


Considerable research has been dedicated towards identifying causes of biodiversity decline in agricultural landscapes and ways to counteract it (Dover 1997; Krebs et al. 1999; Wilson et al. 1999; Tilman et al. 2001; Benton, Vickery & Wilson 2003; Fox 2004). Organic farming has been suggested as one solution (Hole et al. 2005) because an explicit goal of organic farming is to sustain and enhance biological diversity in the agricultural landscape (CAC 1999; IFOAM 2005). This assumption may be one important reason for the continued growth of the organic sector. The goal to protect biodiversity together with the view of organic farming as a more sustainable form of agriculture are factors affecting both consumers’ food choice (Sheperd, Magnusson & Sjödén 2005) and policy makers. In Europe this has led to the use of agri-environment schemes (financial support to farmers to compensate for taking positive environmental actions) to subsidize organic production.

The organic sector has increased in most European countries, including Sweden, since the beginning of the 1990s. Around 3·5% (4·8 million ha) of the arable land in the European Union (EU) was under organic management in 2002 (Willer & Richter 2004). However, Kleijn & Sutherland (2003) have highlighted that the uptake of different types of agri-environment schemes has been highest in areas with extensive agriculture and hence fewer concerns regarding declining biodiversity. In addition, the efficiency of agri-environment schemes in general and organic farming in particular to protect biodiversity has recently been questioned (Trewavas 1999; Kleijn et al. 2001; Kleijn & Sutherland 2003; Green et al. 2005; Hole et al. 2005).

Comparative studies of biodiversity on organic and conventional arable land generally report higher species richness and abundance in organic systems, but the results are variable, with some neutral or even negative effects being found in individual studies (Feber et al. 1997; Hutton & Giller 2003; Wickramasinghe et al. 2003; Bengtsson, Ahnström & Weibull 2005; Fuller et al. 2005; Hole et al. 2005; Roschewitz et al. 2005; Schmidt et al. 2005). The mixed results are likely to be partly the result of inadequate study design and differences in the spatial scale at which the studies were conducted (Kleijn et al. 2004; Bengtsson, Ahnström & Weibull 2005). In addition, the effects of agri-environment schemes may be dependent on various landscape characteristics and these may not have been taken into account in the interpretation of some of the previous studies (Kleijn et al. 2004; Bengtsson, Ahnström & Weibull 2005; Tscharntke et al. 2005). The conclusion from the meta-analysis of Bengtsson, Ahnström & Weibull (2005) is that there is a pressing need for carefully designed farm-scale studies of the effect of organic farming on biodiversity that takes the landscape context into account.

Landscape heterogeneity can be a key factor in promoting biodiversity in the agricultural landscape (Weibull, Bengtsson & Nohlgren 2000; Benton, Vickery & Wilson 2003; Tscharntke et al. 2005). Heterogeneous landscapes may benefit population persistence through several different mechanisms, for example by providing habitats for a diversity of need, by buffering temporal variation in food supply and providing refugee sites in a landscape characterized by ephemeral habitats (Benton, Vickery & Wilson 2003). Some studies have found the effects of landscape heterogeneity on biodiversity to be more important than farming practice (Weibull, Bengtsson & Nohlgren 2000; Purtauf et al. 2005). It has even been suggested that the reason why organic farming affects biodiversity may be because it restores farmland heterogeneity rather than any effects associated with the lack of use of agrochemicals (Benton, Vickery & Wilson 2003). Consequently, the lack of any observable effect of organic farming on biodiversity could in some cases be attributed to landscape heterogeneity having a stronger influence (Weibull, Bengtsson & Nohlgren 2000; Bengtsson, Ahnström & Weibull 2005; Tscharntke et al. 2005).

In this study we evaluated the consequences of both organic farming and landscape heterogeneity on the species richness and abundance of butterflies on farmland in southern Sweden. By using a design in which we compared the consequences of farming practice in replicated landscapes of different complexity, we considered in particular whether the consequence of organic farming depends on the structure of the landscape. We also investigated the occurrence of organic farming in relation to agricultural productivity, to assess whether the distribution depends on farming intensity.

Materials and methods

To evaluate whether the effects of organic farming on biodiversity are conditional on the landscape context, we established a study system in southern Sweden consisting of 12 independent pairs of farms located in two contrasting landscapes. Each pair of farms consisted of one organic and one conventional farm (Fig. 1).

Figure 1.

Location of the study sites in the province of Scania, southern Sweden: organic (open symbols) and conventional (filled symbols) farms in the two landscape types (heterogeneous, triangles; homogeneous, squares).

landscape definition

The two contrasting landscape types were characterized as heterogeneous mixed farmland and intensively farmed homogeneous plain, respectively. As in Roschewitz et al. (2005) and Purtauf et al. (2005), we used the proportion of arable land (i.e. annually tilled fields with annual crops) and the occurrence of grasslands (field borders and pastures) in the landscape as proxies for habitat-type diversity and landscape type. The three landscape measures (proportion of arable land, mean field size and proportion of pasture of all agricultural land) were calculated in ArcView GIS 3·2 using agricultural statistics from the Swedish Board of Agriculture (Jönköping) within a 1-km radius circle from the centre of the farm unit. Six pairs of farms were situated in heterogeneous landscapes [proportion of arable land 0·15 ± 0·038 (mean ± SEM); field size 31600 ± 4600 m2; proportion of pasture 0·19 ± 0·031] and the remaining six pairs in homogeneous landscapes (proportion of arable land 0·70 ± 0·049; field size 60200 ± 6400 m2; proportion of pasture 0·028 ± 0·011). The heterogeneous landscapes held a significantly lower proportion of arable land (t = −8·83, d.f. = 22, P < 0·0001), smaller fields (t = −3·61, d.f. = 22, P= 0·0016) and a significantly larger proportion of pasture (t = 4·92, d.f. = 13·6, P= 0·0002) than the homogeneous landscapes.

farm selection

Spatially explicit information on agricultural land use (from the Swedish Board of Agriculture) was used in ArcView GIS 3·2 to select organic farms and match them with conventional ones. Organic farms were identified through the occurrence of EU-subsidized organically managed fields (according to EEC regulation 2092/91). The initial selection of organic farms focused on farms that had at least 50% of their land area under organic management during 2002. Farms with vegetable crops, fruit cultivations, etc., were then excluded; the dominant crops of the remaining organic farms were therefore cereal, cereal–legume mixture and ley, and more rarely beet and oil seed rape. We included 12 organic farms [area of organic arable land per farm 72 ± 13 ha (mean ± SEM); proportion of organic agricultural land per farm 0·94 ± 0·036] in the study, i.e. six from each of the two landscape types. We used proximity, field size, crop type and occurrence of landscape features (e.g. stone walls, tree rows and small habitat islands) to match carefully, as far as possible, the selected 12 organic farms with conventional ones, to eliminate all differences not directly linked to farming practice (i.e. differences other than use/non-use of pesticides, use/non-use of inorganic fertilizers and a long and diverse crop rotation; Stockdale et al. 2001). The farms within a pair were situated close enough (3–8 km within pairs) to occur in the same type of landscape but were always at least 3 km apart, to avoid systematic exchange of butterflies.

butterfly inventory

We surveyed butterflies (Rhopalocera, including burnet moths, Zygaenidae) along organically and conventionally managed cereal field headlands and margins during June–August 2003 (five visits) and May–August 2004 (six visits) on the selected 24 farms. The farms within a pair were visited on the same day and each farm was visited every 2 weeks. Butterfly species and abundance were recorded in the field headlands (outer 5 m of the cereal fields) and adjacent uncultivated field margins (2-m wide strip of grassy vegetation along the field boundaries) using a modified standard transect count (Pollard 1977). The butterfly data from the two subsampling areas (field headland and uncultivated field margin) were combined in the statistical analyses.

The location of survey transects was established at the beginning of the first survey season. The locations of some transects changed between 2003 and 2004 because of crop rotation on the studied farms. The type of field margins was standardized and matched within farm pairs and transects were placed in two to three cereal fields per farm along the border between the field headland and field margin. Each farm had a total transect length of 400–750 m (matched within pairs of farms) and transects were divided into segments of 50 m using a Silva pedometer Distans (Silva Sweden AB, Box 998, SE-191 29 Sollentuna).

agricultural statistics

Agricultural statistics in Sweden (e.g. yield per hectare, total production of different crops and agricultural financial support) are calculated in eight large production districts (SCB 2004). These production districts are based on so-called natural agricultural areas, which are defined by the conditions (e.g. type of soil and bedrock, topography and climate) that constitute the prerequisites for agricultural activity. Data on the area of arable land, organic arable land and yield of spring barley for the eight large production districts in Sweden that were used in this study were modified from the Yearbook of Agricultural Statistics 2004 (SCB 2004). Yield of spring barley (kg ha−1) was selected as a proxy for productivity of the agricultural land because data were available for all eight production districts and spring barley, together with autumn wheat, accounts for the largest area of cereal cultivation in Sweden (SCB 2004).

statistical analysis

We used generalized linear mixed models (SAS macro Glimmix) with Poisson error distribution and log-link function to analyse the effects of farm practice and landscape type on butterfly species richness. The butterfly abundance (individuals per 50 m of transect) was log-transformed [ln(x + 0·1)] to achieve residual normal distribution, and analysed using general mixed models with normal error distribution. Data were analysed at the segment level to account for the slightly unequal sampling effort at different farms (all results were qualitatively the same if analysed at transect level). The fixed factors in the models were year, landscape type, farm practice and landscape type × farm practice; the random factors were farm pair and farm identity; the repeated factor was visit (nested within year). We selected the covariance structure for the repeated factor based on AIC (Akaike Information Criterion), which in all cases resulted in a first-order autoregressive structure being used. We used the Satterthwaite method (Littell et al. 1996) to approximate denominator degrees of freedom. Pearson correlation was used to assess the association between proportion of organic arable land and productivity of the arable land (yield of spring barley; kg ha−1). All statistical analyses were performed in SAS 9·1 for Windows.


effects of farm practice and landscape context

Both species richness and abundance of butterflies were significantly larger if the farm was under organic management (species richness, F1,9·1 = 40·37, P < 0·001; Fig. 2a; abundance, F1,10·6 = 26·10, P= 0·0004; Fig. 2b) and was significantly higher in the more heterogeneous landscapes (species richness, F1,9·3 = 16·89, P= 0·0025; Fig. 2a; abundance, F1,10 = 17·73, P= 0·0018; Fig. 2b). However, there was a significant interaction between the effects of landscape type and farming practice on species richness (F1,9·1 = 20·78, P= 0·0013; Fig. 2a). Farming practice significantly affected butterfly species richness in the homogeneous (F1,5·1 = 27·10, P= 0·0033) but not in the heterogeneous landscapes (F1,6·5 = 3·21, P = 0·12). There was also a significant interaction between the effects of landscape type and farming practice on butterfly abundance (F1,10·6 = 9·30, P= 0·012; Fig. 2b). A significant effect of farming practice on butterfly abundance was observed in the homogeneous (F1,5·1 = 25·72, P= 0·0037) but not in the heterogeneous (F1,5·5 = 3·01, P= 0·14) landscapes.

Figure 2.

Number of butterfly species and butterfly abundance in relation to farming practice (organic, open bars; conventional, filled bars) and landscape context. (a) Mean number of butterfly species per 50-m transect, visit and year. (b) Mean butterfly abundance (individuals per 50-m transect) per visit and year. Error bars show SEM and n equals six farms per regime and landscape type.

distribution of organic farms

The proportion of organic arable land of all arable land significantly decreased with increasing yield of spring barley (r = −0·86, n= 8, P= 0·0060; Fig. 3) and hence organic farming in Sweden was taken to be negatively associated with productivity of the arable land.

Figure 3.

Proportion of organic arable land in relation to productivity (yield of spring barley) in the eight large agricultural production districts in Sweden (squares, plain district; triangles, central district; circles, forest district; diamonds, northern Sweden) (SCB 2004). The map of Sweden illustrates the locations of the eight production districts (1–8) and numbers in the graph symbols correspond to numbers on the map.


There are several differences between organic and conventional farms that could help explain the observed effects on butterfly species richness and abundance. Because we carefully matched farms, differences are most likely to have been caused by the underlying farm management practices. European organic crop production is mainly differentiated from conventional farming by the prohibition of pesticides and inorganic fertilizers (EEC regulation 2092/91) and this might partly explain the increase in butterfly diversity and abundance on organic farms. The application of pesticides leads to direct mortality and sublethal effects on fecundity and longevity in butterflies, while herbicides remove nectar resources and larval host-plants both in the field and in the edge zone next to the field (Davis et al. 1991; Dover 1997; Longley & Sotherton 1997; de Snoo 1999; Haughton et al. 2003). The alternative methods used in organic farming, for example a long and diverse crop rotation scheme with a high proportion of ley and nitrogen-fixating crops (Stockdale et al. 2001), might in itself promote butterfly diversity and abundance by increasing the amount of temporary food sources. Thus organic farming, with its exclusion of pesticides and longer crop rotation, may on a landscape-scale increase habitat heterogeneity (Benton, Vickery & Wilson 2003).

Our results, together with an increasing body of scientific work (Benton, Vickery & Wilson 2003; Weibull & Östman 2003; Purtauf et al. 2005; Schmidt et al. 2005; Tscharntke et al. 2005), indicate that landscape heterogeneity is a key factor in promoting biodiversity in the agricultural landscape. A mosaic landscape may support a larger number of species in a given area, simply because the landscape contains a larger number of habitats. However, several additional mechanisms may also promote biodiversity in the separate habitats. In our case, the effect of landscape type could be related to differences in the availability of semi-natural grasslands in the form of pastures and uncultivated field boundaries. Such areas may be vital for insects in the agricultural landscape and act as permanent habitat in a landscape characterized by different forms of temporary habitats (Tscharntke et al. 2002; Weibull & Östman 2003; Purtauf et al. 2005; Tscharntke et al. 2005). Heterogeneous landscapes may also support species that use different habitats for different needs or during different life stages (Ouin et al. 2004). For example, butterflies require larval host-plants and adult nectar resources for population persistence (Cowley et al. 2001; Thomas et al. 2001; Boggs 2003; Pywell et al. 2004; Tudor et al. 2004).

Although we can conclude that both farming practice and landscape heterogeneity significantly affects butterfly species richness and abundance, the most interesting result is the effect of the interaction between the two. A similar relationship has been proposed for arable weeds (Roschewitz et al. 2005). Such interactions can explain why earlier comparisons between organic and conventional systems have showed divergent results, as the results will depend on the landscape context. There are at least two possible explanations for this pattern. First, the difference between organic and conventional farming practice is larger in more homogeneous landscapes, for example the use of pesticides (kg ha−1) in conventional farming is higher in more intensively farmed agricultural areas (SCB 2004; but see Roschewitz, Thies & Tscharntke 2005). Secondly, organic farming may recreate some of the landscape heterogeneity in the more intensively farmed agricultural plains by increasing the spatial and temporal occurrence of host and nectar plants. The relative importance of habitat heterogeneity and farming practice may vary both in relation to the spatial configuration of organic farming and with the spatial scale used by different organisms (Fuller et al. 2005; Tscharntke et al. 2005). If our results also generalize to the diversity and abundance of other groups of animals and plants, the effect of organic farming on biodiversity may not be as straightforward as previously thought. For example, the subsidizing of organic farming in Europe has increased the organic sector in most European countries (Willer & Richter 2004) but the uptake of agri-environment schemes has been highest in areas with extensive agriculture (Kleijn & Sutherland 2003). In Sweden, our analyses show that the distribution of organic farms is skewed towards agricultural districts with more extensive farming activity that, as a consequence, are likely to contain more heterogeneous agricultural landscapes (Roschewitz, Thies & Tscharntke 2005). Given our results, the biodiversity gains of organic farming would in fact be largest if farms in areas where biodiversity is threatened by agricultural intensification were converted to organic management, because these are the areas where biodiversity currently is low and where the effect of organic farming on biodiversity is high. Our results also demonstrate a clear effect of landscape heterogeneity on butterfly diversity. Therefore any agri-environment scheme that supports the persistence of farms threatened by marginalization in heterogeneous landscapes may benefit overall farmland biodiversity. However, this is then not a direct effect of organic farming per se but an indirect effect of supporting the persistence of the types of agricultural activity that produce heterogeneous landscapes.

It has been argued that the long-term sustainability of farmland ecosystems depends on the conservation of biodiversity (Tscharntke et al. 2005). If this is the case, the interactive effect of management and landscape heterogeneity on biodiversity should be taken into account when promoting organic farming. The promotion of organic farming and/or landscape heterogeneity therefore has the potential to increase biodiversity significantly in homogeneous landscapes. In Europe this may require considerable changes in the way agri-environment schemes are designed and implemented, because currently conversion to organic farming is mainly occurring in heterogeneous landscapes. We propose that future agri-environment schemes should be context-based. For example, organic farming in intensively farmed landscapes could be promoted if the financial support to organic farming was related to predicted yield loss instead of land area. Furthermore, in homogeneous agricultural landscapes organic farming could be used as a more efficient tool to restore landscape heterogeneity if the creation and management of semi-natural habitat elements was mandatory in the regulations, as is the case in the newly implemented environmental stewardship agri-environment scheme in the UK (DEFRA 2005). For example, requiring an increased amount and width of uncultivated field margins could result in creation of attractive habitat for pollinators (Kells, Holland & Goulson 2001).


We thank the farmers who allowed us to work on their land, U. Sandnes and H. Nilsson for assistance with the butterfly inventory and J. Bengtsson, E. Öckinger, L. Wickramasinghe and one anonymous referee for valuable comments on earlier drafts of the manuscript. The work was financially supported by the Swedish Research Council for Environment, Agricultural Sciences and Spatial Planning (FORMAS) research theme ‘The landscape ecology of organic farming’ and H. G. Smith was supported by a grant from FORMAS.