How Do Landscape Composition and Configuration, Organic Farming And

Journal of Animal Ecology 2010, 79, 491–500 doi: 10.1111/j.1365-2656.2009.01642.x How do landscape composition and conﬁguration, organic farming and fallow strips affect the diversity of bees, wasps and their parasitoids?

Andrea Holzschuh1*, Ingolf Steffan-Dewenter2 and Teja Tscharntke1

1Agroecology, Georg-August University, Waldweg 26, D-37073 Go¨ttingen, Germany; and 2Population Ecology Group, Department of Animal Ecology I, University of Bayreuth, Universita¨tsstraße 30, 95440 Bayreuth, Germany

Summary 1. Habitat destruction and increasing land use intensity result in habitat loss, fragmentation and degradation, and subsequently in the loss of species diversity. The fact that these factors are often highly confounded makes disentangling their effects extremely difficult, if not impossible, and their relative impact on species loss is mostly speculative. 2. In a two-year study, we analysed the relative importance of changed landscape composition (increased areas of cropped habitats), reduced habitat connectivity and reduced habitat quality on nest colonization of cavity-nesting bees, wasps and their parasitoids. We selected 23 pairs of conventional and organic wheat fields in the centre of landscape circles (500 m radius) differing in edge densities (landscape configuration) and % non-crop habitats (landscape composition). Standard- ized trap nests were established in the field centres and in neighbouring permanent fallow strips (making a total of 92 nesting sites). 3. Factors at all three scales affected nest colonization. While bees were enhanced by high proportions of non-crop habitat in the landscape, wasps profited from high edge densities, supporting our hypothesis that wasps are enhanced by connecting corridors. Colonization of herbivore-predating wasps was lower in field centres than in fallow strips for conventional sites, but not for organic sites, indicating a fallow-like connectivity value of organic fields. The relative importance of habitat type and farming system varied among functional groups suggesting that their perception of crop–non-crop boundaries or the availability of their food resources differed. 4. Local and landscape effects on parasitoids were mainly mediated by their hosts. Parasitism rates were marginally affected by local factors. A specialist parasitoid was more sensitive to high land use intensity than its host, whereas generalist parasitoids were less sensitive. 5. We conclude that the conversion of cropland into non-crop habitat may not be a sufficiently successful strategy to enhance wasps or other species that suffer more from isolation than from habitat loss. Interestingly, habitat connectivity appeared to be enhanced by both higher edge densities and by organic field management. Thus, we conclude that high proportions of conventionally managed and large crop fields threaten pollination and biological control services at a landscape scale. Key-words: agri-environment schemes, field margins, pollinators, predators, trophic interactions tion), habitat fragmentation is a measure of connectivity Introduction which is strongly affected by the geometry of habitats (land- Habitat loss, habitat fragmentation and degradation of habi- scape configuration) (Fahrig 2003). Linear habitat strips tat quality are considered to be among the main threats of which connect otherwise isolated habitats can enormously biodiversity (Harrison & Bruna 1999; Fahrig 2003). While reduce habitat fragmentation even if the total area of habitat habitat loss or the amount of remaining habitat types within strips is low (Haddad & Tewksbury 2005). Knowledge of the a landscape are measures of habitat area independently of relative importance of habitat loss, fragmentation and degra- the configuration of habitat patches (landscape composi- dation is essential for understanding changes in diversity and species interactions, and for implementing effective conserva- *Correspondence author. E-mail: [email protected] tion and restoration measures (Collinge 1996; Fahrig 1997).

2009 The Authors. Journal compilation 2009 British Ecological Society 492 A. Holzschuh, I. S. Dewenter & T. Tscharntke

However, the separation of effects of habitat loss and habitat actions (Tscharntke & Brandl 2004; Diekotter et al. 2007). fragmentation has frequently proved difficult in real land- Higher trophic levels such as specialized parasitoids have scapes, because both factors are often highly confounded often been hypothesized to be more sensitive to habitat loss (Fahrig 1997; Trzcinski, Fahrig & Merriam 1999; Debinski & and fragmentation than their hosts, because they have to find Holt 2000; Cushman & McGarigal 2003; Ritchie et al. 2009). habitat patches which are occupied by a host, whereas their Furthermore, habitat fragmentation often results in the deg- hosts only have to find a habitat patch (Kruess & Tscharntke radation of habitat quality in the remaining patches and con- 1994; Holt et al. 1999). Differences in the sensitivity of par- sequently impedes the disentangling of landscape and local asitoids and hosts may result in a release of parasitism in factors (Harrison & Bruna 1999; McGarigal & Cushman dependence on habitat or landscape quality (Kruess & 2002; Fahrig 2003). The aim of our study was to examine the Tscharntke 1994; Roland & Taylor 1997). relative importance of landscape composition, landscape The purpose of our study was to examine how habitat type configuration and local habitat quality which was influenced and farming intensity at the local scale, and landscape com- by land use intensity. Agricultural landscapes can be ideal position and configuration at the landscape scale affect nest study environments for disentangling such effects. This is the colonization of cavity-nesting bees and wasps, and their par- case, if the length of connecting non-crop strips such as field asitoids. Additional analyses were performed for the two banks and margin strips (a measure of configuration) is inde- functional groups of herbivore-predating and spider-predat- pendent of the total amount of non-crop habitat (a measure ing wasps, for the dominant bee species Osmia rufa and other of composition) across landscapes, and if habitats differing bee species, and for the corresponding parasitoids. All cavity- in local quality occur across all landscapes. nesting bees and wasps depend on nesting sites in non-crop To date, effects of landscape composition and connectivity habitats, but forage in multiple habitats including crop fields. have almost always been examined separately. The positive We established standardized nesting sites in organic and con- impact of landscape connectivity is known from studies on ventional wheat fields and adjacent perennial fallow strips. In habitat corridors which connect otherwise isolated habitat a hierarchical design, we analysed effects of landscape config- patches (e.g. Haddad 1999; Baum et al. 2004). In agricultural uration and composition (gradients of edge density and pro- landscapes, grass strip corridors in a crop field matrix portion of non-crop habitats in landscape circles with 500 m enhance the colonization of new nesting sites by wasps (Holz- radius), farming system (organic vs. conventional field man- schuh, Steffan-Dewenter & Tscharntke 2009). However, agement) and habitat type (highly disturbed field centre vs. studies on the effects of corridors in open landscapes are rare permanent fallow strip) on nest colonization of wasps, bees and the majority suggests that the benefits of corridors might and their parasitoids and on parasitism rates. be small compared to the impact of the quality of the remaining habitat patches and of the surrounding matrix (O¨ ckinger We tested the following hypotheses: & Smith 2008). Studies examining the effects of both land- 1. Nest colonization by bees and wasps increases with scape and local scales have focused predominantly on land- increasing proportion of non-crop habitats (landscape scape composition rather than configuration and suggested composition) and with increasing edge density providing that heterogeneous agricultural landscapes with many and connectivity (landscape configuration). diverse non-crop habitats enhance farmland species more 2. Nest colonization by bees and wasps is higher in fallow than local differences in farming intensity (Kremen et al. strips than in field centres and higher under organic than 2004; Bengtsson, Ahnstro¨ m & Weibull 2005; Clough et al. under conventional farming methods. 2005). Landscape and local effects can also interact: benefits 3. Parasitoids are more affected by agricultural intensifica- of low-intensity farming were small in landscapes with many tion at local and landscape scales than their bee and wasp non-crop habitats, but great in landscapes dominated by hosts. crop fields (Roschewitz et al. 2005; Rundlo¨ f & Smith 2006; Holzschuh et al. 2007; Rundlo¨ f, Nilsson & Smith 2008). However, local increases in land use intensity have often been Materials and methods found to affect species diversity enormously. Conventional STUDY REGION AND STUDY SITES farming reduces diversity and abundance of a range of taxa compared to organic farming (reviewed in Bengtsson et al. The study was conducted in 2003 and 2004 in 46 winter wheat fields 2005; Hole et al. 2005), and diversity and abundances are and adjacent permanent fallow strips in the area surrounding Go¨ ttin- lower in field centres than in field edges, field margins or fal- gen, Lower-Saxony (5132¢00¢¢ N00956¢00¢¢ E). In the study region, low strips (Fussell & Corbet 1991; Ba¨ ckmann & Tiainen very intensively used and fertile soils in flat parts of the region, alter- nate with less intensively used agricultural landscapes in hilly parts. 2002; Marshall & Moonen 2002; Pywell et al. 2005; Clough Wheat is the most important arable crop in the study region as well et al. 2007). Here, we assess at once the impact of two local as in most agricultural regions in Germany (Statistisches Bundesamt factors (farming system and habitat type), of landscape com- 2004). position and of landscape configuration. Within the region, 12 study areas were selected to encompass land- The perception of landscape, habitat type and farming scape gradients from crop-dominated to non-crop-dominated land- intensity may strongly differ among species, functional scapes and from low to high edge densities. Within each study area, a groups and trophic levels, resulting in changed species inter- pair of organic and conventional winter wheat fields was selected for

2009 The Authors. Journal compilation 2009 British Ecological Society, Journal of Animal Ecology, 79,491–500 Landscape composition, configuration and farming intensity 493 each year. As a result of crop rotation, field pairs within a study area Table 1. Mean (±SE) number of flowering plant species and % were not the same in 2003 and 2004. In total, 23 field pairs in 12 study flower cover in organic and conventional fields, and adjacent fallow areas were used, because no wheat fields were managed organically in strips one of the study areas in 2004. Organic farmers managed wheat fields according to the European Union regulation 2092 ⁄ 91 ⁄ EEC, which Flowering Flower prohibits the use of synthetic fertilizers and pesticides. Instead of syn- Site plant species cover (%) thetic fertilizers, organic farmers applied animal and green manure Conventional field centre 1Æ7±0Æ4<0Æ1±0Æ1 and included legumes in the crop rotation for replenishing the soil Organic field centre 8Æ1±0Æ61Æ6±0Æ6 resources. Weeds were managed mechanically or by effective crop Conventional fallow strip 12Æ5±1Æ50Æ7±0Æ2 rotations. Organic fallow strip 17Æ4±1Æ31Æ8±0Æ4 Each organic field was paired with the first nearby conventional winter wheat field for a comparison of farming systems which con- trolled for differences in abiotic conditions and landscape context. Dewenter 1998). Trap nests were composed of two trap nest tubes fit- Distances between fields within a pair ranged from 0 to 600 m and ted on a wooden pole at a height of 1Æ0–1Æ2 m and shaded by a between study areas from 3 to 43 km. Mean field size was 41 · 50 cm chipboard roof. Each trap nest tube consisted of 150–180 4Æ5±0Æ5 ha (SE) and did not differ between the two farming types 20 cm long internodes of common reed Phragmites australis, which (anova: F =2Æ6, P =0Æ118, n = 46). were put into a 10Æ5 cm diameter plastic tube. The diameters of reed One side of each field was flanked by a permanent fallow strip internodes ranged from 2 to 10 mm. Trap nests were in the field from between the field boundary and a farm track. Fallow strips were older mid-April until harvest end-July. In the laboratory, all reed inter- than 20 years, had a naturally developed herb and grass layer, and nodes containing nests were opened. For each nest, the genus of bee mostly included a narrow ditch. The occurrence of a ditch and the or wasp larvae, the number of brood cells and the occurrence of natu- management (mowing) of fallow strips did not differ between strips ral enemies were recorded (Tscharntke et al. 1998). Most larvae of adjacent to organic and conventional fields. Mean fallow strip width bees, wasps and natural enemies were identified to the species level. was 3Æ0±0Æ2 m (SE) and did not differ between fallow strips adja- All nests were reared separately to get the adults of bees, wasps and cent to conventional or organic fields (anova: F =1Æ1, P =0Æ309, their natural enemies for final species identification. In some cases, n = 46). no adults emerged or all brood cells were parasitized, so that only the A standardized nesting site was established in the centre of each genus (or the family in case of the eumenids) could be identified. field and each fallow strip (altogether 92 nest sites: 23 field pairs · 2 These reed internodes were included in the analyses as additional spe- farming systems · 2 habitat types). In conventional fields, farmers cies, if no other species of this genus (or family) was found in the same did not apply insecticides within a 15 · 15 m quadrate with the nest- trap nest. If another species of this genus (or family) was found in the ing site in the centre. Diversity of flowering plants and flower cover same trap nest, the unidentified species was assumed to be the same were higher in organic than in conventional sites, and higher in fallow as the identified species. strips than in field centres (linear mixed-effects models: all Species richness represented the total number of species, abun- P <0Æ001, Table 1). dance the total number of brood cells of bees, wasps and natural enemies from four trap nests per study site. The mortality rate was the number of parasitized or predated brood cells divided by the total TRAP NEST COMMUNITIES number of brood cells per study site. Additionally, we performed sep- Standardized nesting sites (trap nests) enabled us to study nest colo- arate analyses for herbivore-predating wasps and spider-predating nization of cavity-nesting bees, wasps and their parasitoids under wasps (Table 2), and for the dominant red mason bee O. rufa (96% standardized nest site conditions (Tscharntke, Gathmann & Steffan- of bee brood cells) and ‘other bees’.

Table 2. Trap-nesting species, their larval food and their parasitoids. Data are from 92 trap nests in organic and conventional wheat ﬁelds and the adjacent fallow strips. Braconidae, Ichneumonidae and Acari were not further identiﬁed

Group Family Larval food Species Natural enemies

Predators of herbivores Eumenidae Caterpillars Ancistrocerus gazella, A. nigricornis, Chrysis ignita, Melittobia acasta, A. parietinus, A. trifasciatus, Megatoma undata, Braconidae, Symmorphus crassocerus, S. gracilis Ichneumonidae Predators of herbivores Sphecidae Aphids Passaloecus corniger, P. gracilis Chrysis ignita, Melittobia acasta, Megatoma undata, Ichneumonidae Predators of spiders Sphecidae Spiders Trypoxylon clavicerum, T. ﬁgulus, Chrysis cyanea, Chrysis ignita, Megatoma undata, Melittobia acasta, Braconidae, Ichneumonidae Predators of spiders Pompilidae Spiders Dipogon intermedius – Osmia rufa Megachilidae Pollen Osmia rufa Anthrax anthrax, Cacoxenus indagator, Megatoma undata, Monodontomerus obsoletus, Trichodes apiarius, Acari Other bees Megachilidae Pollen, nectar Chelostoma ﬂorisomne, Hylaeus communis, Chrysis spp., Megatoma undata, H. confusus, Heriades truncorum, Megachile Melittobia acasta, Trichodes apiarius versicolor, Osmia leaiana Ichneumonidae

2009 The Authors. Journal compilation 2009 British Ecological Society, Journal of Animal Ecology, 79,491–500 494 A. Holzschuh, I. S. Dewenter & T. Tscharntke

dance were linear. All statistical analyses were performed using R (R LANDSCAPE CONTEXT Development Core Team 2005). For each wheat field, the surrounding landscape was characterized in a landscape circle with the field in the centre and a radius of 500 m. The radius was chosen according to the results of previous Results studies on trap-nesting bees and wasps (Gathmann & Tscharntke In total, 11 193 brood cells of eleven wasp species (2567 2002; Steffan-Dewenter 2002). During field inspections, we mapped brood cells) and seven bee species (8644 brood cells) were col- the land use in these landscape circles on the basis of official topo- lected from 92 trap nest sites (184 trap nest tubes). Eight wasp graphical maps (DGK 1:5000) in 2003 and 2004. The edge density species were predators of herbivorous insects (75Æ6% of the (total length of patch edges divided by total area) and the proportion of non-crop habitats [(total area)crop field area) ⁄ total area] in wasp brood cells; 1941 brood cells of six eumenid species spe- each landscape circle were calculated for each year using Geo- cialized on lepidopterous larvae and two sphecid species spe- graphic Information Systems (GIS; Topol 4Æ506, Gesellschaft fu¨ r cialized on aphids, Table 2). Three wasp species were digitale Erdbeobachtung und Geoinformation mbH, Go¨ ttingen, specialized predators of spiders (24Æ4% of the wasp brood Germany and ARC ⁄ View 3Æ2., ESRI Geoinformatik GmbH, Han- cells; 626 brood cells of two sphecid species and one pompilid nover, Germany). We considered all interfaces between different species, Table 2). Bee communities were dominated by the habitat types, between crop fields cultivated with different crops, red mason bee O. rufa L. (95Æ6% of the bee brood cells; 8266 and between grassland parcels or crop fields managed by different brood cells of O. rufa; 378 brood cells of other bee species). farmers as patch edges. The edge areas of managed habitats were We recorded 12 species of natural enemies in 2334 host brood characterized by less intensive management than the field centres cells. Two species of natural enemies (Megatoma undata, (Clough et al. 2007). The interface between crop fields or grasslands Trichodes apiarius), which were found in 82 (3Æ5%) of the consisted of a small strip of bank vegetation. At least one of the edges of crop fields or grasslands was bordered by a perennial fallow brood cells, were predators, all other natural enemies were strip, hedge or tree row. Fallow strips wider than 3 m, hedges and (klepto-) parasitoids. Because of the dominance of parasi- tree rows were mapped as semi-natural non-crop habitats. Land- toids, we refer to natural enemies as parasitoids below. Three scape parameters for landscape circles around organic and conven- species of natural enemies attacked wasps, six species tional fields forming a pair were averaged for each year. Landscape attacked bees, and three species were found in both bee and mosaics were formed by crop fields cultivated with cereals (44% of wasp nests. The drosophilid fly Cacoxenus indagator was the total landscape area), oilseed rape (8%), sugar beet (4%), maize dominant parasitoid of O. rufa (in 94% of parasitized brood (2%), and by non-crop habitats including intensively managed cells), which was its only host species (Table 2). Other bees grassland (13%), semi-natural habitats such as calcareous grass- were mainly parasitized by Melittobia acasta (76%), herbi- lands and orchard meadows (10%), forests (8%), fallows (5%), set- vore-predating wasps by Ichneumonidae (48%) and M. acas- tlements (2%) and others (4%). We used Spearman rank ta (33%), and spider-predating wasps by Ichneumonidaes correlations to test correlations between landscape parameters. The proportion of non-crop habitats in both years was positively corre- (33%), Chrysis spp. (30%) and M. acasta (28%). The mortal- lated with the Shannon-index of habitat type diversity (using the ity by natural enemies was 21Æ9±3Æ2% (mean ± SE) for percentage values of all habitat types; Steffan-Dewenter 2002) wasps and 24Æ2±1Æ8% (mean ± SE) for bees. (R =0Æ9, P <0Æ005), the proportion of grassland (R =0Æ8, P <0Æ01), and the proportion of forest (R =0Æ8, P <0Æ01), but WASPS not with edge density (R =0Æ2, P >0Æ4). The colonization of trap nests in organic and conventional STATISTICS fields and in adjacent fallow strips was related to the surrounding landscape, the farming system (organic vs. conven- Linear mixed-effects models (Pinheiro & Bates 2000) were used to tional) and the habitat type (wheat field centre vs. fallow analyse effects on species richness and number of brood cells of wasps strip). At the landscape scale, the species richness of total and bees. Edge density, proportion of non-crop habitats, farming system (organic vs. conventional) and habitat type (field centre vs. wasps and of herbivore-predating wasps, and the total num- fallow strip) were considered as fixed factors, study area and year as ber of wasp brood cells increased with increasing edge density random factors. (landscape configuration) in landscape circles with 500 m The following error structure was incorporated in the models radius (Table 3, Fig. 1a,b). The landscape effect was inde- (number of levels indicated in parentheses): ‘study area’ (12) ⁄ ‘year’ pendent of farming system and habitat type. The proportion (2) ⁄ ‘farming system’ (2) ⁄ ‘habitat type’ (2). Statistics for the random of non-crop habitats (landscape composition) had no addi- factors are shown in Appendix S1. We used Wald tests to test for sig- tional explanatory power. Our data show that a doubling of nificance of fixed effects and twofold interactions among them. Fixed edge density from 350 to 700 m edge per ha resulted in 260% factors and interactions that did not contribute to the model with more wasp brood cells in trap nests. P <0Æ05 were removed in a backward stepwise procedure from the At the local scale, the species richness of wasps and the full model. Response variables were transformed [log10(x +1)].In total number of wasp brood cells were higher in organic than addition, we tested for polynomial effects of the landscape factors by adding the fixed factors (edge density)2 and (proportion of non-crop in conventional sites and higher in fallow strips than in field habitats)2 to the model. None of these factors had additional explan- centres (Table 3, Fig. 2a). The species richness of herbivore- atory power (P >0Æ1) suggesting that the relationships between predating and spider-predating wasps reflected the results landscape factors and log-transformed insect diversity and abun- found for the species richness of total wasps (Table 3). An

2009 The Authors. Journal compilation 2009 British Ecological Society, Journal of Animal Ecology, 79,491–500 Landscape composition, conﬁguration and farming intensity 495

Table 3. Final linear mixed-effects models describing the effects of of bees was higher in organic than in conventional sites (fal- edge density and proportion of non-crop habitat (in landscape circles low strips and field centres), and higher in fallow strips than with 500 m radius), farming system (organic vs. conventional), in field centres (Table 4, Fig. 2d). For the most abundant bee habitat type (field centre vs. fallow strips) and their interactions on species O. rufa, organic farming enhanced the number of species richness and number of brood cells of wasps brood cells significantly, resulting in 30% more brood cells in d.f. FP organic than in conventional fields and 107% more brood cells in fallow strips adjacent to organic than in fallow strips Total species richness of wasps adjacent to conventional fields (Table 4, Fig. 2e). For other Edge density 10 7Æ40Æ022 bees, the number of brood cells was marginally higher in fal- Farming system 22 8Æ30Æ009 low strips than in field centres (Table 4, Fig. 2f). The total Habitat type 45 22Æ6<0Æ001 number of bee brood cells was marginally enhanced by Species richness of predators of herbivores Edge density 10 7Æ50Æ021 organic farming and in fallow strips (Table 4). Farming system 22 4Æ00Æ059 Habitat type 45 12Æ9<0Æ001 Species richness of predators of spiders PARASITOIDS Farming system 22 5Æ10Æ035 Habitat type 45 23Æ6<0Æ001 Species richness of parasitoids and number of parasitized Total brood cell number of wasps brood cells increased with increasing host species richness Edge density 10 5Æ20Æ046 and increasing number of host brood cells for herbivore-pre- Farming system 22 6Æ90Æ016 dating wasps and spider-predating wasps (models with two Habitat type 44 19Æ8<0Æ001 fixed factors; all P <0Æ005). For O. rufa and for ‘other bees’ Farming system · habitat type 44 4Æ60Æ038 species richness of parasitoids and number of parasitized Brood cell number of predators of herbivores Farming system 22 3Æ70Æ068 brood cells increased with increasing number of host brood Habitat type 44 10Æ50Æ002 cells only (all P <0Æ004). Farming system · habitat type 44 4Æ60Æ036 Although effects of local and landscape factors on parasi- Brood cell number of predators of spiders toids were mediated by host abundance and host species rich- Habitat type 44 21Æ8<0Æ001 ness, patterns of parasitized brood cells slightly differed from Non-significant factors and interactions (P >0Æ1) were removed in a what we expected according to host patterns. Parasitoids of stepwise backward procedure from the full model. the bee O. rufa were more sensitive to habitat quality than their hosts, parasitoids of ‘other bees’ and of herbivore-predating wasps were less sensitive. For O. rufa, the number of interaction between farming system and habitat type indi- parasitized brood cells, but not of total host brood cells was cated that the farming system influenced the number of wasp lower in field centres than in fallow strips, resulting in a mar- brood cells in fields and fallow strips differently (Table 3): ginally lower parasitism rate in field centres (Table 5). The The mean number of wasp brood cells was more than 200% number of brood cells of ‘other bees’ and of herbivore-pre- higher in organic than in conventional fields, whereas the dating wasps, but not the number of parasitized brood cells number of brood cells in fallow strips were not influenced by was lower in field centres and in conventional sites, respec- the farming system of the adjacent field. Organic field centres tively. These divergences resulted in marginally higher para- and fallow strips adjacent to both organic and conventional sitism rate in field centres for ‘other bees’, and in fields had similar numbers of wasp brood cells. Herbivore- conventional sites for herbivore-predating wasps (Table 5). predating wasps showed the same pattern as found for the total species richness and total number of brood cells of Discussion wasps (Table 3, Fig. 2b). Spider-predating wasps were influenced by habitat type only, with higher brood cell numbers in The results of our study support our hypotheses that factors fallow strips than in organic or conventional field centres at all spatial scales, including farming system, habitat type, (Table 3, Fig. 2c). landscape composition and connectivity, contribute to the explanation of nest colonization patterns. Cavity-nesting bees and wasps depend on nesting sites in non-crop habitats, BEES but forage in multiple habitats including crop fields in a com- The species richness of bees and the number of brood cells of plex and little known way. The relative importance of land- total bees, of the red mason bee O. rufa and of other bees scape composition vs. configuration and of habitat type vs. were positively related to the proportion of non-crop habitats farming system depended on the species group tested. Differ- (landscape composition) in a 500 m radius, but not to edge ences between parasitoids and their hosts resulted in changes density (landscape configuration, Table 4, Fig. 1). The land- in parasitism rates. scape effect was independent of farming system and habitat Bees were enhanced by high proportions of non-crop habi- type. A doubling of non-crop habitats from 30% to 60% in a tat (landscape composition), whereas wasps were enhanced landscape circle with 500 m radius resulted in an increase of by high edge densities (landscape configuration). Edge total bee brood cells of more than 100%. The species richness density is positively related to the length of linear non-crop

2009 The Authors. Journal compilation 2009 British Ecological Society, Journal of Animal Ecology, 79,491–500 496 A. Holzschuh, I. S. Dewenter & T. Tscharntke

(a) (c) 5 3

3 2 2

1 1 Number of bee species Number of wasp species 0 300 400 500 600 700 800 0 20406080 Edge density (m ha–1) % Non-crop (b) (d) 1000 200 400 50 100

10 40 5 10 Fig. 1. Effects of the landscape context (edge density and % non-crop ﬁelds in 500 m radi- Number of bee brood cells Number of wasp brood cells 0 us) on species richness and brood cell num- 300 400 500 600 700 800 0 20406080 bers of wasps and bees. Results are based on Edge density (m ha–1) % Non-crop mixed-effects models (see Table 2). Data of the two study years were averaged for each Conventional fields Organic fields of the four site types per study area (conven- Conventional fallow strips Organic fallow strips tional ⁄ organic ﬁeld centre ⁄ fallow strip).

features such as bank vegetation, fallow strips, hedges and trap-nesting bees seem to be limited by a lack of nesting tree rows which lie parallel to the field edges. The positive resources (affecting bees in both organic and conventional effect of edge density rather than of total area of non-crop fields) rather than by a lack of flower resources at the land- habitat for nest colonization by wasps suggests that field scape scale (affecting bees in conventional fields more edges provided connectivity and facilitated wasp movements strongly). We conclude from our study that for cavity-nesting between trap nests and source habitats where dispersal bees and wasps organic fields cannot compensate for missing started (Fried, Levey & Hogsette 2005). This supports experi- nesting resources in non-crop habitats at the landscape scale. mental findings showing benefits of grass strip corridors for While landscape characteristics determine where dispers- nest colonization by wasps (Holzschuh et al. 2009). In con- ing individuals can come from, local factors may affect their trast, bee colonization may have been enhanced by a high final colonization decision. We hypothesized that nest coloni- number of source populations, when the amount of non-crop zation is higher in fallow strips than in field centres, and habitat in the landscapes is high, independent of connectivity. higher under organic than under conventional farming meth- The differing response of bees and wasps to landscape factors ods. Both organic farming and fallow strips enhanced the may have consequences for pollination and predator–prey diversity and abundance of wasps and bees compared to con- systems: while bees may provide pollination services within ventional farming and wheat field centres. A meta-analysis at least 500 m radius around their non-crop nesting habitats, revealed that there are great beneficial effects of organic predation by wasps may decline and predator–prey interac- farming on plants and predatory insects, while effects on tions may shift in favour of prey in landscapes where bound- non-predatory insects and pests are small (Bengtsson et al. ary structures are missing. 2005). While bees may benefit from the absence of agrochem- Landscape configuration and composition affected trap- icals and a higher abundance and diversity of flowering nesting wasps and bees in both habitat types and both farm- plants in organic fields (Holzschuh et al. 2007), predatory ing systems similarly. In studies on flower-visiting bees land- wasps may prefer organic fields due to a higher abundance of scape and local factors were found to interact with each spiders, aphids and lepidopteran larvae, which are used for other: diversity decreased with decreasing landscape hetero- nest provision (Tscharntke et al. 1998; Schmidt et al. 2005). geneity in conventional fields, but not in organic fields, indi- Our results show that even non-crop nesting specialists do cating that organic fields compensated for lacking non-crop not generally perceive cereal fields as hostile landscape foraging habitats in homogeneous landscapes (Holzschuh matrix. The positive effect of organic compared to conven- et al. 2007; Rundlo¨ f et al. 2008). In contrast to the majority tional farming underlines the impact of local food availability of flower-visiting bees, which are mainly ground-nesting, on nest colonization (Tscharntke et al. 1998) and revealed

2009 The Authors. Journal compilation 2009 British Ecological Society, Journal of Animal Ecology, 79,491–500 Landscape composition, conﬁguration and farming intensity 497

(a)2·4 Farming system** (d) 2 Farming system* Habitat type*** Habitat type*

1·5 1·6

0·8 0·5 Number of bee species Number of wasp species

0 0 (b) (e) 40 Farming system x Habitat type* 200 Farming system* Habitat type**

30 150 Osmia rufa 20 100

10 50 Brood cells of

0 0 Brood cells of predators herbivores

Fig. 2. Species richness of wasps (a) and bees (c)20 Habitat type*** (f) 8 Habitat type (d), and brood cell numbers of wasps special- P = 0·06 ized on herbivores (b), of wasps specialized on spiders (c), of the most abundant bee spe- 15 6 cies Osmia rufa (e) and of other bees (without O. rufa) (f) in conventional (black bars) and organic (white bars) ﬁelds and adjacent fal- 10 4 low strips. Data of the two study years were averaged for each of the four site types per s- tudy area (conventional ⁄ organic ﬁeld cen- 5 2 tre ⁄ fallow strip). Means and standard errors Brood cells of other bees

are shown. Results are based on mixed-effec- Brood cells of predators spiders 0 0 ts models (see Table 2) with *P <0Æ05, Conv Org Conv Org Conv Org Conv Org **P <0Æ01, ***P <0Æ001. Field centre Fallow strip Field centre Fallow strip

the potential importance of cereal fields in providing those ered to be valuable for bees does not necessarily enhance bee food resources. abundance. Low (food-resource) quality of the adjacent crop Interestingly, the positive effect of organic field manage- habitat can reduce the quality of the non-crop habitat enor- ment also influenced adjacent fallow strips which did not dif- mously. fer in their management. Positive effects of organic farming Based on a previous study showing that eumenid wasps on adjacent habitats have been recorded in plants (Aude prefer flying along fallow strips instead of crossing conven- et al. 2004) and from pollinators (Feber et al. 2007; Holz- tionally managed cereal fields (Holzschuh et al. 2009) and schuh, Steffan-Dewenter & Tscharntke 2008). They may our results on landscape scale effects (see above), we expected result from the absence of agrochemical drift from fields into great differences between trap nests in field centres and fallow adjacent habitats (Marshall & Moonen 2002) and ⁄ or – in the strips for herbivore-predating wasps. Our data on local case of pollinators – from the benefits provided by the flower effects confirm this expectation for conventional fields, but resources in organic fields. For the bee O. rufa, positive not for organic fields. Organic farming increased the value of effects of organic farming on abundances in adjacent fallow wheat fields for predators of herbivores to the level of fallow strips were even stronger than positive effects of fallow strips strips. This is remarkable, because fields were more homoge- vs. field centres. This was in contrast to our expectation that neous than fallow strips, even despite the higher weed diver- fallow strips are generally better bee habitats than cereal sity in organic fields. Apparently, the hostility of fields, because they provide higher plant diversity and might conventional wheat fields does not result from the structure be perceived as less hostile than crop fields, which never pro- of a cereal monoculture per se, but rather from the lower vide nesting sites (Fussell & Corbet 1991; Ba¨ ckmann & Tiai- food availability compared to organic fields. nen 2002; Pywell et al. 2005, 2006; Feber et al. 2007). Our Food availability may influence the abundances in trap results indicate that a habitat type that is generally consid- nests in two ways: reproduction may be enhanced, if high

2009 The Authors. Journal compilation 2009 British Ecological Society, Journal of Animal Ecology, 79,491–500 498 A. Holzschuh, I. S. Dewenter & T. Tscharntke

Table 4. Final linear mixed-effects models describing the effects of female. In contrast, high species diversity can only be edge density and proportion of non-crop habitat (in landscape circles explained by a high number of females that colonized the trap with 500 m radius), farming system (organic vs. conventional), nests, but not by high reproduction rates per female. Thus, habitat type (field centre vs. fallow strips) and their interactions on adults presumably, prefer nests in the vicinity of abundant species richness and number of brood cells of bees food resources, so high numbers of brood cells may be mainly d.f. FPexplained by a preference for those nesting sites. Finally, we tested the hypothesis that parasitoids are more Total species richness of bees affected by agricultural intensification than their hosts. Para- Proportion of non-crop habitats 10 13Æ80Æ004 sitoid species richness and abundance were well-explained by Farming system 22 4Æ70Æ041 both host species richness and host abundance suggesting Habitat type 45 6Æ60Æ014 that parasitoids were most abundant and diverse in heteroge- Total brood cell number of bees Proportion of non-crop habitats 10 6Æ90Æ025 neous landscapes, organic sites and fallow strips. However, Farming system 22 4Æ10Æ056 parasitoids of herbivore-predating wasps and bees other than Habitat type 45 3Æ20Æ080 O. rufa were less sensitive to agricultural intensification and Brood cell number of Osmia rufa reduced habitat quality than their hosts resulting in margin- Proportion of non-crop habitats 10 4Æ50Æ058 Farming system 22 5Æ10Æ034 ally higher parasitism rates in conventional sites and field Brood cell number of other bees centres respectively. Only parasitoids of the bee O. rufa were Proportion of non-crop habitats 10 7Æ40Æ022 found to be more sensitive than their hosts resulting in mar- Habitat type 45 3Æ70Æ060 ginally lower parasitism rates in field centres than in fallow strips. Parasitism rates of trap-nesting bees and wasps Non-significant factors and interactions (P >0Æ1) were removed in a decreased with increasing distance from extensively managed stepwise backward procedure from the full model. grasslands in Germany and Switzerland (Tscharntke et al. 1998; Albrecht et al. 2007) and from forest edges in Indonesia food availability reduces the time spent on collecting food for (Klein et al. 2006) suggesting that parasitoids were disadvan- larval provision; colonization is enhanced, if colonizing taged when following their hosts into more disturbed habi- females prefer nests in the vicinity of flowering plants or tats. In contrast, parasitism rates in trap nests did not abundant prey (Klein, Steffan-Dewenter & Tscharntke 2004, decrease in smaller and more isolated orchard meadows in 2006; Albrecht et al. 2007). A high number of brood cells can central Europe (Steffan-Dewenter & Schiele 2008) and even be explained either by a high number of females that colo- increased with increasing habitat modification from rain for- nized the trap nest or by a high reproduction per colonizing est to rice paddies in Ecuador (Tylianakis, Tscharntke &

Table 5. Final linear mixed-effects models describing the effects of farming system (organic vs. conventional) and habitat type (ﬁeld centre vs. fallow strips) on numbers of parasitized brood cells, numbers of host brood cells, and resulting parasitism rates. Landscape factors and interactions did not explain additional variance (P >0Æ05)

Response variables

Number of parasitized brood Number of host brood cells Parasitism rate cells

Explanatory variables d.f. FP P d.f. FP

Herbivore-predating wasps Farming system NS (*) 16 3Æ50Æ079 Habitat type 45 6Æ00Æ018 ** NS Effect direction: farming system No effect Org > conv Org < conv Effect direction: habitat type Fallow > field Fallow > field No effect Spider-predating wasps Habitat type 45 13Æ2<0Æ001 *** NS Effect direction: habitat type Fallow > field Fallow > field No effect Osmia rufa Farming system 22 5Æ10Æ034 * NS Habitat type 45 6Æ40Æ015 NS 38 3Æ40Æ071 Effect direction: farming system Org > conv Org > conv No effect Effect direction: habitat type Fallow > field No effect Fallow > field Other bees Habitat type NS (*) 4 7Æ30Æ054 Effect direction: habitat type No effect Fallow > field Fallow < field

For comparison, P-values of the number of host brood cells are indicated by ***P <0Æ001, **P <0Æ01, *P <0Æ05 and (*)P <0Æ1, corresponding to the values shown in Tables 3 and 4.

2009 The Authors. Journal compilation 2009 British Ecological Society, Journal of Animal Ecology, 79,491–500 Landscape composition, conﬁguration and farming intensity 499

Lewis 2007) showing that generalist parasitoids did not suffer (QLK5-CT-2002-01495) coordinated by David Kleijn and the BESS as strongly from agricultural intensification as their hosts. (Biotic Ecosystem Services) project of Carsten Dormann. Based on theoretical considerations, specialists, not general- ists, of higher trophic levels are expected to be more affected References by disturbances than their hosts, suffering from smaller population sizes and patchy distributions (Holt et al. 1999; Albrecht, M., Duelli, P., Schmid, B. & Mu¨ ller, C.B. (2007) Interaction diversity within quantified insect food webs in restored and adjacent intensively man- Tscharntke et al. 2005). Local population dynamics of host aged meadows. Journal of Animal Ecology, 76, 1015–1025. species make them an unstable resource for species of higher Aude, E., Tybirk, K., Michelsen, A., Ejrnaes, R., Hald, A.B. & Mark, S. trophic levels. Theory predicts that specialist parasitoids (2004) Conservation value of the herbaceous vegetation in hedgerows – does organic farming make a difference? Biological Conservation, 118,467– depend on recolonization processes more strongly than their 478. hosts (Kruess & Tscharntke 1994). Our results underline that Ba¨ ckmann, J.-P.C. & Tiainen, J. (2002) Habitat quality of field margins in a changes in land use intensity and habitat quality can result in Finnish farmland area for bumblebees (Hymenoptera: Bombus and Psithy- rus). Agriculture, Ecosystems & Environment, 89, 53–68. changed trophic interactions, but the direction of changes are Baum, K.A., Haynes, K.J, Dillemuth, F.P. & Cronin, J.T. (2004) The matrix not generally predictable. Host species which are dominantly enhances the effectiveness of corridors and stepping stones. Ecology, 85, parasitized by rather specialized species such as the drosophi- 2671–2676. Bengtsson, J., Ahnstro¨ m, J. & Weibull, A.-C. (2005) The effects of organic agri- lid fly C. indagator (host: O. rufa) may be able to escape from culture on biodiversity and abundance: a meta-analysis. Journal of Applied parasitism by colonizing more disturbed habitats. In con- Ecology, 42, 261–269. trast, species parasitized by generalist species like the para- Clough, Y., Kruess, A., Kleijn, D. & Tscharntke, T. (2005) Spider diversity in cereal fields: comparing factors at local, landscape and regional scales. Jour- sitic wasp M. acasta appeared to suffer twice from nal of Biogeography, 32, 2007–2014. agricultural intensification: from direct negative effects and Clough, Y., Holzschuh, A., Gabriel, D., Purtauf, T., Kruess, A., Steffan-Dew- additionally from increased parasitism rates in more dis- enter, I., Kleijn, D. & Tscharntke, T. (2007) Alpha and beta diversity of ar- thropods and plants in organically and conventionally managed wheat turbed habitats. fields. Journal of Applied Ecology, 44, 804–812. Habitat loss, habitat fragmentation and reduced habitat Collinge, S.K. (1996) Ecological consequences of habitat fragmentation: impli- quality contributed to the reduction of diversity and abun- cations for landscape architecture and planning. Landscape and Urban Plan- ning, 36, 59–77. dance of cavity-nesting bees, wasps and parasitoids in agri- Cushman, S. A. & McGarigal, K. (2003) Hierarchical, multi-scale decomposi- cultural landscapes. However, differences in the perception tion of species–environment relationships. Ecological Applications, 17,637– of landscape and local factors among species groups indicate 646. Debinski, D.M. & Holt, R.D. (2000) A survey and overview of habitat frag- that effects of agricultural intensification, which take place at mentation experiments. Conservation Biology, 14, 342–355. several scales simultaneously, are not easily predictable. Diekotter, T., Haynes, K., Mazeffa, D. & Crist, T. (2007) Direct and indirect Non-crop nesting specialists do not necessarily perceive crop effects of habitat area and matrix composition on species interactions among flower-visiting insects. Oikos, 116, 1588–1598. fields as a hostile landscape matrix – at least if fields are less Fahrig, L. (1997) Relative effects of habitat loss and fragmentation on popula- intensively managed, and do not necessarily favour fallow tion extinction. Journal of Wildlife Management, 61, 603–610. strips, because high management intensity in the adjacent Fahrig, L. (2003) Effects of habitat fragmentation on biodiversity. Annual Review of Ecology, Evolution, and Systematics, 34, 487–515. crop field can reduce their quality. Perennial fallow strips and Feber, R.E., Johnson, P.J., Firbank, L.G., Hopkins, A. & Macdonald, D.W. other edge habitats are not only important for local recolon- (2007) A comparison of butterfly populations on organically and conven- ization processes from field edges into field centres (Landis, tionally managed farmland. Journal of Zoology, 273, 30–39. Fried, J., Levey, D. & Hogsette, J. (2005) Habitat corridors function as both Wratten & Gurr 2000), but facilitated even nest colonization drift fences and movement conduits for dispersing flies. Oecologia, 143,645– by wasps at the landscape scale. Interestingly, habitat con- 651. nectivity appeared to be enhanced by both higher edge densi- Fussell, M. & Corbet, S.A. (1991) Forage for bumble bees and honey bees in farmland: a case study. Journal of Apicultural Research, 30,87–97. ties (higher perimeter-area ratios) and higher percentage of Gathmann, A. & Tscharntke, T. (2002) Foraging ranges of solitary bees. Jour- organic farming. Conversion of crop land into non-crop hab- nal of Animal Ecology, 71, 757–764. itat may not be a sufficiently successful strategy to enhance Haddad, N.M. (1999) Corridor and distance effects on interpatch movements: a landscape experiment with butterflies. Ecological Applications, 9, 612–622. wasps or other species that suffer more from isolation than Haddad, N.M. & Tweksbury, J.J. (2005) Low-quality habitat corridors as loss of habitat. Consequently, related ecosystem services such movement conduits for two butterfly species. Ecological Applications, 15, as pollination and biological control by bees and wasps are 250–257. Harrison, S. & Bruna, E. (1999) Habitat fragmentation and large-scale conser- threatened by habitat loss and land use intensification at dif- vation: what do we know for sure? Ecography, 22, 225–232. ferent spatial scales. Hole, D.G., Perkins, A.J., Wilson, J.D., Alexander, I.H., Grice, P.V. & Evans, A.D. (2005) Does organic farming benefit biodiversity? Biological Conserva- tion, 122, 113–130. Acknowledgments Holt, R.D., Lawton, J.H., Polis, G.A. & Martinez, N.D. (1999) Trophic rank and the species–area relationship. Ecology, 80, 1495–1504. We thank the farmers for their participation in the project, Yann Holzschuh, A., Steffan-Dewenter, I., Kleijn, D. & Tscharntke, T. (2007) Diver- sity of flower-visiting bees in cereal fields: effects of farming system, landscape Clough for help with the site selection and for statistical advice, composition and regional context. Journal of Applied Ecology, 44,41–49. Doreen Gabriel for GIS-support, Gundula Kolb for help with the Holzschuh, A., Steffan-Dewenter, I. & Tscharntke, T. (2008) Agricultural land- dissection of trap nests, Reiner Theunert for help with species identi- scapes with organic crops support higher pollinator diversity. Oikos, 117, fication and Rachel Gibson, Emily Martin, Ken Norris and two 354–361. anonymous reviewers for their helpful comments. This research was Holzschuh, A., Steffan-Dewenter, I. & Tscharntke, T. (2009) Grass strip corridors in agricultural landscapes enhance nest site colonization by solitary carried out within the framework of the EU-funded project ‘EASY’ wasps. Ecological Applications, 19, 123–132.

2009 The Authors. Journal compilation 2009 British Ecological Society, Journal of Animal Ecology, 79,491–500 500 A. Holzschuh, I. S. Dewenter & T. Tscharntke

Klein, A.M., Steffan-Dewenter, I. & Tscharntke, T. (2004) Foraging trip dura- Schmidt, M.H., Roschewitz, I., Thies, C. & Tscharntke, T. (2005) Differential tion and density of megachilid bees, eumenid wasps and pompilid wasps in effects of landscape and management on diversity and density of ground- tropical agroforestry systems. Journal of Animal Ecology, 73, 517–525. dwelling farmland spiders. Journal of Applied Ecology, 42, 281–287. Klein, A.M., Steffan-Dewenter, I. & Tscharntke, T. (2006) Rain forest pro- Statistisches Bundesamt (2004) Landwirtschaft in Zahlen 2003, motes trophic interactions and diversity of trap-nesting hymenoptera in Wiesbaden. Available at: http://www.destatis.de/jetspeed/portal/cms/site/ adjacent agroforestry. Journal of Animal Ecology, 75, 315–323. destatis/Internet/DE/Content/Publikationen/Fachveroeffentlichungen/ Kremen, C., Williams, N.M., Bugg, R.L., Fay, J.P. & Thorp, R.W. (2004) The LandForstwirtschaft/LandundForstwirtscaft2003.property-file.pdf. Statis- area requirements of an ecosystem service: crop pollination by native bee tisches Bundesaunt, Wiesbade. communities in California. Ecology Letters, 7, 1109–1119. Steffan-Dewenter, I. (2002) Landscape context affects trap-nesting bees, wasps Kruess, A. & Tscharntke, T. (1994) Habitat fragmentation, species loss, and and their natural enemies. Ecological Entomology, 27, 631–637. biological-control. Science, 264, 1581–1584. Steffan-Dewenter, I. & Schiele, S. (2008) Do resources or natural enemies drive Landis, D.A., Wratten, S.D. & Gurr, G.M. (2000) Habitat management to con- bee population dynamics in fragmented habitats. Ecology, 89, 1375–1387. serve natural enemies of arthropod pests in agriculture. Annual Review of Trzcinski, M.K., Fahrig, L. & Merriam, G. (1999) Independent effects of forest Entomology, 45, 175–201. cover and fragmentation on the distribution of forest breeding birds. Ecolog- Marshall, E.J.P. & Moonen, A.C. (2002) Field margins in northern Europe: ical Applications, 9, 586–593. their functions and interactions with agriculture. Agriculture, Ecosystems Tscharntke, T. & Brandl, R. (2004) Plant–insect interactions in fragmented and Environment, 89, 5–21. landscapes. Annual Review of Entomology, 49, 405–430. McGarigal, K. & Cushman, S. A. (2002) Comparative evaluation of experi- Tscharntke, T., Gathmann, A. & Steffan-Dewenter, I. (1998) Bioindication mental approaches to the study of habitat fragmentation effects. Ecological using trap-nesting bees and wasps and their natural enemies: community Applications, 12, 335–345. structure and interactions. Journal of Applied Ecology, 35, 708–719. O¨ ckinger, E. & Smith, H.G. (2008) Do corridors promote dispersal in grassland Tscharntke, T., Klein, A.M., Kruess, A., Steffan-Dewenter, I. & Thies, C. butterflies and other insects? Landscape Ecology, 23, 27–40. (2005) Landscape perspectives on agricultural intensification and biodiver- Pinheiro, J.B. & Bates, D.M. (2000) Mixed-Effects Models in S and S-Plus. sity – ecosystem service management. Ecology Letters, 8, 857–874. Springer, New York, USA. Tylianakis, J., Tscharntke, T. & Lewis, O. (2007) Habitat modification Pywell, R.F., Warman, E.A., Carvell, C., Sparks, T.H., Dicks, L.V., Bennett, alters the structure of tropical host-parasitoid food webs. Nature, 445,202– D., Wright, A., Critchley, C.N.R. & Sherwood, A. (2005) Providing foraging 205. resources for bumblebees in intensively farmed landscapes. Biological Con- servation, 121, 479–494. Received 2 February 2009; accepted 2 November 2009 Pywell, R.F., Warman, E.A., Hulmes, L., Hulmes, S., Nuttall, P., Sparks, T.H., Handling Editor: Ken Wilson Critchley, C.N.R. & Sherwood, A. (2006) Effectiveness of new agri-environment schemes in providing foraging resources for bumblebees in intensively farmed landscapes. Biological Conservation, 129, 192–206. R Development Core Team (2005) R: A Language and Environment for Statisti- Supporting information cal Computing. Foundation for Statistical Computing, Vienna, Austria. Ritchie, L., Betts, M., Forbes, G. & Vernes, K. (2009) Effects of landscape com- Additional Supporting Information may be found in the online ver- position and configuration on northern flying squirrels in a forest mosaic. sion of this article. Forest Ecology and Management, 257, 1920–1929. Roland, J. & Taylor, P. (1997) Insect parasitoid species respond to forest struc- Appendix S1. Variance components analysis showing the importance ture at different spatial scales. Nature, 386, 710–713. of the random factors ‘study area’, ‘year’ and ‘farming system’ in the Roschewitz, I., Gabriel, D., Tscharntke, T. & Thies, C. (2005) The effects of models presented in Table 2. Variances are expressed as percentages landscape complexity on arable weed species diversity in organic and con- of the total variance. ventional farming. Journal of Applied Ecology, 42, 873–882. Rundlo¨ f, M. & Smith, H.G. (2006) The effect of organic farming on butterfly As a service to our authors and readers, this journal provides sup- diversity depends on landscape context. Journal of Applied Ecology, 43, porting information supplied by the authors. Such materials may be 1121–1127. Rundlo¨ f, M., Nilsson, H. & Smith, H.G. (2008) Interacting effects of farming re-organized for online delivery, but are not copy-edited or typeset. practice and landscape context on bumblebees. Biological Conservation, 141, Technical support issues arising from supporting information (other 417–426. than missing files) should be addressed to the authors.

2009 The Authors. Journal compilation 2009 British Ecological Society, Journal of Animal Ecology, 79,491–500