Biological Conservation 172 (2014) 56–64

Contents lists available at ScienceDirect

Biological Conservation

journal homepage: www.elsevier .com/locate /b iocon

Landscape composition and configuration differently affect trap-nestingbees, wasps and their antagonists

http://dx.doi.org/10.1016/j.biocon.2014.02.0150006-3207/� 2014 Elsevier Ltd. All rights reserved.

⇑ Corresponding author. Tel.: +49 551 39 33907.E-mail address: julianesteckel@yahoo.de (J. Steckel).

Juliane Steckel a,⇑, Catrin Westphal b, Marcell K. Peters a, Michaela Bellach a, Christoph Rothenwoehrer b,Stefan Erasmi c, Christoph Scherber b, Teja Tscharntke b, Ingolf Steffan-Dewenter a

a Department of Animal Ecology and Tropical Biology, Biocenter, University of Wuerzburg, Am Hubland, 97074 Wuerzburg, Germanyb Agroecology, Georg-August-University Goettingen, Grisebachstrasse 6, 37077 Goettingen, Germanyc Department of Cartography GIS & Remote Sensing, Institute of Geography, Goldschmidtstr. 5, 37077 Goettingen, Germany

a r t i c l e i n f o

Article history:Received 8 July 2013Received in revised form 4 February 2014Accepted 11 February 2014Available online 7 March 2014

Keywords:Agricultural intensificationEcosystem functioningLandscape heterogeneityManagement schemesSpatial scalesTrophic interactions

a b s t r a c t

Intensification of agriculture reduces heterogeneity at local and landscape levels and thereby impact bio-diversity and ecosystem processes. We studied a host-antagonist system of cavity-nesting bees, waspsand their antagonists and hypothesised that hosts and antagonists show different responses to localland-use intensity, the diversity of landscape in terms of composition and the spatial structure of land-scape in terms of configuration.

In a highly replicated study, we established nesting resources on 95 grasslands in three geographicregions across Germany and measured species richness and abundance of hosts (bees and wasps) andtheir antagonists, and rates of parasitism. For each site, we quantified local land-use intensity as wellas landscape heterogeneity in terms of composition and configuration at spatial scales ranging from250 m to 2000 m.

Increasing landscape heterogeneity enhanced species richness, abundance and parasitism rate,whereas local land-use intensity only marginally negatively affected total abundance. Bee and waspabundance as well as wasp species richness were enhanced by landscape composition at 250 m, whereastheir antagonists were enhanced by landscape configuration at 1500 m.

In conclusion, landscape composition and configuration affect trophic levels differently and are morerelevant than local land-use intensity. Solitary bees and wasps, which offer important pollination andpest control services, could be supported by enhancing landscape diversity, while their antagonists couldbenefit from measures that promote landscape connectivity. Hence, scale-dependent and trophic groupspecific conservation management schemes are required, that address different components of landscapeheterogeneity to enhance functional diversity and trophic interactions in agricultural landscapes.

� 2014 Elsevier Ltd. All rights reserved.

1. Introduction

Interacting species of different trophic levels may respond todifferent components of agricultural intensification and at differ-ent spatial scales (Holland et al., 2004; Kruess, 2003; Tscharntkeet al., 2012). The negative effects of agricultural intensification onbiodiversity and biotic interactions on local and landscape levelswere the focus of several studies (Hendrickx et al., 2007; Karpet al., 2012; Tscharntke et al., 2005) but little is known if intensifi-cation affects trophic levels differently and thereby may disruptbiotic interactions (Holt et al., 1999; Rand et al., 2012; Thieset al., 2003).

Moreover, different components of landscape heterogeneity,such as the composition and configuration of landscape, are ex-pected to have distinct effects on different functional groups orecosystem processes, but this remains largely unexplored (Fahriget al., 2011; Holzschuh et al., 2010; Kennedy et al., 2013). Whilecomposition reflects the number and proportions of different hab-itat types in a landscape, configuration refers to the spatialarrangement of habitats and their shapes (Fahrig et al., 2011; Liand Reynolds, 1995). Bee abundance and species richness for in-stance, is enhanced by landscape composition (percentage ofnon-crop habitats), whereas wasps benefit from high edge density,i.e. landscape configuration (Holzschuh et al., 2010). Still, the roleof landscape composition versus configuration for species richnessand biotic interactions at different trophic levels remains unclear.Furthermore, species may respond to landscape heterogeneity at

J. Steckel et al. / Biological Conservation 172 (2014) 56–64 57

different spatial scales depending on species-specific dispersal andforaging distances (Steffan-Dewenter, 2002). For example bodysize (Greenleaf et al., 2007; Westphal et al., 2006), trophic level(Thies et al., 2005, 2003) and resource or habitat specialisation(Tscharntke et al., 2005) may determine scale-dependent re-sponses to landscape heterogeneity. Moreover, specialists of highertrophic levels are assumed to be more vulnerable to habitat frag-mentation and reduced landscape heterogeneity than their hosts(Brueckmann et al., 2011; Chaplin-Kramer et al., 2011; Holt et al.,1999; Rand et al., 2012). Antagonists also have more complexrequirements because they have to synchronise their activities inspace and time with host abundance (Durrer and Schmid-Hempel,1995; Steffan-Dewenter, 2003). Structurally diverse landscapeswith high connectivity between habitats could improve thechances of finding habitats with host populations, thereby particu-larly benefiting higher trophic levels. Semi-natural habitats mayoffer overwintering sites or host species for natural enemies inadjacent crop fields (Rand et al., 2006) and thus serve as refugehabitats in agricultural landscapes. Calcareous grasslands arenotably one of the most species-rich habitats in Central Europe(Poschlod and WallisDeVries, 2002; van Swaay, 2002). Here, wechose grassland habitats of different land-use intensities to studyland-use effects on arthropods. The management of grasslandsdetermines vegetation structure and richness and thus the avail-ability of resources for arthropod communities (Borer et al.,2012; Socher et al., 2012). Generally, there is a lack of studiesdealing with insect diversity and biotic interactions in grasslandhabitats (Tscharntke et al., 2012) although grassland habitatsaccount for 29% of the farmed area in Germany (http://www.bmelv-statistik.de) and 40.5% of the terrestrial area of theworld (http://www.fao.org).

We used cavity-nesting bees, wasps and their antagonists intrap nests as a model system to study the responses of differentfunctional groups to local grassland management intensity andlandscape heterogeneity. Trap-nesting arthropod species can serveas biodiversity indicator taxa and provide exceptional insights intomultitrophic biotic interactions (Steffan-Dewenter and Schiele,2008; Tscharntke et al., 1998; Westphal et al., 2008). Hosts in thissystem are solitary bees, serving as pollinators of wild plants andinsect-pollinated crops and predatory wasps that fulfil a crucialrole as predators of pest insects (Klein et al., 2004). These host spe-cies depend on different habitat types within their foraging rangefor food supply and nest building (Westrich, 1996).

Due to logistic constraints landscape-scale studies are oftenconducted in only one study region. However, to allow more gen-eral conclusions about impacts of different factors of landscapeheterogeneity on functional biodiversity, a replication of studiesin several regions is desirable (Fahrig et al., 2011; Holzschuhet al., 2007). Here, we present results from a well replicated studyconducted in 95 study plots in three distinct regions in Germany(Fischer et al., 2010, http://www.biodiversity-exploratories.de)covering three spatial levels in our study: (1) local level of studyplots, (2) landscape level represented by eight spatial scales(250 m up to 2000 m radius around study plots), and (3) regionallevel (represented by three geographic regions across Germany).Within the framework of our study the following questions wereaddressed: (1) What is the relative importance of local land-useintensity versus landscape heterogeneity for bees, wasps and theirantagonists? and (2) Are there different responses of hosts (beesand wasps) and their antagonists to landscape composition andconfiguration and are these responses scale-specific? Related tothese questions, we tested the following hypotheses:

i. Species richness and abundance of hosts (bees and wasps)and their antagonists are negatively correlated with localland-use intensity.

ii. Bees, wasps and their antagonists are enhanced by increas-ing landscape heterogeneity.

iii. Bees and wasps are more strongly affected by landscapecomposition and antagonists by landscape configuration.

iv. The patterns found are independent from the study region.

2. Materials and methods

2.1. Study plots

The study was conducted within the framework of the DFG-funded project ‘Biodiversity Exploratories’ (Fischer et al., 2010).The Exploratories are represented by three research regions in Ger-many (the Biosphere Reserve Schorfheide-Chorin to the NationalPark Hainich-Dün to the Biosphere Reserve Schwaebische Alb,henceforth referred to as Schorfheide, Hainich and Alb, http://www.biodiversity-exploratories.de). We established 3.5 m � 15 mstudy plots on the study sites of the Exploratories (KML-file A1,Table A2). The study sites of the Exploratories differed in theirland-use intensities, ranging from extensively managed calcareousgrasslands to intensively used pastures and meadows with highmowing or grazing frequencies or both. The study plots withinthe study sites were fenced with electric wire when necessary toexclude cattle.

2.2. Trap nests

We constructed 760 trap nests using PVC tubes of 10.5 cmdiameter, filled with reed internodes of Phragmites australis (Cav.)Trin. To sample the entire community of cavity-nesting species,we used reed of internodes with different diameters (0.2–1.2 cm)(Gathmann et al., 1994). At each study plot four wooden poleswere placed in a staggered pattern with a distance of 4 m. On eachpole, two trap nests were mounted at 1.5 m height. Trap nests wereinstalled between the middle of April and the middle of May 2008and recollected at the end of September and beginning of October2008. The traps were stored outside in a dry, unheated cabin to letthe animals develop under natural conditions. After a diapause of amonth, which served as a cold impulse to develop, we started todissect nests of bees and wasps in an early developmental stageto be able to record exact numbers of parasitized brood cells, cellswithout content due to predation and cells with dead offspring ofdifferent developmental stages (Gathmann and Tscharntke, 1999;Westphal et al., 2008). For identification to species level, nestswere closed again and then stored at room temperature untilhatching of imagoes.

Altogether, we quantified nine response variables: (1) the totalnumber of brood cells, hereafter referred to as total abundance, (2)number of brood cells of bees and (3) wasps, (4) number of broodcells of antagonists, (5) total species richness, (6) number of beespecies, (7) number of wasp species, (8) number of antagonist spe-cies and (9) parasitism rate. Parasitism rate was calculated bydividing the number of brood cells attacked by antagonists perstudy plot by the total number of brood cells per study plot. Emptynests of multivoltine species were not taken into account for theabundance data. For species richness data, individuals from a studyplot that could only be classified to higher taxonomic ranks, likegenus or family rank, were only counted as additional species incase there was no other species representing that genus or familyfrom the study plot.

2.3. Metrics of local land-use intensity

Local land-use intensity was assessed by annual questionnairesand interviews with land-users and land owners (Fischer et al.,2010). Based on this information, we calculated for each

A

A

A

AA A

A A A

A

B

AB

A

ABB

A

AB

B

A

A A

(a)

(b)

(c)

Fig. 1. (a) Effects of region on means (±SE) of bee, wasp and antagonist speciesrichness on plot level. (b) Effects of region on means (±SE) of bee, wasp andantagonist brood cell numbers on plot level. (c) Effects of region on means (±SE) ofparasitism rate on plot level. Different letters indicate significant differencesbetween means (a = 0.0167, post hoc tests).

58 J. Steckel et al. / Biological Conservation 172 (2014) 56–64

Exploratory study site a compound, additive index of land-useintensity, hereafter termed LUI (Supplement A3; Blüthgen et al.,2012). The LUI enabled the incorporation of the quantity of eachcomponent such as livestock units per hectare and duration ofgrazing period, number and time of mowing events and numberand amount of fertilizer applications in one continuous index.The species number of flowering plants (FP) was calculated asthe arithmetic average of two to three individual measurements(between May and September) of the species number of floweringplants on each of the 3.5 m � 15 m study plots.

2.4. Landscape heterogeneity metrics

We mapped different land-use types in the field within a radiusof 2000 m around all study plots in 2008 and 2009. This maximumdistance was chosen due to flight and foraging distances reportedfor solitary bees (Gathmann and Tscharntke, 2002; Zurbuchenet al., 2010) and relevant landscape scales in previous studies(Holzschuh et al., 2007; Steffan-Dewenter and Schiele, 2008). Fordigitalization we used high resolution aerial photographs (nominalspatial resolution 0.4 m) and topographic maps (1:10,000). Using aGeographical Information System (ArcGIS™ 9.3, ESRI) we classifiedeight general land-use types: arable land, forest, grassland, semi-natural habitat, road, woodland, settlement and water bodies.Intensively managed pastures and meadows were defined as grass-land. Semi-natural habitats consisted of habitats of extensive land-use, such as extensively managed meadows, marshland, shrubland,hedges (>5 m width), calcareous grasslands and orchards. Due toexpected scale-specific responses of our study organisms, we cal-culated metrics of landscape heterogeneity in the surroundinglandscape of the study plots within concentric circles of 250 m,500 m, 750 m, 1000 m, 1250 m, 1500 m, 1750 m and 2000 m ra-dius. Calculations were performed using the software FRAGSTATS3.3 (McGarigal et al., 2002) and were based on raster maps of3 m � 3 m grid cells. Landscape composition was characterisedusing Shannon’s Diversity Index (McGarigal et al., 2002; Shannonand Weaver, 1949). When the landscape consists of only oneland-use type, the Shannon’s Diversity Index equals 0, indicatingno diversity in land-use types. With eight distinct land-use typesit increases to a maximum of 2.08 in case all land-use types haveequal cover in the landscape. As a second metric of landscape com-position, we calculated the percentage of semi-natural habitats inthe landscape surrounding the study plots. It ranges from 0% (nosemi-natural habitats within the landscape) to 100%, when the en-tire landscape only consists of semi-natural habitats. The Shape In-dex was used as metric of landscape configuration (McGarigalet al., 2002). For each distinct patch (continuous area of oneland-use type within the landscape) the ratio of patch perimeterdivided by the minimum perimeter possible for a maximally com-pact patch (a raster square) of the corresponding patch area wascalculated. The Shape Index is the median of those ratios and a con-venient solution of the size bias of the perimeter-area ratio indexby adjusting for a square standard. When the Shape Index equalsone, all patches within the landscape are maximally compact whilehigher Shape Indices characterize more complex shapes of habitatpatches (Forman, 1995; McGarigal et al., 2002).

2.5. Statistical analyses

We used the software R 2.14.0 for Windows (R DevelopmentCore Team, 2012) for statistical analyses. First, we compared localand landscape metrics among regions using Analysis of Variance(ANOVA) (Table A4). We performed t-tests with Bonferroni correc-tion (a = 0.0167) to test for significant differences of local and land-scape metrics between regions at P < 0.05 (Table A4). Second, wetested for regional differences of the response variables species

richness and abundance of bees, wasps and antagonists by per-forming ANOVA analysis with region as single explanatory variable(Fig. 1).

Third, we analysed the effects of the local level metrics land-useintensity and floral richness, as well as their interaction on the nineresponse variables mentioned above. We constructed ordinary lin-ear models. Models of antagonist richness and abundance includedhost species richness and abundance, respectively, as covariates.Additionally, we added landscape level metrics, such as

J. Steckel et al. / Biological Conservation 172 (2014) 56–64 59

composition, configuration and region as explanatory variables tothe models already including local variables. To test for divergenteffects of these variables among regions, we included interactionterms with region for each explanatory variable. We ln-trans-formed count data and arcsine square-root transformed percent-age data when necessary to fulfil the assumption of normality ofresiduals (Crawley, 2007; Sokal and Rohlf, 1994). Species numberof flowering plants was square root transformed as it generally in-creased the fit of models to the data. Collinearity among theexplanatory variables was checked using variance inflation factorsand Pearson correlation coefficients. Using thresholds of <3 for thevariance inflation factors and <0.65 for the Pearson correlationcoefficients, none of the explanatory variables were found to cov-ary as strongly with other variables to prohibit their parallel usein models (Zuur et al., 2010, 2009). We used the dredge functionof the R package MuMIn to automatically construct all possiblemodels based on the set of explanatory variables in the full model,including the null model, and to identify a minimum-adequatemodel using the Akaike Information Criterion (AICc) for modelevaluation (Burnham and Anderson, 2004). First, minimum-ade-quate models solely based on local level factors were identified.Second, minimum-adequate models including local and landscapefactors as well as region were identified for each of the eight land-scape scales to estimate the scale on which each process and func-tional group responded. The landscape scale for each responsevariable was chosen by AICc comparison of the minimal adequatemodel of each scale. The model that was best supported by the databased on the lowest AICc was chosen. The improvement of modelfit due to consideration of landscape level factors could then beevaluated by comparing AICc values of the best models for all com-binations of factors on local, landscape and regional level(Table A5). Following Kissling and Carl (2008), spatial autocorrela-tion was tested calculating Moran’s I values for distance classes be-tween 1 km and 10 km. Where spatial autocorrelation wasdetected, we additionally calculated spatial linear models. The spa-tial linear models were calculated as ‘spatial simultaneous autore-gressive error models’ using the R library ‘spdep’, v. 0.5-41 (Bivand,2012). To compare the effect sizes of predictor variables, we z-transformed all continuous variables in the final models by thefunction scale (Package ‘base’, version 2.15.1).

3. Results

A total of 3672 nests with 19,603 brood cells were collectedyielding 12,786 individuals, of which 8070 were males and 4716females. Most of the brood cells were occupied by host species(18,082). Antagonists were found in 1390 of the brood cells and103 cells harboured generalist predators. Altogether, 74 speciescould be identified, comprising 47 host species (24 wasps, 23 bees)and 27 antagonists. The 27 antagonist species consisted of threebee species, two predatory beetles, two parasitoid flies and 20 spe-cies of parasitoid wasps (Table A6). The total number of speciesand composition of communities varied between regions. The Albcomprised 51, Hainich 35 and Schorfheide 43 species. Alb andSchorfheide overlapped in 24 species, Alb and Hainich in 24 speciesas well, Hainich and Schorfheide shared 25 same species(Table A6).

3.1. Regional differences in hymenoptera communities and land-useintensity

Total species richness varied significantly between regions(F2,92 = 5.054, P = 0.008). This was due to significant differences inhost species richness (F2,92 = 8.779, P < 0.001), whereas antagonistspecies richness did not vary significantly between regions

(F2,92 = 0.879, P = 0.419) (Fig. 1a). Bee and wasp abundance as wellas the abundance of antagonists (measured as the number of broodcells per study plot) did not vary significantly between regions(Fig. 1b), whereas parasitism rates varied significantly between re-gions (F2,92 = 3.74, P = 0.027; Fig. 1c).

The range of the local land-use index (LUI) was similar in allthree regions. However, the three regions partly varied in flower-ing plant richness and landscape metrics (Shape Index, percentageof semi-natural habitats) depending on the considered spatialscale. For all tested landscape scales, Shannon’s Diversity Indexdid not significantly differ among regions (Table A4).

3.2. Local and landscape level effects

Minimum-adequate models including landscape factors and re-gion as explanatory variables were generally better supported interms of AICc values than models incorporating only local levelvariables (Table A5), underscoring the importance of landscapeand regional scale factors in determining diversity, abundanceand the strength of host-antagonist-interactions in grassland ani-mal communities. For models in which both local and landscape le-vel variables were retained in the final minimum-adequatemodels, standardised effect sizes of most landscape metrics and re-gion are higher than local level variables (Table 2), suggesting ahigher importance of large level variables compared to those mea-sured at small spatial level. For each response variable the selectedminimum-adequate model is given in Table 2. In the following, wepresent the effect directions and strengths of variables of the min-imum-adequate models. Total species richness and wasp speciesrichness were best explained by landscape composition (SHDI)and the percentage of semi-natural habitats at a radius of 250 m(Tables 1 and 2). The percentage of semi-natural habitats affectedtotal species richness less positively than the Shannon’s DiversityIndex with an effect size of 22% and 36% respectively. Total speciesrichness was marginally significantly enhanced by the speciesnumber of flowering plants.

For species richness of bees no explanatory variable was signif-icant across all three regions. In contrast to hosts, species richnessof antagonists responded to larger landscape scales and to differentlandscape metrics. Antagonist species richness and abundance in-creased at a scale of 1500 m with configurational complexity(Shape Index) of the surrounding landscape.

Bee species richness was differently affected by semi-naturalhabitats at a scale of 1000 m, depending on region (Fig. 2a): Inthe Alb the effect was marginally positively (N = 33, P = 0.052), inthe Hainich it was not significant (N = 28, P > 0.1) and in the Scho-rfheide it was negative (N = 34, P < 0.05). Due to these contradict-ing results, we differentiated the dominating types of semi-natural habitats in each region (Table A7). We found that semi-nat-ural habitats within 1000 m around the study plots in the Albmainly comprised calcareous (5.35%) and extensively used(6.08%) grasslands, in the Hainich extensively used grassland(6.44%) and shrubland (4.26%) and in the Schorfheide wetland(4.88%).

Total abundance marginally significantly decreased withincreasing local land-use intensity and increased at the landscapelevel with Shannon’s Diversity Index. The final model for bee abun-dance included Shannon’s Diversity Index at a landscape scale of250 m and region. Bee abundance was enhanced by Shannon’sDiversity Index in Hainich (N = 28, P < 0.01) and Schorfheide(N = 34, P < 0.001). The effect in the Alb was not significant(N = 33, P > 0.1). Local level variables were not part of the finalmodel (Fig. 2b).

Wasp species profited more from semi-natural habitats (42%)than from Shannon’s Diversity Index (36%) at a landscape scale of250 m (Fig. 2c). No further predictors remained in the final model.

Table 1Overview over minimal adequate models for each response variable. Multiple R2 and AICc values are given for Ordinary-Least-Squares-Models (OLS). Moran’s I values refer to thesmallest spatial scale of 1 km and indicate the degree of spatial autocorrelation of residuals derived from the OLS model. First local explanatory variables are given (LUI = land-useindex, FP = species number of flowering plants, SR host = species richness of hosts, BC host = brood cell number of hosts), then the identified adequate landscape scale for thefollowing landscape factors (SHDI = Shannon’s Diversity Index, SHAPE = median of the Shape Index, % SNH = percentage of semi-natural habitats). Arrows indicate that a variablewas part of the minimal adequate model and give the effect direction. Significant interactions between region and any other explanatory variable are shown by differing arrowsbetween regions. (grey, A: Alb, black, H: Hainich, dashed, S: Schorfheide). Thinner arrows indicate non-significant effects (P > 0.05). The incorporation of region in the minimum-adequate model is indicated by a circle.

OLS Model

Response variable R2 AICc Moran’s I Explanatory variable

Local level variables Scale (m) Landscape level variables Region

Species richness LUI FP SR host BC host SHDI SHAPE % SNH

Total species richness 0.2862 494.49 0.368 250

Species richness of bees 0.1858 320.91 0.326 1000 A H S

Species richness of wasps 0.3747 81.24 0.251 250

Species richness of antagonists 0.4036 124.48 0.136 1500

AbundanceTotal number of brood cells 0.3090 271.06 0.314 250

A H SBrood cell number of bees 0.2234 306.07 0.427 250

A H SBrood cell number of wasps 0.3072 274.63 0.009 250

Brood cell number of antagonists 0.3862 278.02 0.091 1500

Parasitism rate 0.2303 101.01 �0.016 1500

60 J. Steckel et al. / Biological Conservation 172 (2014) 56–64

Also wasp abundance was determined by the same predictor vari-ables at 250 m (Fig. 2 d). It was enhanced by the percentage ofsemi-natural habitats to the same extent as Shannon’s Diversity In-dex (see Table 2).

In contrast, species richness and abundance of antagonists aswell as parasitism rate were neither correlated with Shannon’sDiversity Index, nor the percentage of semi-natural habitats butwere significantly enhanced by the Shape Index (Fig. 2e, and f; Ta-bles 1 and 2). Species richness of antagonists additionally re-sponded positively to species richness of hosts with a relativecontribution of 64%. Likewise, the abundance of hosts also had asignificant positive impact on the abundance of antagonists witha high relative contribution of 52% (Table 2). Antagonist abundancefurther profited from the species number of flowering plants –though to a smaller extent (17%). Moreover, local diversity ofantagonists was positively correlated with parasitism rate (lm:F1,93 = 30.86, P < 0.001).

Generally, Moran’s I correlograms revealed spatial autocorrela-tion in the minimal adequate models for total, bee and wasp spe-cies richness as well as total and bee abundance (Table 1).However, corrected SARerr models provided very similar resultsso that it was justified to rely on Ordinary-Least-Squares-Modelswithout correction (Tables 2, A8 and A9).

4. Discussion

In this study we aimed to disentangle effects of local land-useintensification, landscape heterogeneity across spatial scales andgeographic region on interacting species groups of solitary beesand wasps and their antagonists to be able to draw up scale- andtaxon-specific conservation schemes. Generally, landscape levelwas more relevant than the local level in explaining patterns ofspecies richness, abundance and biotic interactions. As hypothe-sised, total species richness and total abundance were enhancedby landscape heterogeneity, in particular by Shannon’s DiversityIndex and the percentage of semi-natural habitats, whereas at locallevel increasing land-use intensity had a marginally significantlynegative effect on total abundance, though the effect size was

low. Thus, low local flower cover and vegetation disturbance bymowing or grazing as well as low plant richness due to frequentfertiliser applications create unfavourable local conditions. How-ever, these local impacts could be mitigated by small-scale land-scape heterogeneity and provision of additional food resourcesfor bees and wasps in the surrounding landscape.

At the landscape level, richness and abundance of bees andwasps were positively affected by a high Shannon’s Diversity Indexand a high percentage of semi-natural habitats (landscape compo-sition), whereas local level predictors were not included in theminimum-adequate models. Thus, bees and wasps rather dependon landscape level predictors than on local level ones (Swiftet al., 2004; Tscharntke et al., 2005). If a variety of different habi-tats, like hedges with potential nesting opportunities and food re-sources, were reachable for these central-place-foragers, theyoccurred in higher abundances and species numbers. Therefore,their occurrence and thus species richness was mediated by land-scape characteristics. Likewise, for other insect taxa landscape het-erogeneity was found to be more important than the local farmingsystem (Fahrig et al., 2011; Tscharntke et al., 2012; Weibull et al.,2000).

We found that bee species responded differently to the percent-age of semi-natural habitats, depending on region. The negativecorrelation of semi-natural habitats in the Schorfheide could be ex-plained by the high amount of wetlands within the landscape type‘semi-natural habitat’. In contrast, in the regions Alb and Hainich,semi-natural habitats were dominated by calcareous and exten-sively used grasslands. Semi-natural habitats surrounding agricul-tural habitats are considered to serve as source for (re)colonisationof managed areas (Tscharntke et al., 2012). Since bees profit morefrom calcareous and extensively used grasslands than from wet-lands, bee diversity was presumably enhanced in the Alb but notin the Schorfheide (Osborne et al., 1991; Steffan-Dewenter et al.,2002).

Importantly, we found contrasting responses of host species(bees and wasps) and antagonists to components of landscape het-erogeneity. Our results indicate that antagonists benefit more fromlandscape configuration than from landscape composition. Speciesrichness and abundance of antagonists, as well as parasitism rate,

Table 2Model estimates of minimal adequate OLS models for each response variable on the corresponding landscape scale. Explanatory variables incorporated in each model are listedfor each response variable individually. Standardised estimates are given indicating the relative contribution of each variable. Abundances were ln-transformed as well asparasitism rate and species richness of antagonists, % semi-natural habitats was arcsine-square-root-transformed. FP was square-root-transformed. SHDI = Shannon’s DiversityIndex, SHAPE = Median of the Shape Index.

Response variable Scale (m) Explanatory variable Standardised estimate Estimate t p

Species richness

Total species richness 250 Intercept <0.0001 1.9295 1.454 0.1494SHDI index 0.3552 4.0905 3.899 0.0002***

% Seminat. habitat 0.2154 2.7971 1.991 0.0495*

Number of plant species 0.1629 0.7884 0.512 0.1273

Species richness of bees 1000 Alb 0.3589 2.9677 7.026 <0.0001***

Hainich �0.0748 3.3039 0.497 0.6204Schorfheide �0.4586 4.4625 1.884 0.0628Alb: % seminat. habitat 0.2457 2.2144 1.846 0.0682Hainich: % seminat. habitat �0.1066 �0.9602 �1.657 0.1010Schorfheide: % seminat. habitat �0.7479 �6.7396 �3.247 0.0017**

Species richness of wasps 250 Intercept <0.0001 0.4292 3.964 0.0001***

SHDI index 0.3604 0.5098 4.251 <0.0001***

% Seminat. habitat 0.4177 0.6661 4.927 <0.0001***

Species richness of antagonists 1500 Alb �0.1878 �1.0246 �1.625 0.1077Hainich �0.2266 �1.0465 �0.185 0.8537Schorfheide 0.3689 �0.7092 2.558 0.0122*

SHAPE index 0.1807 0.7226 2.094 0.0391*

Spec. richn. of hosts 0.6393 0.1532 7.195 <0.0001***

AbundanceTotal number of brood cells 250 Alb 0.0124 5.0010 7.509 <0.0001***

Hainich �0.0175 3.3451 �2.022 0.0462*

Schorfheide 0.1708 3.1174 �2.444 0.0165*

Alb: SHDI index 0.0127 0.0444 0.068 0.9463Hainich: SHDI index 0.5521 1.9217 2.233 0.0281*

Schorfheide: SHDI index 0.6974 2.4271 2.823 0.0059**

LUI index �0.1806 �1.2056 �1.954 0.0538

Brood cell number of bees 250 Alb �0.1400 4.4280 5.759 <0.0001***

Hainich 0.1631 3.0508 �1.420 0.1592Schorfheide 0.1972 2.2422 �2.343 0.0214*

Alb: SHDI index �0.1155 �0.4619 �0.583 0.5610Hainich: SHDI index 0.395 1.5786 2.047 0.0436*

Schorfheide: SHDI index 0.6416 2.5645 2.962 0.0039**

Brood cell number of wasps 250 Intercept <0.0001 1.2699 3.790 0.0003***

SHDI index 0.3581 1.4468 3.891 0.0002***

% Seminat. habitat 0.3447 1.5319 3.746 0.0003***

Brood cell number of antagonists 1500 Intercept <0.0001 �5.5402 �3.998 0.0001***

SHAPE index 0.2817 2.5216 3.427 0.0009***

Number of plant species 0.1665 0.2781 2.027 0.0456*

Brood cell number of hosts 0.5213 0.5855 6.344 <0.0001***

Parasitism rate 1500 Alb 0.2859 �3.7989 �6.916 <0.0001***

Hainich �0.4116 �4.1101 �3.044 0.0031**

Schorfheide 0.0615 �3.8990 �0.986 0.3265SHAPE index 0.4063 1.2794 4.169 <0.0001***

*** Signif. codes: P < 0.001.** Signif. codes: P < 0.01.* Signif. codes: P < 0.05.

J. Steckel et al. / Biological Conservation 172 (2014) 56–64 61

were enhanced by high landscape configuration (Shape Index) at alarger landscape scale (1500 m). These findings are in accordancewith studies on local prey/host losses that revealed that antago-nists depend more than their hosts on habitat connectivity(Tscharntke and Kruess, 1999). Antagonists require habitats withphenologically matching occurrence of (certain) hosts, which wasalso reflected by the high effect sizes found for the predictor vari-ables host species richness and abundance, respectively, in themodels for antagonist species richness and abundance. Presum-ably, the probability to find suitable habitats is higher in land-scapes with sufficient non-arable habitats and edge structuresthat connect potential host habitats. Connective elements and edgestructures should not only maintain dispersal of antagonists butalso supply natural nesting resources in hedges or forest edgesfor potential hosts. Antagonists attracted by these conditions con-sequently occur in higher abundances if the Shape Index is high.

The abundance of antagonists was enhanced by species num-ber of local flowering plants, contrary to findings of a recentstudy of (Ebeling et al., 2012). An enhanced floral richness mightalso imply a structural richness of the vegetation, offering foodsupply and shelter for predators and parasitoids and for antago-nists that directly depend on flowers for feeding on nectar (Wes-trich, 1989).

The analysis of multiple landscape scales (Scherber et al., 2012)was useful to account for different dispersal and foraging distancesof species groups that might respond to landscape parameters atdifferent spatial scales. The antagonists responded to larger scales(1500 m) than their lower trophic level hosts (wasps 250 m, bees1000 m) (Holt, 1996). The dependency of antagonists on host pop-ulations that fluctuate in time and space, might result in higherdispersal rates of antagonists (Thies et al., 2003) to compensatefor local resource limitation. Indeed, species richness and

Fig. 2. Relationship between (a) bee species richness and landscape composition (SHDI) in three regions. (b) Bee abundance (ln-transformed) and landscape composition(SHDI) in three regions. (c) Wasp species richness (ln-transformed) and landscape composition (SHDI). (d) Wasp abundance (ln-transformed) and landscape composition(SHDI). (e) Species richness of antagonists (ln-transformed) and landscape configuration (Shape Index) in three regions. (f) Abundance of antagonists (ln-transformed) andlandscape configuration (Shape Index) in three regions. In case no significant differences between regions were found, region was not incorporated in the final model andtherefore not differentiated in figures. If there were additional variables in the model, they were set to mean values for calculating regression lines.

62 J. Steckel et al. / Biological Conservation 172 (2014) 56–64

abundance of antagonists were mainly driven by species richnessof hosts and abundance of host, respectively. Other patterns werefound in a former trap-nest study where the abundance of antago-nists was positively affected by the percentage of semi-naturalhabitats on a smaller scale (500 m) than host species richness(up to 750 m) (Steffan-Dewenter et al., 2002). Additionally, a this-tle-study (Cirsium arvense (L.)) revealed larger home ranges for her-bivores (>3 000 m) than for their parasitoids (750 m) (Kruess,2003). The differing responses of hosts and antagonists may haveconsequences for pollination services by bees and control of pestorganisms by predatory wasps and generalist antagonists: polli-nating bees may offer their service within several hundred meters(Ricketts et al., 2008) if a landscape provides a diversity of habitats

while their antagonists can only thrive if habitat connecting corri-dors over large spatial scales are available.

The general relation of biodiversity and ecosystem functioningis still controversially discussed. Depending on four mechanisms(redundancy, idiosyncrasy, species complementarity and samplingeffects) there might be a correlation of species richness and ecosys-tem functioning or not (Tscharntke et al., 2005). Our study indi-cates that local species richness of antagonists is positivelycorrelated with ecosystem functions like parasitism. This is inaccordance with former trap-nest studies (Tscharntke et al.,1998; Veddeler et al., 2010).

It might depend on the region studied whether a landscapecharacteristic affects species richness or not (Fahrig et al., 2011;

J. Steckel et al. / Biological Conservation 172 (2014) 56–64 63

Gimona et al., 2009). The mean species richness per region de-clined from south to north. This might be explained by differingbiotic (below- and aboveground species pool) and abiotic (climate)or socio-economic (agricultural practices) conditions across re-gions (Fahrig et al., 2011; Gaston, 2000; Tscharntke et al., 2012).Nevertheless, for most dependent variables we could find consis-tent, general patterns independently of region, because there wereno significant interactions with region in the final models. Thisindicates that our results regarding landscape composition andconfiguration are of general significance across different regions.Future studies could benefit from an even higher number of repli-cates at the region level and the selection of regions with morecontrasting landscape characteristics.

To conclude, we could show with a highly replicated large-scalefield study that hosts and their antagonists respond to differentcomponents of landscape heterogeneity at different spatial scales.To facilitate host species (pollinators and predators), small-scalemeasures that involve a diversification of the surrounding matrixare needed. The requirements of their antagonists could be betterpromoted by management schemes that address landscape config-uration, e.g. by creating edge habitats and connecting corridors atlarger spatial scales. Hence, to facilitate trophic interactions andvaluable ecosystem services offered by solitary bees, wasps andtheir antagonists, management measures should specificallyaddress landscape diversity and spatial complexity by creating amultitude of different, complex-shaped habitats on a landscapescale.

Acknowledgements

We thank the managers of the three exploratories, Swen Ren-ner, Sonja Gockel, Andreas Hemp and Martin Gorke and SimonePfeiffer for their work in maintaining the plot and project infra-structure, and Markus Fischer, the late Elisabeth Kalko, Eduard Lin-senmair, Dominik Hessenmoeller, Jens Nieschulze, Daniel Prati,Ingo Schoening, François Buscot, Ernst-Detlef Schulze and Wolf-gang W. Weisser for their role in setting up the Biodiversity Explor-atories project. We thank Rainer Theunert, Józan Zsolt, KlausHorstmann and Christoph Saure for taxonomic support and forpre-identification and preparation Eva Stangler, Sebastian Hop-fenmueller, Marco Eckl, Katharina Kallnik. We thank Anna Hövel-born and Franziska Engelen for digitalization of the landscapemapping. The work has been funded by the DFG Priority Program1374 ‘Infrastructure-Biodiversity-Exploratories’ (STE 957-7) andby the program ‘Chancengleichheit für Frauen in Forschung undLehre’ supplied by the Frauenbuero of the University ofWuerzburg. Field work permits were issued by the responsiblestate environmental offices of Baden-Wuerttemberg, Thueringen,and Brandenburg (according to § 72 BbgNatSchG).

Appendix A. Supplementary material

Supplementary data associated with this article can be found, inthe online version, at http://dx.doi.org/10.1016/j.biocon.2014.02.015. These data include Google maps of the most important areasdescribed in this article.

References

Bivand, R., 2012. spdep: Spatial dependence: weighting schemes, statistics andmodels [WWW Document]. <http://cran.r-project.org/>.

Blüthgen, N., Dormann, C.F., Prati, D., Klaus, Valentin.H., Kleinebecker, T., Hoelzel, N.,Alt, F., Boch, S., Gockel, S., Hemp, A., Mueller, J., Nieschulze, J., Renner, S.C.,Schoening, I., Schumacher, U., Socher, S.A., Wells, K., Birkhofer, K., Buscot, F.,Oelmann, Y., Rothenwoehrer, C., Scherber, C., Tscharntke, T., Weiner, C.N.,Fischer, M., Kalko, E.K.V., Linsenmair, K.E., Schulze, E.-D., Weisser, W.W., 2012. Aquantitative index of land-use intensity in grasslands: integrating mowing,grazing and fertilization. Basic Appl. Ecol. 13, 207–220.

Borer, E.T., Seabloom, E.W., Tilman, D., 2012. Plant diversity controls arthropodbiomass and temporal stability. Ecol. Lett. 15, 1457–1464.

Brueckmann, S.V., Krauss, J., van Achterberg, C., Steffan-Dewenter, I., 2011. Theimpact of habitat fragmentation on trophic interactions of the monophagousbutterfly Polyommatus coridon. J. Insect Conserv. 15, 707–714.

Burnham, K., Anderson, D., 2004. Multimodel inference – understanding AIC and BICin model selection. Sociol. Methods. Res. 33, 261–304.

Chaplin-Kramer, R., O’Rourke, M.E., Blitzer, E.J., Kremen, C., 2011. A meta-analysis ofcrop pest and natural enemy response to landscape complexity. Ecol. Lett. 14,922–932.

Crawley, M.J., 2007. The R Book. John Wiley & Sons Ltd., The Atrium, Southern Gate,Chichester, West Sussex PO19 8SQ, England.

Development Core Team, R., 2012. R: A language and environment for statisticalcomputing. R Foundation for Statistical Computing, Vienna, Austria, ISBN 3-900051-07-0 [WWW Document]. <http://www.R-project.org/>.

Durrer, S., Schmid-Hempel, P., 1995. Parasites and the regional distribution ofbumblebee species. Ecography 18, 114–122.

Ebeling, A., Klein, A.-M., Weisser, W.W., Tscharntke, T., 2012. Multitrophic effects ofexperimental changes in plant diversity on cavity-nesting bees, wasps, andtheir parasitoids. Oecologia 169, 453–465.

Fahrig, L., Baudry, J., Brotons, L., Burel, F.G., Crist, T.O., Fuller, R.J., Sirami, C.,Siriwardena, G.M., Martin, J.-L., 2011. Functional landscape heterogeneity andanimal biodiversity in agricultural landscapes. Ecol. Lett. 14, 101–112.

Fischer, M., Bossdorf, O., Gockel, S., Haensel, F., Hemp, A., Hessenmoeller, D., Korte,G., Nieschulze, J., Pfeiffer, S., Prati, D., Renner, S., Schoening, I., Schumacher, U.,Wells, K., Buscot, F., Kalko, E.K.V., Linsenmair, K.E., Schulze, E.-D., Weisser, W.W.,2010. Implementing large-scale and long-term functional biodiversity research:the biodiversity exploratories. Basic Appl. Ecol. 11, 473–485.

Forman, R.T.T., 1995. Some general principles of landscape and regional ecology.Landscape Ecol. 10, 133–142.

Gaston, K.J., 2000. Global patterns in biodiversity. Nature 405, 220–227.Gathmann, A., Tscharntke, T., 1999. Landschafts-bewertung mit bienen und wespen

in nisthilfen: artenspektrum, interaktionen und bestimmungsschlüssel.Naturschutz Landschaftspflege Baden-Württemberg 73, 277–305.

Gathmann, A., Tscharntke, T., 2002. Foraging ranges of solitary bees. J. Anim. Ecol.71, 757–764.

Gathmann, A., Greiler, H.-J., Tscharntke, T., 1994. Trap-nesting bees and waspscolonizing set-aside fields: succession and body size, management by cuttingand sowing. Oecologia 98, 8–14.

Gimona, A., Messager, P., Occhi, M., 2009. CORINE-based landscape indices weaklycorrelate with plant species richness in a northern European landscape transect.Landscape Ecol. 24, 53–64.

Greenleaf, S.S., Williams, N.M., Winfree, R., Kremen, C., 2007. Bee foraging rangesand their relationship to body size. Oecologia 153, 589–596.

Hendrickx, F., Maelfait, J.-P., Van Wingerden, W., Schweiger, O., Speelmans, M.,Aviron, S., Augenstein, I., Billeter, R., Bailey, D., Bukacek, R., Burel, F., Diekoetter,T., Dirksen, J., Herzog, F., Liira, J., Roubalova, M., Vandomme, V., Bugter, R., 2007.How landscape structure, land-use intensity and habitat diversity affectcomponents of total arthropod diversity in agricultural landscapes. J. Appl.Ecol. 44, 340–351.

Holland, J.D., Bert, D.G., Fahrig, L., 2004. Determining the spatial scale of species’response to habitat. Bioscience 54, 227–233.

Holt, R.D., 1996. Food Webs in Space: An Island Biogeographic Perspective. FoodWebs. Chapman & Hall, New York, USA.

Holt, R.D., Lawton, J.H., Polis, G.A., Martinez, N.D., 1999. Trophic rank and thespecies–area relationship. Ecology 80, 1495–1504.

Holzschuh, A., Steffan-Dewenter, I., Kleijn, D., Tscharntke, T., 2007. Diversity offlower-visiting bees in cereal fields: effects of farming system, landscapecomposition and regional context. J. Appl. Ecol. 44, 41–49.

Holzschuh, A., Steffan-Dewenter, I., Tscharntke, T., 2010. How do landscapecomposition and configuration, organic farming and fallow strips affect thediversity of bees, wasps and their parasitoids? J. Anim. Ecol. 79, 491–500.

Karp, D.S., Rominger, A.J., Zook, J., Ranganathan, J., Ehrlich, P.R., Daily, G.C., 2012.Intensive agriculture erodes b-diversity at large scales. Ecol. Lett. 15, 963–970.

Kennedy, C.M., Lonsdorf, E., Neel, M.C., Williams, N.M., Ricketts, T.H., Winfree, R.,Bommarco, R., Brittain, C., Burley, A.L., Cariveau, D., Carvalheiro, L.G., Chacoff,N.P., Cunningham, S.A., Danforth, B.N., Dudenhöffer, J.-H., Elle, E., Gaines, H.R.,Garibaldi, L.A., Gratton, C., Holzschuh, A., Isaacs, R., Javorek, S.K., Jha, S., Klein,A.M., Krewenka, K., Mandelik, Y., Mayfield, M.M., Morandin, L., Neame, L.A.,Otieno, M., Park, M., Potts, S.G., Rundlöf, M., Saez, A., Steffan-Dewenter, I., Taki,H., Viana, B.F., Westphal, C., Wilson, J.K., Greenleaf, S.S., Kremen, C., 2013. Aglobal quantitative synthesis of local and landscape effects on wild beepollinators in agroecosystems. Ecol. Lett. 16, 584–599.

Kissling, W.D., Carl, G., 2008. Spatial autocorrelation and the selection ofsimultaneous autoregressive models. Glob. Ecol. Biogeogr. 17, 59–71.

Klein, A.-M., Steffan-Dewenter, I., Tscharntke, T., 2004. Foraging trip duration anddensity of megachilid bees, eumenid wasps and pompilid wasps in tropicalagroforestry systems. J. Anim. Ecol. 73, 517–525.

Kruess, A., 2003. Effects of landscape structure and habitat type on a plant-herbivore-parasitoid community. Ecography 26, 283–290.

Li, H., Reynolds, J.F., 1995. On definition and quantification of heterogeneity. Oikos73, 280–284.

McGarigal, K., Cushman, S.A., Neel, M.C., Ene, E., 2002. FRAGSTATS: Spatial PatternAnalysis Program for Categorical Maps. Computer software program producedby the authors at the University of Massachusetts, Amherst.

64 J. Steckel et al. / Biological Conservation 172 (2014) 56–64

Osborne, J.L., Williams, I.H., Corbet, S.A., 1991. Bees, pollination and habitat changein the European Community. Bee World 72, 99–116.

Poschlod, P., WallisDeVries, M.F., 2002. The historical and socioeconomicperspective of calcareous grasslands—lessons from the distant and recentpast. Biol. Conserv. 104, 361–376.

Rand, T.A., Tylianakis, J.M., Tscharntke, T., 2006. Spillover edge effects: the dispersalof agriculturally subsidized insect natural enemies into adjacent naturalhabitats. Ecol. Lett. 9, 603–614.

Rand, T.A., van Veen, F.J.F., Tscharntke, T., 2012. Landscape complexity differentiallybenefits generalized fourth, over specialized third, trophic level naturalenemies. Ecography 35, 97–104.

Ricketts, T.H., Regetz, J., Steffan-Dewenter, I., Cunningham, S.A., Kremen, C.,Bogdanski, A., Gemmill-Herren, B., Greenleaf, S.S., Klein, A.M., Mayfield, M.M.,Morandin, L.A., Ochieng’, A., Viana, B.F., 2008. Landscape effects on croppollination services: are there general patterns? Ecol. Lett. 11, 499–515.

Scherber, C., Lavandero, B., Meyer, K.M., Perovic, D., Visser, U., Wiegand, K.,Tscharntke, T., 2012. Scale effects in biodiversity and biocontrol: methods andstatistical analysis. In: Gurr, G.M., Wratten, S.D., Snyder, W.E., Read, D.M.Y.(Eds.), Biodiversity and Insect Pests: Key Issues for Sustainable Management.John Wiley & Sons.

Shannon, C.E., Weaver, W., 1949. Mathematical Theory of Communication.University of Illinois Press.

Socher, S.A., Prati, D., Boch, S., Mueller, J., Klaus, V.H., Hoelzel, N., Fischer, M., 2012.Direct and productivity-mediated indirect effects of fertilization, mowing andgrazing on grassland species richness. J. Ecol. 100, 1391–1399.

Sokal, R.R., Rohlf, F.J., 1994. Biometry: The Principles and Practices of Statistics inBiological Research, third ed. Palgrave Macmillan.

Steffan-Dewenter, I., 2002. Landscape context affects trap-nesting bees, wasps, andtheir natural enemies. Ecol. Entomol. 27, 631–637.

Steffan-Dewenter, I., 2003. Importance of habitat area and landscape context forspecies richness of bees and wasps in fragmented orchard meadows. Conserv.Biol. 17, 1036–1044.

Steffan-Dewenter, I., Schiele, S., 2008. Do resources or natural enemies drive beepopulation dynamics in fragmented habitats? Ecology 89, 1375–1387.

Steffan-Dewenter, I., Münzenberg, U., Bürger, C., Thies, C., Tscharntke, T., 2002.Scale-dependent effects of landscape context on three pollinator guilds. Ecology83, 1421–1432.

Swift, M.J., Izac, A.-M.N., van Noordwijk, M., 2004. Biodiversity and ecosystemservices in agricultural landscapes—are we asking the right questions? Agr.Ecosyst. Environ. 104, 113–134.

Thies, C., Steffan-Dewenter, I., Tscharntke, T., 2003. Effects of landscape context onherbivory and parasitism at different spatial scales. Oikos 101, 18–25.

Thies, C., Roschewitz, I., Tscharntke, T., 2005. The landscape context of cereal aphid–parasitoid interactions. Proc. R. Soc. B 272, 203–210.

Tscharntke, T., Kruess, A., 1999. Habitat fragmentation and biological control. In:Hawkins, B.A., Cornell, H.V. (Eds.), Theoretical Approaches to Biological Control.Cambridge University Press, Cambridge, pp. 190–205.

Tscharntke, T., Gathmann, A., Steffan-Dewenter, I., 1998. Bioindication using trap-nesting bees and wasps and their natural enemies: community structure andinteractions. J. Appl. Ecol. 35, 708–719.

Tscharntke, T., Klein, A.M., Kruess, A., Steffan-Dewenter, I., Thies, C., 2005.Landscape perspectives on agricultural intensification and biodiversity –ecosystem service management. Ecol. Lett. 8, 857–874.

Tscharntke, T., Tylianakis, J.M., Rand, T.A., Didham, R.K., Fahrig, L., Batáry, P.,Bengtsson, J., Clough, Y., Crist, T.O., Dormann, C.F., Ewers, R.M., Fruend, J., Holt,R.D., Holzschuh, A., Klein, A.M., Kleijn, D., Kremen, C., Landis, D.A., Laurance, W.,Lindenmayer, D., Scherber, C., Sodhi, N., Steffan-Dewenter, I., Thies, C., van derPutten, W.H., Westphal, C., 2012. Landscape moderation of biodiversity patternsand processes – eight hypotheses. Biol. Rev. 87, 661–685.

Van Swaay, C.A., 2002. The importance of calcareous grasslands for butterflies inEurope. Biol. Conserv. 104, 315–318.

Veddeler, D., Tylianakis, J., Tscharntke, T., Klein, A.-M., 2010. Natural enemydiversity reduces temporal variability in wasp but not bee parasitism. Oecologia162, 755–762.

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

Westphal, C., Steffan-Dewenter, I., Tscharntke, T., 2006. Bumblebees experiencelandscapes at different spatial scales: possible implications for coexistence.Oecologia 149, 289–300.

Westphal, C., Bommarco, R., Carre, G., Lamborn, E., Morison, N., Petanidou, T., Potts,S.G., Roberts, S.P.M., Szentgyoergyi, H., Tscheulin, T., Vaissiere, B.E.,Woyciechowski, M., Biesmeijer, J.C., Kunin, W.E., Settele, J., Steffan-Dewenter,I., 2008. Measuring bee diversity in different european habitats andbiogeographical regions. Ecol. Monogr. 78, 653–671.

Westrich, P., 1989. Die Wildbienen Baden-Württembergs. E. Ulmer, Stuttgart.Westrich, P., 1996. The conservation of bees. in: Habitat Requirements of Central

European Bees and the Problems of Partial Habitats, Linnean SocietySymposium Series. Academic Press, London, pp. 1–16.

Zurbuchen, A., Landert, L., Klaiber, J., Müller, A., Hein, S., Dorn, S., 2010. Maximumforaging ranges in solitary bees: only few individuals have the capability tocover long foraging distances. Biol. Conserv. 143, 669–676.

Zuur, A.F., Ieno, E.N., Walker, N., Saveliev, A.A., Smith, G.M., 2009. Mixed EffectsModels and Extensions in Ecology with R. Springer.

Zuur, A.F., Ieno, E.N., Elphick, C.S., 2010. A protocol for data exploration to avoidcommon statistical problems. Methods Ecol. Evol. 1, 3–14.