Mixed effects of organic farming and landscape complexity on farmland biodiversity and biological control potential across Europe


Correspondence author. E-mail: camilla.winqvist@slu.se


1. Organic farming in Europe has been shown to enhance biodiversity locally, but potential interactions with the surrounding landscape and the potential effects on ecosystem services are less well known.

2. In cereal fields on 153 farms in five European regions, we examined how the species richness and abundance of wild plants, ground beetles and breeding birds, and the biological control potential of the area, were affected by organic and conventional farming, and how these effects were modified by landscape complexity (percentage of arable crops within 1000 m of the study plots). Information on biodiversity was gathered from vegetation plots, pitfall traps and by bird territory mapping. The biological control potential was measured as the percentage of glued, live aphids removed from plastic labels exposed in cereal fields for 24 h.

3. Predation on aphids was highest in organic fields in complex landscapes, and declined with increasing landscape homogeneity. The biological control potential in conventional fields was not affected by landscape complexity, and in homogenous landscapes it was higher in conventional fields than in organic fields, as indicated by an interaction between farming practice and landscape complexity.

4. A simplification of the landscape, from 20% to 100% arable land, reduced plant species richness by about 16% and cover by 14% in organic fields, and 33% and 5·5% in conventional fields. For birds, landscape simplification reduced species richness and abundance by 34% and 32% in organic fields and by 45·5% and 39% in conventional fields. Ground beetles were more abundant in simple landscapes, but were unaffected by farming practice.

5.Synthesis and applications. This Europe-wide study shows that organic farming enhanced the biodiversity of plants and birds in all landscapes, but only improved the potential for biological control in heterogeneous landscapes. These mixed results stress the importance of taking both local management and regional landscape complexity into consideration when developing future agri-environment schemes, and suggest that local-regional interactions may affect other ecosystem services and functions. This study also shows that it is not enough to design and monitor agri-environment schemes on the basis of biodiversity, but that ecosystem services should be considered too.


Agricultural intensification in Europe has resulted in a loss of biodiversity (Donald, Green & Heath 2001; Robinson & Sutherland 2002) through management changes and landscape alterations (Chamberlain et al. 2000; Robinson & Sutherland 2002; Benton, Vickery & Wilson 2003; Geiger et al. 2010). There is growing concern that this loss in biodiversity will result in declining ecosystem services (Kremen, Williams & Thorp 2002; Hooper et al. 2005; Tscharntke et al. 2005). Organic farming, with restricted use of pesticides and mineral fertilizers, has been promoted as an agri-environment scheme (AES) to mitigate the negative effects of intensified agricultural management on biodiversity [Council Regulation (EC) No 834/2007]. The effectiveness of AES has received much attention and is under debate (Kleijn & Sutherland 2003). Organic farming often increases species richness and abundance, but the effects vary among taxa, and the composition of the landscape surrounding arable fields needs to be taken into account when evaluating the effects of organic farming (Bengtsson, Ahnström & Weibull 2005; Gabriel et al. 2006; Rundlöf & Smith 2006). In addition, the effects of organic farming and landscape complexity on ecosystem services remain largely unexplored.

The overall effectiveness of organic farming varies among taxa (Bengtsson, Ahnström & Weibull 2005; Fuller et al. 2005). Separate case studies show increased diversity of birds, plants and predatory arthropods in organic fields and farms, whereas other studies, for instance on non-predatory insects and pests, show no effect or decreased diversity (Bengtsson, Ahnström & Weibull 2005; Hole et al. 2005). Similar inconsistent results of organic farming have been found in other habitats such as grasslands (Hole et al. 2005).

Studies that simultaneously consider multiple taxa and trophic levels are rare (but see Billeter et al. 2008; Geiger et al. 2010). Even if organic farming in general promotes biodiversity, it remains unclear whether increased species richness leads to a higher provisioning of ecosystem services (Letourneau & Bothwell 2008; Macfadyen et al. 2009).

Agricultural landscapes can be structurally simple and dominated by annual or perennial arable fields, or more complex with arable fields interspersed with non-crop habitats such as grasslands, hedgerows and ditches. The effects of landscape composition on local biodiversity have been shown to be pervasive and are sometimes even more important than the farming practice (Weibull, Östman & Granqvist 2003; Purtauf et al. 2005; Aavik & Liira 2010). It has also been suggested that structurally complex landscapes provide higher levels of ecosystem services than structurally simple landscapes (Thies & Tscharntke 1999; Östman, Ekbom & Bengtsson 2001; Gardiner et al. 2009).

Recent efforts have been made to explore how the effects of organic farming on biodiversity may be modified by the surrounding landscape composition. Studies show that for wild bee, butterfly and weed diversity, organic farming increased biodiversity to the greatest extent in simple landscapes (Roschewitz et al. 2005a; Rundlöf & Smith 2006; Holzschuh et al. 2007). Only limited components of biodiversity in the agroecosystem have been examined in this context. Studies on the potential interactions of local versus landscape management on ecosystem services are still scarce (but see Roschewitz et al. 2005b; Diekötter et al. 2010), and more experimental studies are needed on the potential interactions of management practices and landscape composition on ecosystem services and functions.

In this study we analysed organic and conventional farms along a landscape complexity gradient in five regions across Europe, in a design that enabled us to separate the effects of landscape complexity and farming practice. We examined species richness and abundance of wild plants, ground beetles (Coleoptera, Carabidae) and breeding birds. We also investigated the potential for biological control measured as the removal rate of aphids glued on labels placed in cereal fields as a measure of impact on an ecosystem service. With this sample design we tested the hypothesis that both organic farming and an increased landscape complexity increase species richness and abundance of plants, ground beetles and breeding birds, as well as the potential for biological control of aphids.

Materials and methods

Study area and farm selection

The study was carried out in May, June and July 2007 in five regions in Europe: Sweden, Estonia, the Netherlands, Western Germany and Eastern Germany (Table 1). We selected c. 30 farms in each region, with 1–5 cereal fields per farm. On each farm we selected five sampling points spread across as many cereal fields as possible (see below), and one bird survey plot. Fields on different farms were located at least 1-km apart and fields within one farm were always closer than 1 km to each other. Surveyed fields were sown mainly with winter wheat Triticum aestivum L., and managed either organically or conventionally. We selected farms along both a cereal yield gradient and a landscape complexity gradient. Yield (at 14% moisture content) was used as a proxy to select farms of different management intensities. Fields were selected so that landscape composition did not differ between farming practices (F1,198 = 1·18, = 0·28). To avoid differences in species pools within regions, the size of each study region was a maximum of 50 × 50 km, and no smaller than 30 × 30 km to reduce spatial autocorrelations. Winter wheat was grown at 80·7% of the sampling points, spring wheat at 9·8% and winter rye at 8·8%, while triticale and winter barley were grown at less than 1% of the sampling points. In Flevoland in the Netherlands most sampled organic fields had spring wheat, whereas conventional fields had winter wheat.

Table 1.   Information about the organic and conventional fields in the five European regions
 No. sampling pointsField size (ha)Yield (kg ha−1)% Arable crop
  1. Numbers of sampling points, field size, yield and percentage of arable crops in a buffer zone with 1000-m radius and regional location coordinates are shown. Minimum, maximum and mean values (in brackets) are given

Uppsala, Sweden 59°51′0″N, 17°37′60″E
 Organic481·0–21·5 (7·9)2500–6000 (3980)23·2–90·7 (65·4)
 Conventional1110·8–66·7 (13·0)3450–8500 (6480)28·6–96·8 (63·1)
Tartu, Estonia 58°21′0″N, 26°34′0″E
 Organic401·3–29·2 (12·7)1310–4290 (2290)35·1–81·6 (60·9)
 Conventional1101·9–103·2 (25·2)1750–6694 (4262)38·9–88·5 (67·5)
Flevoland, Netherlands 52°32′0″N, 5°43′0″E
 Organic750·8–13·4 (6·3)1750–5250 (3370)50·8–99·5 (83·6)
 Conventional751·1–27·0 (8·7)5680–8310 (7330)40·9–98·5 (85·8)
Goettingen, Germany 51°32′2″N, 9°56′8″E
 Organic700·4–22·8 (4·5)1200–5600 (3470)21·0–92·9 (56·5)
 Conventional650·6–17·1 (4·5)2750–9800 (7140)19·6–93·3 (57·5)
Jena, Germany 51°6′29″N, 10°38′48″E
 Organic303·7–40·0 (18·7)3150–5500 (4250)32·1–95·5 (73·3)
 Conventional1200·5–85·7 (28·6)4040–9250 (7020)15·3–100·0 (80·1)

Landscape complexity and organic farming

We used three different measures of the landscape complexity in buffer zones with 500-m and 1000-m radius around each sampling point: mean field size, the percentage of land covered by arable crops, and a Shannon habitat diversity index. Mean field sizes were either calculated using GIS or determined from direct measurements of maps. Field size differed between farming types (F1,164 = 7·38, = 0·0073), with smaller fields on organic farms (11·09 ha, SE = 4·12) compared to conventional farms (16·22 ha, SE = 3·99).

The percentage of arable land and the habitat classes were calculated using the definitions from the European Topic Centre for Land Use and Spatial Information (Büttner, Feranec & Jaffrain 2002). Habitat classes used for the diversity index were: continuous urban fabrics, discontinuous urban fabrics, cultivated arable lands, fallow lands under rotation systems, permanent crops, pastures, forests, transitional woodland-shrub and water. The habitat diversity in the landscape tended to differ between the two farming practices (F1,199 = 3·60, = 0·0593), with a slightly higher diversity around organic fields. We used Geographical Information System ESRI® ArcMapTM 9·1 (ESRI Inc., Redlands, CA, USA) for the landscape analyses.

Pearson correlation was used for the selection of the landscape variable to use in the mixed model. We selected the percentage of arable land as it was correlated with the other landscape measures (log mean field size: r = 0·62; < 0·0001; n = 744, and Shannon habitat diversity index r = −0·84; < 0·0001; n = 744), and it has been used in landscape analyses in other studies (Roschewitz et al. 2005b; Thies, Roschewitz & Tscharntke 2005). In addition, we only used landscape measures estimated at a 1000-m radius as there was high correlation between 500-m and 1000-m radius (r = 0·86; < 0·0001; n = 744), and the 1000-m radius has been shown to be ecologically important for a number of organisms and biological control agents (Roschewitz et al. 2005b; Thies, Roschewitz & Tscharntke 2005). For the analysis of landscape complexity surrounding each bird survey plot, we used the mean value of the percentage of arable land of the five sampling points per farm.

Information about farming practice and management was collected by means of a standardized questionnaire answered by all participating farmers except four from Goettingen, Germany. Of the 151 farms, 51 were organically managed, and 263 of the 744 sampling points were organically managed. There was a significant difference in yield (F1,202 = 284·3, < 0·0001) between the organic fields (3559 kg ha−1, SE = 471·93) and conventional fields (6273 kg ha−1, SE = 464·00).

Species richness and abundance

Wild plants and ground beetles were sampled at the five sampling points at each farm. Three of these sampling points were also used for the assessment of biological control potential, and the largest field on each farm was used for the bird survey. Except for the bird survey, the sampling points were situated at the middle of the longest side of the sampled field, 10 m into the field from a grass covered field margin. When farms had fewer than five winter wheat fields, more than one sampling point was located in the same field. In such cases, sampling points were placed at opposite sides of the field and with a minimum distance of 50 m between them. The timing of sampling of plants, ground beetles and the biological control experiment were synchronized across countries using the phenological stages (BBCH) of winter wheat as a time reference (Federal Biological Research Centre for Agriculture and Forestry 2001).


Wild plants (predominantly arable weeds) were surveyed in three 2 × 2 m2 plots per sampling point. The distance between plots was 5 m and plots were placed parallel to the field margin. Plants were surveyed at the flowering stage of winter wheat (BBCH65). All plants within the plots were determined to species, except for individuals with fewer than two true leaves after the cotyledon. Plant cover was recorded per species using an abundance scale (1: <1%; 2: 1–5%; 3: 5–12·5%; 4: 12·5–25%; 5: 25–50%; 6: >50%) and, prior to the analysis values were re-calculated into percentage of mean cover per sampling point, using the mid value of the abundance scale classes. Total plant cover per sampling point was then calculated.

Ground beetles

Ground dwelling arthropods were caught with two pitfall traps (90-mm diameter) per sampling point. The pitfall traps were filled with 150 mL of 50% ethylene glycol and placed in the middle of the two outer vegetation plots. Roofs made of cardboard fixed with needles prevented flooding by rain. Traps were open for two periods of 7 days each. The first sampling period was 1 week after the appearance of spikes of winter wheat (BBCH55) and the second sampling period was at the milk ripening stage of winter wheat (BBCH75). One randomly selected pitfall trap per sampling point was analysed, the other trap was kept as a backup. Invertebrates caught were preserved in 70% ethanol. All ground beetles were identified to species. Ground beetle abundance was measured as the total number of individuals per sampling point. Data were missing for 29 sampling points in Jena in Germany, for eight sampling points in Flevoland, the Netherlands and for two sampling points in Uppsala, Sweden.

Breeding birds

Birds were monitored three times according to a modified version of the British Trust for Ornithology’s Common Bird Census (Bibby, Burgess & Hill 1992). A quadrat of 500 × 500 m was centred from the middle of the largest selected field at each farm, and divided into a transect grid, so that no part of the studied field was further away than 100 m from the route of the surveyor. In the bird survey plots the organic fields averaged 10·63 ha (SE = 4·07) and conventional fields 16·58 ha (SE = 3·92, F1,139 = 8·69, = 0·0038).

The first census started according to local information on the phenology of breeding birds, and the survey was then repeated every 3 weeks. Surveys were undertaken from one hour after dawn until noon on calm, sunny days with no precipitation. All birds were identified to species, and behaviour and nesting sites were noted on maps from which the number of breeding bird territories were determined, using bird behaviour observed on the three survey rounds. For the analyses of this paper we have used the species richness and abundance of breeding birds in the whole bird survey plots. Data were not obtained for three fields in Goettingen in Germany.

Biological control potential

The potential for biological control was estimated in the field as the predation of aphids glued on plastic labels. Three live pea aphids Acyrthosiphon pisum Harris of the third or fourth instar were glued with odourless superglue (Superglue UHU-Sekundenkleber®, UHU GmbH & Co KG, Bühl/Baden, Germany) on one side of an 11·5 cm long plastic label. In each of the three vegetation plots at three of the sampling points per farm, three labels were placed, one central and two at diagonally opposite corners, so that there were 27 labels with 81 aphids in total at each farm. The plastic labels were positioned in the soil with the glued aphids parallel to the ground, facing downwards, so that ground dwelling arthropod predators could detect them, whereas they were protected from the weather. Labels were placed in the field at 12·00. The percentage of surviving aphids was calculated from the number of aphids remaining after 24 h. Using live aphids stuck on labels has previously been used to estimate predator impact on aphid establishment (e.g. Östman, Ekbom & Bengtsson 2001). The experiment was performed twice within 8 days during the first inflorescence emergence of winter wheat (BBCH50). The pink clone of pea aphids used in this study was first collected in 1997 in the Bayreuth area in Germany on red clover Trifolium pratense L., and was denoted as clone BP (Weisser, Braendle & Minoretti 1999). They were reared on broad beans Vicia faba L. at 20 °C and long-day conditions in greenhouses. Data from both rounds were used for all regions except Flevoland (the Netherlands), from which we only used data from the second round, because of sampling errors during the first round. Data could not be gathered for nine farms in Goettingen, Germany, and on the remaining farms in Goettingen only one sampling point per farm was used. Data were also missing for one farm and from three other sampling points in Jena, Germany, one farm in Sweden, and from eight farms in the Netherlands.

Data analysis

We tested our hypothesis about effects of organic farming in relation to landscape complexity for each dependent variable separately by analysis of covariance (ancova) using mixed models based on normal error distributions. As fixed effects we used landscape complexity (measured as percentage of arable crops in the buffer zone of 1000 m around each sampling point), farming practice and the interaction between landscape complexity and farming practice. In cases where the interaction between landscape complexity and farming practice was not significant, we removed it from the model and continued to analyse the model for main effects as is common practice in an ancova (Littell et al. 2006).

As random terms we used region, farm nested within region and the three-way interaction between farming practice, landscape complexity and region when analysing plants, ground beetles and biological control potential. For the bird species richness and abundance we only used region and the three-way interaction between farming practice, landscape complexity and region as random terms, as birds were only observed at one sampling point per farm. By including this three-way interaction in the random terms of the model we allowed slopes to vary between regions. We selected this random structure because we were mainly interested in the overall effect across Europe of farming practice and landscape complexity on biodiversity and biological control. Specific regional differences in the responses to organic farming and landscape complexity have therefore not been considered in this study. The random part of the model was never reduced during analyses. Random models on five data sets result in quite low power. A P-value below 0·05 is therefore a strong indication of a true difference, and P-values near 0·05 have also been considered in the Results and Discussion.

Ground beetle abundance and the percentage of arable crops were log transformed [log10 (x + 1)], and plant cover was square-root transformed, to achieve residual normal distributions.

Figures were created by plotting the model estimates plus the residual values from the full model including the random terms per sampling point along the landscape gradient, as follows:


where x is the model estimate for farming practice (the intercept), PAL is the proportion arable land, z is the model estimate for landscape complexity (the slope) and residual is the model residual.

For all statistical analyses we used SAS 9.1 for Windows (SAS Institute Inc., Cary, NC, USA).


Species richness and abundance

There were more plant and breeding bird species in organic fields, while the number of ground beetle species did not differ between farming systems (Table 2, Fig. 1). Organic fields had a greater total cover of wild plants and a higher number of breeding bird territories than conventional fields, but not a greater abundance of ground beetles than conventional fields (Table 2, Fig. 2). More plant species and breeding bird species were found as landscape complexity increased (Table 2, Fig. 1), while ground beetle species richness was not affected by landscape composition. The cover of wild plants and the abundance of breeding birds increased as the proportion of arable crops decreased, but the opposite was found for ground beetles; more individuals were found in simpler landscapes with a higher proportion of arable crops (Table 2, Fig. 2). No interaction between landscape complexity and farming practice was detected for any of the analysed organism groups regarding either species richness or abundance.

Table 2.   Mixed-effects model describing relationships between farming practice (organic farming – conventional farming), landscape complexity (percentage of arable land in a 1000-m radius buffer zone) and the interaction between farming practice and landscape complexity, on the species richness and abundance of plants, ground beetles and birds, and on the removal of glued aphids
  1. P-values for terms that were included in the final models are in bold lettering. d.f., the approximate denominator degree of freedom from the mixed model analysis.

Weed species
 Farming practice1367·38<0·0001
 Farming practice × landscape310·060·81
Ground beetle species
 Farming practice7·061·500·26
 Farming practice × landscape42·20·010·91
Breeding bird species
 Farming practice7·027·380·030
 Farming practice × landscape55·60·030·87
Weed cover
 Farming practice4937·18<0·0001
 Farming practice × landscape14·70·440·52
Ground beetle individuals
 Farming practice7·643·140·12
 Farming practice × landscape41·20·000·99
Breeding bird individuals
 Farming practice5·7510·720·018
 Farming practice × landscape1371·440·23
Biocontrol potential (Aphid mortality)
 Farming practice2333·140·077
 Farming practice × landscape56·73·770·057
Figure 1.

 Plant species richness and breeding bird species richness model estimates and residuals plotted against landscape complexity (percentage of arable crops in a buffer zone with 1000-m radius). Organic fields: open circles and dotted line. Conventional fields: filled circles and solid line.

Figure 2.

 Ground beetle abundance, plant cover and breeding bird territories model estimates and residuals plotted against landscape complexity (percentage of arable crops in a buffer zone with 1000-m radius). Organic fields: open circles and dotted line. Conventional fields: filled circles and solid line.


On average 14·5 wild plant species (SE = 1·98, d.f. = 4·5) were recorded in organic fields compared to 6·3 species (SE = 1·97, d.f. = 4·35) in conventional fields. Organic fields had a 20·1% cover of wild plants (SE = 0·92, d.f. = 4·85) compared to a 2·7% cover in conventional fields (SE = 0·92, d.f. = 4·8). Landscape simplification decreased the species richness (slope = −0·028) and abundance (slope = −0·019) of plants. Increasing landscape homogenization, from complex landscapes with 20% arable fields to simple landscapes with 100% arable fields, decreased species richness of plants from an average of 16·1 (SE = 2·03, d.f. = 4·97) to 13·5 (SE = 2·08, d.f. = 5·2) in organic fields, and from 7·9 (SE = 1·97, d.f. = 4·52) to 5·3 (SE = 2·08, d.f. = 5·2) in conventional fields. At the same time wild plant cover decreased from 29·3% (SE = 0·86, d.f. = 4·56) to 15·4% (SE = 1·06, d.f. = 5·85) in organic fields and from 6·7% (SE = 0·83, d.f. = 4·11) to 1·2% (SE = 1·08, d.f. = 6·04) in conventional fields. In total 238 species of wild vascular plants were found, the most frequent species being Stellaria media L., Myosotis arvensis L., Viola arvensis Murr., Galium aparine L. and Circium arvense L.

Ground beetles

The species richness of ground beetles was on average 7·4 (SE = 1·17, d.f. = 4·63) in organic fields and 6·7 (SE = 1·15, d.f. = 4·45) in conventional fields, and beetle abundance was 48·5 (SE = 1·47, d.f. = 4·94) and 33·1 (SE = 1·46, d.f. = 4·61) respectively. Landscape complexity did not affect the species richness of ground beetles, whereas the abundance increased in more simple landscapes (slope = 0·0042). We caught in total 63 940 ground beetles, belonging to 151 different species. The most abundant species was Pterostichus melanarius Illiger, making up almost 60% of the ground beetles caught. Other abundant species were Pterostichus niger Schaller, Poecilus cupreus L., Anchomenus dorsalis Pontoppidan and Pseudoophonus rufipes De Geer.

Breeding birds

Organic fields held on average 9·0 breeding bird species (SE = 1·47, d.f. = 5·0) and 18·6 territories (SE = 2·80, d.f. = 4·72), compared to 6·0 species (SE = 1·43, d.f. = 4·3), and 13·9 territories (SE = 2·71, d.f. = 4·12), in conventional fields. Landscape simplification reduced the species richness (slope = −0·059) and abundance (slope = −0·11) of breeding birds. The average species richness of breeding birds decreased from 12·0 (SE = 1·81, d.f. = 11·1) to 7·9 (SE = 1·60, d.f. = 4·93) in organic fields, and from 9·0 (SE = 1·76, d.f. = 10·7) to 4·9 (SE = 1·55, d.f. = 5·32) in conventional fields, when comparing complex landscapes with 20% arable fields to simple landscapes with 90% arable fields. At the same time the abundance of territories decreased from 24·0 (SE = 3·40, d.f. = 10·1) to 16·4 (SE = 2·95, d.f. = 5·7) in organic fields and from 19·3 (SE = 3·36, d.f. = 9·94) to 11·7 (SE = 2·84, d.f. = 4·83) in conventional fields. We recorded 86 breeding bird species, of which 17 species were considered as true farmland birds. The most abundant farmland species was skylark Alauda arvensis L., making up 37% of the sightings. Other common farmland species were yellowhammer Emberiza citrinella L. and whinchat Saxicola rubetra L. We found in total 2349 territories in the bird survey plots, ranging from 1 to 51 per plot.

Biological control potential

In heterogeneous landscapes the potential for biological control was highest on organic fields. However, as the proportion of arable land increased, the biological control potential in organic fields dropped to a lower value than in conventional fields, because of an interaction between farming practice and landscape complexity (Table 2, Fig. 3). Overall predation rates were lower in homogeneous landscapes. Disregarding landscape complexity, the removal of glued aphids was nearly significantly different between farming practices (Table 2). In organic fields 78·3% (SE = 8·44, d.f. = 4·58) of the glued aphids were removed, and 80·8% (SE = 8·32, d.f. = 4·33) in conventional fields. For organic fields the intercept of the regression was 99·80 (SE = 11·36, d.f. = 14·6) and the slope was −0·30 (SE = 0·12, t1,102 = −2·52, = 0·013) and on conventional fields the intercept was 82·73 (SE = 9·61, d.f. = 7·57) and the slope was not different from zero (−0·028; SE = 0·081, t1,39,9 = −0·34, = 0·73).

Figure 3.

 Percentage of eaten aphid model estimates and residuals plotted against landscape complexity (percentage of arable crops in a buffer zone with 1000-m radius). Organic fields: open circles and dotted line. Conventional fields: filled circles and solid line.


Our study design enabled us to separate the effects of organic farming and landscape complexity on farmland biodiversity and the potential for biological control across five regions in Europe. We did not find any interactions between organic farming and landscape complexity for the species richness or abundance of our study organisms. However, when we examined biological control potential, the two terms interacted. This is the first time that an interaction between farming practice and landscape complexity has been demonstrated for an ecosystem service across Europe, highlighting the importance of taking both local farm management and regional landscape management into account when promoting biological control.

The biological control potential of the landscape was highest in organic fields in heterogeneous landscapes. In conventional fields the biological control potential was consistently high independent of landscape complexity, as measured by a reduction in glued aphids by c. 80% in 24 h. In organic fields, the biological control potential decreased as the landscape became increasingly simplified, resulting in lower biological control potential in organic fields than in conventional fields in homogenous landscapes. These findings support our hypothesis that organic farming combined with high landscape complexity would be advantageous for biological control. However, the fact that the biological control potential of organic fields was reduced in homogenous landscapes implies that our initial hypothesis was too simplistic, and organic farming may not result in higher biological control in all landscapes.

So why were aphid predation rates lower in organic fields than in conventional fields in homogenous landscapes? First, different predator groups may respond differently to management practices and landscape complexity. Generalist predators may benefit from more simple landscapes with higher primary production, leading to an increase in numbers (Tscharntke et al. 2007). An increase in intraguild predation with increased numbers of generalist predators (Letourneau et al. 2009), such as predation of smaller ground beetles by the larger prevalent generalistic ground beetle Pterostichus melanarius, may reduce the overall potential for biological control (Prasad & Snyder 2006). Macfadyen et al. (2009) showed that although species richness of parasitoids was higher on organic farms compared to conventional farms the natural pest control rate was no higher. Studies of such shifts in predator community composition can reveal the mechanism behind the interaction found in this study. Secondly, aphid densities, and the densities of alternative prey, may vary along these gradients, which can modify predation rates of glued aphids. Aphid densities, and parasitism rates, have been found to be higher in fields in complex landscapes compared to fields in simple landscapes (Roschewitz et al. 2005b), and both intraguild predation and alternative prey may limit the biological control potential of generalist predators (Prasad & Snyder 2006).

What about the effects on biodiversity? Organic farming and complex landscapes with a low proportion of arable fields increased both species richness and the abundance of wild plants and breeding birds, whereas ground beetle species richness was unrelated to both farming practice and landscape complexity. Ground beetle numbers were similar in organic and conventional fields, but were higher in simple landscapes with a high proportion of arable fields. These results highlight the fact that organic farming and increasing landscape complexity increase the species richness and abundance of some taxa, and suggest that agri-environment schemes need to be designed either in a general way that benefits most taxa, or should focus on certain taxa of conservation value or high ecosystem service potential.

The species richness and abundance of plants was increased both by organic farming and by landscape complexity. Organic fields harboured more than twice as many plant species as conventional fields, and the cover of plants was more than seven times higher. Our results are consistent with the larger scale analysis of nine regions in Europe (Geiger et al. 2010), which found strong effects of agriculture intensification, especially pesticide use, on biodiversity, and where agri-environment schemes increased the species richness of wild plants. Other European studies have also found higher species richness of plants in areas with a high proportion of semi-natural areas and low intensity land-use (Billeter et al. 2008). In organic farming, synthetic pesticides are not allowed and farmers rely on other methods such as crop rotation, mechanical weed control and enhancement of natural enemies of pests to reduce pests and weeds [Council Regulation (EC) No 834/2007]. The lack of herbicide use may explain why organic farming is especially beneficial to wild plants. The higher number of plant species in fields in complex landscapes may also be a result of a greater species pool in varied landscapes, and the shorter distances from natural areas that could act as sources for colonization (Gabriel et al. 2006). The smaller field size of organic farms may also support the establishment of more plant species (Geiger et al. 2010).

Greater plant biodiversity has been shown to have a number of effects on related biodiversity, ecosystem functions and services. Wild plants may, for instance, promote natural enemies and help to reduce crop damage by herbivores, while plant competition may reduce yields (Poveda, Gómez & Martínez 2008), an adjustment that may be especially important to organic farmers.

Higher numbers of breeding bird species and territories were found on organic than on conventional survey plots, and the numbers increased with landscape complexity, again supporting our initial hypothesis. Several studies have reported similar effects (Bengtsson, Ahnström & Weibull 2005; Hole et al. 2005). However, a recent study including nine regions in Europe found no difference in bird species richness between organic and conventional fields (Geiger et al. 2010). A possible explanation for this discrepancy could be that Geiger et al. (2010) used data from Poland and Spain, two regions characterized by high levels of species richness on conventional farms.

Organic farming has been associated with higher densities of bird species only for those that are more or less obligate breeders in crop fields, such as skylarks and lapwings (Piha et al. 2007). This may be because of the absence of pesticides, directly by reducing the risk of birds being exposed to chemicals, and indirectly by providing more diverse and abundant food resources (Piha et al. 2007). Also, predation risk may be reduced and foraging efficiency enhanced in shorter and sparser vegetation (Whittingham & Evans 2004). In heterogeneous landscapes a variety of resources are accessible within a short distance. This can enhance breeding or foraging success for birds that depend on different habitats, or prefer vegetation of different height: should a nesting attempt fail, then birds in complex landscapes can more easily find a new nesting site within their territory (Wilson et al. 1997).

We found no effects of organic farming on ground beetle species richness or abundance, and ground beetle species richness was unaffected by landscape complexity, whereas ground beetle abundance increased with landscape homogenization. These results oppose our initial hypothesis, but are partly supported by other studies (Holland & Luff 2000; Bengtsson, Ahnström & Weibull 2005). The European-wide study by Geiger et al. (2010) found that agri-environment schemes increased ground beetle species richness, whereas insecticide use reduced species richness. Ground beetles are affected by a range of factors, from abiotic soil factors, ploughing, pesticide and fertilizer use to non-crop habitats in the landscape, but it is still not known which of these factors are most important in controlling their distribution (Holland & Luff 2000). The effects often vary between species and studies, making it hard to draw general conclusions. Pterostichus melanarius, the most common species found in our study, has been shown to be negatively affected by ploughing, a management practice used by both organic and conventional farmers (Holland & Reynolds 2003). Specific management practices may be more important in explaining the distribution of ground beetles than general farming schemes.

Increases in ground beetle numbers in simpler landscapes have previously been found (Weibull, Östman & Granqvist 2003). Pterostichus melanarius and some other common ground beetles have been shown to hibernate in the field, and are therefore not dependent on field margins and semi-natural habitats for overwintering (Holland & Luff 2000). Another explanation may be that natural predators in heterogeneous landscapes spend more time in adjacent habitats, and are not so abundant in crop fields (Östman, Ekbom & Bengtsson 2001).

Different spider families have been found to respond in different ways to the surrounding landscape (Öberg, Ekbom & Bommarco 2007) possibly because of different feeding modes. A consideration of life-history traits may clarify how different components of the predator community are affected by landscape complexity and farming practice.

An increase in predator numbers does not necessarily increase the biological control potential of an area (Macfadyen et al. 2009). In our study, the biological control potential decreased in homogenous landscapes, whereas ground beetle abundance increased. There are several other predator and parasitoids groups in arable fields that contribute to the potential for biological control and further experimental studies are needed to establish best management practice for these groups.

We found that organic farming is beneficial for plants and breeding birds in all landscapes, but we did not find any interactions between organic farming and landscape complexity for the species richness or abundance of our study organisms. Such interactions have previously been found in studies on plants (Roschewitz et al. 2005a), migratory birds (Dänhardt et al. 2010), and pollinators such as butterflies, wild bees and bumblebees (Rundlöf & Smith 2006; Holzschuh et al. 2007; Rundlöf, Nilsson & Smith 2008), where organic farming in homogenous landscapes increased biodiversity the most.

The way in which local-landscape effects on biodiversity will translate into similar patterns for related ecosystem services and functions remains largely unexplored. Diekötter et al. (2010) found an interaction between local farming practice and landscape context on the activity density of decomposers, and on the species richness of ground beetles, but no such interactions were detected for the associated ecosystem services of litter decomposition and seed predation.

In our case the opposite situation occurred: the effect of local management on the potential for biological control was mediated by the complexity of the surrounding landscape, even without detection of interacting effects on biodiversity. In order to fully understand and properly manage other ecosystem services or functions in other ecosystems, potential interactions between local management and the surrounding landscape need to be taken into account.

Management implications

Agri-environment schemes aim at local enhancement of biodiversity and associated ecosystem services, but little is known about the importance of landscape complexity for the effectiveness of agri-environment schemes across Europe. Our study provides Europe-wide evidence for the effects of both organic farming and landscape complexity on biodiversity and the potential for biological control. We found that biodiversity and the potential for biological control were highest when organic farming was combined with complex, heterogeneous landscapes. In homogeneous landscapes, organic farming can help to maintain biodiversity, but the biological control potential of the area may be reduced. In homogenous landscapes, measures should be taken to increase landscape complexity. This should be considered in the development of future agri-environment schemes, which generally only focus on the field or farm scale.

The fact that the effects of organic farming and landscape complexity differ between biodiversity and the potential for biological control shows that agri-environment schemes designed to enhance or preserve biodiversity may not necessarily enhance ecosystem services. In order to sustain ecosystem services, both the design and monitoring of agri-environment schemes need to focus on processes in addition to biodiversity per se.

It is likely that interactions between local management and landscape complexity exist for other ecosystem services in other habitats so this must be considered in order to optimize future conservation actions. A challenge for the future is to design conservation schemes that act at both local and regional scales, enhancing both biodiversity and ecosystem services.


We would like to thank all the farmers and field assistants involved, and Johan Ahnström, Sandra Lindström, Ola Lundin and Maj Rundlöf for valuable comments on the manuscript. We also thank the European Science Foundation and the connected National Science Foundations (Estonian Scientific Foundation, German Federal Ministry of Education and Science BMBF, German Research Foundation, Netherlands Organization for Scientific Research and Swedish Research Council) for funding through the EuroDiversity AgriPopes programme.