Persistent negative effects of pesticides on biodiversity and biological control potential on European farmland.

ARTICLE IN PRESS

1439-1791/$ - se

doi:10.1016/j.ba

�CorrespondE-mail addr

Basic and Applied Ecology 11 (2010) 97–105 www.elsevier.de/baae

Persistent negative effects of pesticides on biodiversity and biological

control potential on European farmland

Flavia Geigera, Jan Bengtssonb, Frank Berendsea,�, Wolfgang W. Weisserc,Mark Emmersond,e, Manuel B. Moralesf, Piotr Ceryngierg, Jaan Liirah,Teja Tscharntkei, Camilla Winqvistb, Sonke Eggersb, Riccardo Bommarcob,Tomas Partb, Vincent Bretagnollej, Manuel Plantegenestk, Lars W. Clementc,Christopher Dennisd,e, Catherine Palmerd,e, Juan J. Onatef, Irene Guerrerof,Violetta Hawrog, Tsipe Aavikh, Carsten Thiesi, Andreas Flohrei, Sebastian Hankei,Christina Fischeri, Paul W. Goedhartl, Pablo Inchaustij

aNature Conservation and Plant Ecology Group, Wageningen University, Wageningen, The NetherlandsbDepartment of Ecology, Swedish University of Agricultural Sciences, Uppsala, SwedencInstitute of Ecology, Friedrich-Schiller-University Jena, Jena, GermanydDepartment of Zoology, Ecology and Plant Sciences, University College Cork, Cork, IrelandeEnvironmental Research Institute, University College Cork, Cork, IrelandfDepartment of Ecology, c/ Darwin, 2, Universidad Autonoma de Madrid, Madrid, SpaingCentre for Ecological Research, Polish Academy of Sciences, Lomianki, PolandhInstitute of Ecology and Earth Sciences, University of Tartu, Tartu, EstoniaiAgroecology, Department of Crop Science, Georg-August-University, Gottingen, GermanyjCentre for Biological Studies of Chize CNRS, Villiers-en-Bois, FrancekUMR BiO3P 1099 INRA/Agrocampus Ouest/Universite Rennes 1, Rennes, FrancelBiometris, Wageningen University and Research Centre, Wageningen, The Netherlands

Received 14 November 2009; accepted 4 December 2009

Abstract

During the last 50 years, agricultural intensification has caused many wild plant and animal species to go extinct regionallyor nationally and has profoundly changed the functioning of agro-ecosystems. Agricultural intensification has manycomponents, such as loss of landscape elements, enlarged farm and field sizes and larger inputs of fertilizer and pesticides.However, very little is known about the relative contribution of these variables to the large-scale negative effects onbiodiversity. In this study, we disentangled the impacts of various components of agricultural intensification on speciesdiversity of wild plants, carabids and ground-nesting farmland birds and on the biological control of aphids.

In a Europe-wide study in eight West and East European countries, we found important negative effects ofagricultural intensification on wild plant, carabid and bird species diversity and on the potential for biological pestcontrol, as estimated from the number of aphids taken by predators. Of the 13 components of intensification wemeasured, use of insecticides and fungicides had consistent negative effects on biodiversity. Insecticides also reducedthe biological control potential. Organic farming and other agri-environment schemes aiming to mitigate the negative

e front matter & 2009 Gesellschaft fur Okologie. Published by Elsevier Gmbh. All rights reserved.

ae.2009.12.001

ing author. Tel.: þ31 317 484973; fax: þ31 317 419000.

ess: [email protected] (F. Berendse).

ARTICLE IN PRESSF. Geiger et al. / Basic and Applied Ecology 11 (2010) 97–10598

effects of intensive farming on biodiversity did increase the diversity of wild plant and carabid species, but – contraryto our expectations – not the diversity of breeding birds.

We conclude that despite decades of European policy to ban harmful pesticides, the negative effects of pesticides onwild plant and animal species persist, at the same time reducing the opportunities for biological pest control. Ifbiodiversity is to be restored in Europe and opportunities are to be created for crop production utilizing biodiversity-based ecosystem services such as biological pest control, there must be a Europe-wide shift towards farming withminimal use of pesticides over large areas.& 2009 Gesellschaft fur Okologie. Published by Elsevier Gmbh. All rights reserved.

Zusammenfassung

Durch die Intensivierung der Landwirtschaft in den letzten 50 Jahren sind viele Pflanzen- und Tierarten aufregionaler und nationaler Ebene ausgestorben und ist die Funktion des Agrarokosystems beeintrachtigt. Dielandwirtschaftliche Intensivierung umfasst viele verschiedene Faktoren, wie zum Beispiel die Homogenisierung derLandschaft, die Vergroßerung von landwirtschaftlichen Betrieben und Ackern und den zunehmenden Gebrauch vonDungern und Pestiziden. Uber den relativen Beitrag der einzelnen Faktoren zu den weitgehenden Auswirkungen derIntensivierung auf die Biodiversitat ist jedoch wenig bekannt. In dieser Studie haben wir den Einfluss dieserverschiedenen Faktoren auf die Diversitat von Pflanzen, Laufkafern und bodenbrutenden Ackervogeln sowie auf diebiologische Schadlingsbekampfung von Blattlausen, entwirrt.In einer europaweiten Studie, in acht West- und Ost-Europaischen Landern, haben wir weitgehende, negative

Effekte der landwirtschaftlichen Intensivierung auf Pflanzen, Laufkafer, bodenbrutende Ackervogel und diebiologische Schadlingsbekampfung - die Anzahl durch naturliche Feinde gefressener Blattlause - gefunden. Von dendreizehn Faktoren der landwirtschaftlichen Intensivierung die wir gemessen haben, hatte der Gebrauch vonInsektiziden und Fungiziden konsequent negative Effekte auf die Biodiversitat. Insektizide reduzierten ebenfalls diebiologische Schadlingsbekampfung. Organische Bewirtschaftung und andere Formen von Okologischem Ausgleich, diezum Ziel haben, die negativen Effekte der Intensivierung auf Biodiversitat abzuschwachen, erhohten die Pflanzen- undLaufkaferdiversitat, jedoch – entgegen unseren Erwartungen - nicht die Diversitat der Brutvogel.

Wir stellen fest, dass trotz jahrzehntelanger europaischer Politik gegen schadliche Pestizide, die negativenAuswirkungen von Pestiziden auf Pflanzen- und Tierarten andauern und damit auch die Moglichkeit biologischerSchadlingsbekampfung abnimmt. Wenn die Biodiversitat in Europa erhalten werden soll und die Chance aufbiodiversitatsgebundenen Okosystemfunktionen, wie biologische Schadlingsbekampfung, beruhenden Ackerbaugeschaffen werden soll, ist eine europaweite Veranderung zu einer Bewirtschaftung mit minimalem Gebrauch vonPestiziden uber eine große Flache notwendig.& 2009 Gesellschaft fur Okologie. Published by Elsevier Gmbh. All rights reserved.

Keywords: Agricultural intensification; Organic farming; Agri-environment schemes; Vascular plants; Carabids; Birds

Introduction

Farmland is the most extensive habitat for wild plantand animal species in Europe, covering 43% of the EUmembers states’ surface area (EU-27) and still harbour-ing a large share of European biodiversity, e.g., 50% ofall European bird species (Pain & Pienkowski, 1997) and20–30% of the British and German flora (Marshallet al., 2003). In recent decades, however, agriculturalintensification has unquestionably contributed to theimpoverishment of European farmland biodiversity(Donald, Green, & Heath, 2001; Krebs, Wilson, Brad-bury, & Siriwardena, 1999; Robinson & Sutherland,2002; Stoate et al., 2001). There is considerable concernthat declines in biodiversity affect the delivery ofecosystem services (Hooper et al., 2005). In agriculturallandscapes, the services considered most at risk from

agricultural intensification are biological pest control(Tscharntke, Klein, Kruess, Steffan-Dewenter, & Thies,2005), crop pollination (Biesmeijer et al., 2006) andprotection of soil fertility (Brussaard et al., 1997).

Agricultural intensification takes place at various spatialscales, from increased application of herbicides, insecticides,fungicides and chemical fertilizer on local fields to loss ofnatural and semi-natural habitats and decreased habitatheterogeneity at the farm and landscape levels (Attwood,Maron, House, & Zammit, 2008; Benton, Vickery, &Wilson, 2003; Billeter et al., 2008; Hendrickx et al., 2007;Tscharntke et al., 2005; Weibull, Bengtsson, & Nohlgren,2000). So far it has been difficult to disentangle the impactsof intensified management of local fields from changes inland use at the landscape level, since both occursimultaneously in most agricultural landscapes (Robinson& Sutherland, 2002).

ARTICLE IN PRESSF. Geiger et al. / Basic and Applied Ecology 11 (2010) 97–105 99

In addition, previous assessments have generallyfocused on a few taxa or countries and hardly anystudy has simultaneously addressed the effects ofagricultural intensification on key ecosystem servicessuch as the biological control of agricultural pests.

Since the early 1990s the EU has promoted initiatives toprevent and reduce the negative effects of intensive farming.In 1991, legislation limiting the use of pesticides with highrisks to the environment came into force (Council Directive91/414/EEC of 15 July 1991). The reform of the CommonAgricultural Policy (CAP) in 1992 aimed to reduce thenegative consequences of agricultural intensification byfinancially supporting agri-environment schemes andorganic farming (Council Regulation 2078/92/EEC of 30June 1992). However, several studies have shown that agri-environment schemes and organic farming do not alwaysdeliver the expected benefits (Bengtsson, Ahnstrom, &Weibull, 2005; Berendse, Chamberlain, Kleijn, & Schekker-man, 2004; Kleijn, Berendse, Smit, & Gilissen, 2001). So, animportant, but yet unanswered question is whether policieshave significantly reduced the adverse effects of intensive

Fig. 1. Effects of cereal yield (ton/ha) on: (A) the number of wild plan

of carabid species per sampling point (per trap during 2 sampling pe

(one survey plot of 500� 500m2), and (D) the median survival ti

including the two surrounding landscape variables as covariates and

farming on biodiversity and, closely linked to this, on thedelivery of key ecosystem services such as biological pestcontrol. In this study, we investigated in nine Europeanareas the effects of agricultural intensification and itscomponents on the species diversity of wild plants, carabidsand ground-nesting farmland birds (thus considering threedifferent trophic levels) and biological control potential. Wemeasured eight landscape structure variables and 13components of agricultural intensification at farm and fieldlevel and disentangled their different effects on biodiversityloss. Moreover, we tested the hypothesis that both organicfarming and agri-environment schemes reduce the negativeeffects of intensive farming on biodiversity.

Material and methods

Study area

The nine areas studied were located in eight countries:Sweden, Estonia, Poland, the Netherlands, Germany

t species per sampling point (in 3 plots of 4m2), (B) the number

riods), (C) the number of ground-nesting bird species per farm

me of aphids (h). Trend lines were calculated using GLMM

field, farm and study area as nested random effects.

ARTICLE IN PRESSF. Geiger et al. / Basic and Applied Ecology 11 (2010) 97–105100

(two areas: close to Gottingen, West-Germany and Jena,East-Germany, respectively), France, Spain and Ireland(see Appendix A, Fig. 1 for the locations of the ninestudy areas). Each area was between 30� 30 and50� 50 km2 in size to minimize differences in theregional species pools among farms within each area.In each area, 30 arable farms were selected along anintensification gradient using cereal yield as a proxy foragricultural intensification. The farms were selected sothat the previous year’s yield and landscape compositionwere uncorrelated within the study area in question.

Sampling protocol

On each farm, five points distributed over no morethan five arable fields were selected for sampling wildplants and carabids and estimating the biologicalcontrol potential. Most (80%) of the sampling pointswere in fields of winter wheat (the major cereal crop inmuch of Europe). The remainder was in winter barley(9%), spring wheat (5%), winter rye (5%) or triticale(o1%). To avoid field margin effects on observations,the sampling points were positioned 10m from thecentre of one side of the field. Whenever two samplingpoints were located in the same field, they were placed atopposite sides of the field. On each farm, one area of500� 500m2 was selected around one of the sampledfields, for the survey of breeding birds. All sampled fieldsfrom different farms were at least 1 km apart. Samplingwas performed during spring and summer 2007 and wassynchronized using the phenological stages of winterwheat in each study area as a time reference.

Wild plants

At each sampling point, vegetation relevees weremade once during the flowering to the milk-ripeningstage of winter wheat, using three plots of 2� 2m2. Theplots were placed 5m apart on a line parallel to the fieldborders. All species with at least the first two leaves(after the cotyledon) were recorded per plot. To avoidphenological effects of sampling, the sequence of farmsurveys was randomized over the intensification gradientwithin each study area.

Carabids

Carabids were caught with two pitfall traps persampling point, which were opened during two periodsof 7 days. The first sampling period occurred 1 weekafter the appearance of spikes of winter wheat(immediately after the biological control experiment,see below) and the second sampling period coincidedwith the milk-ripening stage of winter wheat. The twopitfall traps (90mm diameter, filled with 50% ethyleneglycol) were placed in the middle of the two outervegetation plots. The invertebrates caught were fixed in

the lab with 70% ethanol. We identified all the speciescaught in one trap randomly selected from each pair oftraps.

Birds

The bird surveys were conducted according to amodified version of the British Trust for Ornithology’sCommon Bird Census (Bibby, Burgess, & Hill, 1992),starting according to local information on thephenology of breeding birds. They were conductedthree times, at intervals of 3 weeks. Bird inventoryquadrats of 500� 500m2 were surveyed in such away that each spot within the quadrat was no morethan 100m from the surveyor’s route. The surveys tookplace between 1 h after dawn and noon, but only if itwas not windy, cloudy, or raining. Breeding birdterritories for ground-nesting farmland species weredetermined using the three survey rounds (Appendix A,Table 1). Three different criteria were used to definebreeding bird territories, depending on the species’detectability and breeding behaviour (Appendix A,Table 1).

Biological control potential

Biological control potential was estimated experimen-tally during the emergence of the first inflorescence ofwinter wheat (Ostman, 2004). The experiment lasted 2days and was repeated once within 8 days. In themorning of the first day, live pea aphids (Acyrthosiphon

pisum) of the third or fourth instar were glued to plasticlabels (three per label) by at least two of their legs andpart of their abdomen. Odourless superglue was used.At noon, the labels were placed in the three vegetationplots at three of the five sampling points per farm. Thelabels were bent over slightly, so the aphids were on thelower surface, protected from rain. Three labels wereplaced along the diagonal of each plot. Hence, at eachfarm there were 27 labels, with 81 aphids in total.Immediately after the aphids had been placed in thefield, the numbers of aphids present were recorded.Thereafter, the labels were checked four more timesduring 30 h: around 6 p.m. of the first day, at 8 a.m., 1p.m. and about 6 p.m. on the following day, the exacttime varying depending on the study area. After the lastcheck, the labels with the remaining aphids were takento the lab and checked under stereo microscopes tocheck whether remaining aphids could not have beenremoved by predators because they were covered withglue. The data used for the analyses was from one orboth of the rounds, depending on what was availablefrom each study area.

ARTICLE IN PRESSF. Geiger et al. / Basic and Applied Ecology 11 (2010) 97–105 101

Farmers’ questionnaire

Information about yields and farming practices(pesticide and fertilizer use, ploughing and mechanicalweed control regime) and farm layout (number of crops,percentage land covered by an agri-environmentscheme, field size) was collected by means of aquestionnaire sent out to all participating farmers. Theresponse was 98%.

Landscape structure

On the landscape scale, eight landscape structurevariables were estimated within circles around eachsampling point (with radii of 500 and 1000m) andadditionally four variables around each bird quadrat(used for the analysis of the bird data; only 500mradius): mean field size and its standard deviation, thepercentage of land planted with arable crops within thearea and the Shannon habitat diversity index. Thefollowing habitat classes were used to estimate thehabitat diversity (according to the definitions from theEuropean Topic Centre on Land Use and SpatialInformation (Buttner, Feranec, & Jaffrain, 2000):continuous urban fabrics, discontinuous urban fabrics,cultivated arable lands, fallow lands under rotationsystems, permanent crops, pastures, forests, transitionalwoodland-scrub and water.

Statistics

Generalized linear mixed models (GLMM) (Breslow& Clayton, 1993) were used to analyse the effects ofagricultural intensification on biodiversity and thebiological control potential in GenStat 11.1 (Payneet al., 2008). All explanatory variables were included asfixed effects. Because of the sampling structure of thedata, fields nested within farms, farms within study area,and study area, were included as random effects. Toidentify the distribution of the species diversity data(Appendix A, Table 5), the variance to mean relation-ship was explored at the lowest stratum. This revealedthat a Poisson distribution was appropriate for thenumbers of plant, carabid and bird species. The mediansurvival time of aphids was heavily skewed and wastherefore assumed to follow a lognormal distribution.

Heavily skewed explanatory variables were log-transformed and the percentage of land planted witharable crops was logit transformed (Appendix A,Table 4). All variables were standardized according to(x�m)/s, with x=measurement, m=mean and s=stan-dard deviation to enable comparison of the magnitudeof their effects.

To reduce the number of landscape variables, aprincipal component analysis (PCA) was done with

Canoco for Windows 4.5 (Ter Braak & Smilauer, 2002).This revealed two distinct groups of variables (seeAppendix A, Fig. 2): the first group was mean field sizeand its standard deviation, the second was the percen-tage of land planted with arable crops and the Shannonhabitat diversity index. Because it had the highestcorrelation with the plane defined by the two axes(longest PCA arrow), mean field size within a radius of500m was selected from the first group to be included inthe analyses. In the second group, the highest correla-tion with the plane was the percentage land planted witharable crops. We chose the variable with the same radiusas mean field size (selected from the first group), i.e.500m.

Only a few intensification variables showed highcorrelations between each other (Pearson correlationcoefficient 40.7; see Appendix A, Tables 6A and B).The number of insecticide applications correlated withthe amount of fungicides applied (r=0.75) and thenumber of fungicide applications (r=0.73) at the farmlevel only. Number of fungicide applications and theamount of fungicides applied were correlated both at thefarm level and at the sampling point level (r=0.87 and0.83, respectively), as were the amount of inorganicfertilizer applied and the amount of fungicides used (forboth, r=0.72). These correlations cannot be consideredproblematic under the present modelling approach(Brotons, Thuiller, Araujo, & Hirzel, 2004).

The mean values and standard deviations of allresponse and explanatory variables included in theanalyses are given in Appendix A, Table 7.

Three separate analyses were done using different setsof explanatory variables, but always including the twolandscape variables (mean field size and percentage landplanted with arable crops within 500m) as covariates.The first analysis included yield (as a summary variablefor agricultural intensification, see Tilman, Cassman,Matson, Naylor, & Polasky (2002)) and its interactionwith study area. The second analysis included 13components of agricultural intensification related tofarming practices: amounts of chemical N fertilizer,amounts of organic fertilizer, number of applications ofherbicides, insecticides and fungicides, applied amountsof the active ingredients of herbicides, insecticides andfungicides, number of crops per farm, field size,frequency of ploughing, frequency of mechanical weedcontrol and the percentage of arable land under agri-environment schemes (Appendix A, Table 4). Modelswere derived using forward and backward selection. Theforward selection started with an empty model (exceptfor the two landscape variables) and at each step thevariable with the most significant effect was included, onthe basis of the results of Wald tests (po0.05). Thisprocedure was reiterated until variables no longer addedsignificant effects to the model. The backward selectionstarted with a full model and at each step the most

ARTICLE IN PRESSF. Geiger et al. / Basic and Applied Ecology 11 (2010) 97–105102

non-significant variable, i.e. the variable with the highestp-value, was removed. Forward, as well as backwardselection, resulted in identical models in all cases. In athird analysis, we investigated the single effects of farmtype (conventional or organic) and the percentage ofland under agri-environment schemes. Organic farmersdo not apply chemical fertilizers and use only a very

Fig. 2. Effects of cereal yield (ton/ha) on the number of wild plant sp

areas. Trend lines were calculated using GLMM including the two su

study area as nested random effects and are plotted, whenever the r

limited set of pesticides. Agri-environment schemesinclude commitments to lower fertilizer and pesticideapplications, while in some countries, margins or entirefields are excluded from fertilizer or pesticide applica-tions.

We emphasize that in the text the term ‘effect’ is usedfor statistical associations and relationships, and does

ecies per sampling point (in 3 plots of 4m2) in each of the study

rrounding landscape variables as covariates and field, farm and

elationship was significant (po0.05).

ARTICLE IN PRESS

Table 1. Effects of different components of agricultural intensification on the diversity of plants, carabids and birds and median

aphid survival time. The models selected after considering all 13 intensification variables using forward selection (backward selection

produced identical models) are presented. All models included the two landscape variables (mean field size and percentage of land

planted with arable crops within a radius of 500m, in italics), even if these had no significant effects (non-significant effects are not

shown). Intensification variables were only included if they had significant effects using the Wald test (po0.05). AES=agri-

environment scheme, amount of a.i.=amount of active ingredients.

Response variable Explanatory variable Standardized effect v21 p-value

Number of plant species Mean field size �0.094 6.09 0.014

% of land under AES 0.149 12.23 o0.001

Frequency of herbicide application �0.1061 8.88 0.003

Frequency of insecticide application �0.105 6.15 0.013

Applied amounts of a.i. of fungicides �0.262 31.45 o0.001

Number of carabid species % of land under AES 0.062 6.31 0.012

Applied amounts of a.i. of insecticides �0.061 10.87 0.001

Number of breeding bird species Frequency of fungicide application �0.127 5.71 0.017

Median survival time of aphids % of land under AES �0.144 9.43 0.002

Applied amounts of a.i. of insecticides 0.114 11.17 0.001

F. Geiger et al. / Basic and Applied Ecology 11 (2010) 97–105 103

not necessarily mean a causal relation between twovariables.

Data availability

For the second analysis (13 variables of farmingpractices as explanatory variables), sampling points withone or more missing variables were removed. Bird datawere not collected in France. There were no data onaphid survival time for Spain and France.

Results

We first investigated the relationship between cerealyield, a variable closely related to many differentintensification measures (Donald et al., 2001; Tilmanet al., 2002), and wild plant, carabid and breeding birdspecies diversity (Appendix A, Table 1) on arable fields.We found strong negative relationships between cerealyields and the species diversity of wild plants, carabidsand ground-nesting farmland birds (Wald tests:w21=141.42, po0.001; w21=23.33, po0.001; w21=7.33,p=0.007; Appendix A, Table 2). On average, in thesampled area, an increase in cereal yield from 4 to 8 ton/ha results in the loss of five of the nine plant species, twoof the seven carabid species and one of the three birdspecies (Fig. 1A–C).

Crop yield correlated positively with median aphidsurvival time (w21=6.85; p=0.009), suggesting a negativeeffect on the biological control potential (Fig. 1D).

The effects of wheat yield on wild plant and carabidspecies diversity and aphid survival time differed amongstudy areas (yield� study area interaction: w28=36.87,po0.001; w28=24.35, p=0.002; w26=17.84, p=0.007,respectively). Comparison of the yield effects amongstudy areas revealed that in some countries, yield hadnegative effects on these variables, but in other countriesthere was no relationship (Fig. 2; Appendix A, Table 3).In two of the three study areas where we found norelationship, the variation in yield among fields andfarms was much smaller than in the other countries,which probably explains the lack of significant effects.There were no consistent differences between West andEast European countries.

As a second step, we investigated the relativeimportance of 13 variables we considered as relevantcomponents of agricultural intensification (Appendix A,Table 4). The characteristics of the surrounding land-scape had significant effects on wild plant speciesdiversity only (Table 1). The number of plant specieswas inversely related to average size of fields within aradius of 500m, emphasizing the importance of fieldmargins for the establishment of wild plant species onarable land. The number of wild plant species declinedas the frequency of herbicide and insecticide applicationand the amounts of active ingredients of fungicidesincreased (Table 1). The number of carabid species wasnegatively affected by the amounts of active ingredientsof insecticide applied. Bird species diversity declinedwith increasing frequency of fungicide application, avariable closely correlated with the frequency ofinsecticide application (Pearson’s correlation coefficientr=0.732; po0.001). The predation on aphids declined

ARTICLE IN PRESSF. Geiger et al. / Basic and Applied Ecology 11 (2010) 97–105104

as the applied amounts of insecticides increased(Table 1).

Thirdly, we examined the effects of organic farmingand the implementation of agri-environment schemes onbiodiversity and the biological control potential. Or-ganic farms comprised 22% of the total number ofselected farms in our study and occurred in five of thenine study areas. These farms harboured more wildplant and carabid species (single effects of farm type:w21=164.96, po0.001; w21=3.98, p=0.046, respectively;Appendix A, Table 2), but not significantly more birdspecies (w21=1.31, p=0.252). Aphid survival was notsignificantly lower on organic farms as compared withconventional farms (w21=2.93, p=0.087). 45% of theselected farms had agri-environment schemes. Theseschemes had positive effects on the number of wild plantand carabid species and the predation of aphids (singleeffects of percentage of land with agri-environmentscheme: w21=51.97, po0.001; w21=6.91, p=0.009;w21=13.24, po0.001, respectively; Appendix A, Table 2),but not on bird species diversity (w21=1.56, p=0.211).

Discussion

We studied the effects of agricultural intensificationon biodiversity across three trophic levels and thepotential for biological pest control in eight Europeancountries. Out of the 13 studied components ofagricultural intensification, use of pesticides, especiallyinsecticides and fungicides, had the most consistentnegative effects on the species diversity of plants,carabids and ground-nesting farmland birds, and onthe potential for biological pest control. We concludethat despite several decades of implementing a Europe-wide policy intended to considerably reduce the amountof chemicals applied on arable land, pesticides are stillhaving disastrous consequences for wild plant andanimal species on European farmland. Importantly, thisimpact is also manifested as a reduction of the potentialof natural enemies to control pest organisms.

It is noteworthy that both organic farms, which applyonly those pesticides considered harmless to the environ-ment, and agri-environment schemes had positive effectson plant and carabid diversity, but did not show theexpected positive effects on bird species diversity. Apossible explanation for the lack of such positive effectsis the large spatial scale of the pollution associated withpesticide use across Europe, which inevitably leads tonegative effects of pesticides – even in areas where theapplication of these substances has been reduced orterminated. Such large-scale effects will be especiallyrelevant for taxa that utilize large areas, such as birds,mammals, butterflies (Rundlof, Bengtsson, & Smith, 2008)

and bees (Clough et al., 2007; Holzschuh, Steffan-Dewenter, & Tscharntke, 2008).

We conclude that if biodiversity is to be restored inEurope and opportunities are to be created for cropproduction utilizing biodiversity-based ecosystemservices such as biological pest control, a Europe-wideshift towards farming with minimal use of pesticidesover large areas is urgently needed.

Acknowledgements

We thank the European Science Foundation and theconnected national science foundations (Polish Ministryof Science and Higher Education, Spanish Ministry ofScience and Education, Estonian Scientific Foundation,Irish Research Council for Science, German FederalMinistry of Education and Science BMBF, GermanResearch Foundation, Netherlands Organization forScientific Research, Swedish Research Council, CentreNational de la Recherche Scientifique, DepartementEcologie et Developpement Durable) for funding thepresented study through the EuroDiversity AgriPopesprogramme.

Author contributions

F.G., C.W., L.W.C., C.D., C.P., I.G., M.B.M., P.C.,V.H., O.A., J.R., T.A., J.L., A.F., S.H., C.F., V.B. andM.P. collected the field data on plant, carabid and birddiversity and aphid survival; J.B., F.B., P.I., W.W.W.,M.E., M.B.M., P.C., P.K., J.L., T.T., S.E., R.B., T.P.,J.J.O. and C.T. designed the study and developed theprotocol for data collection; P.W.G., F.B., J.B., F.G.,P.I., W.W.W., M.E., R.B. and J.L. performed theanalysis of the data sets; F.B., F.G., J.B. and P.I. wrotethe manuscript. All authors discussed the results andcommented on the manuscript.

Appendix A. Supplementary material

Supplementary data associated with this article can befound in the online version at doi:10.1016/j.baae.2009.12.001.

References

Attwood, S. J., Maron, M., House, A. P. N., & Zammit, C.

(2008). Do arthropod assemblages display globally con-

sistent responses to intensified agricultural land use and

management? Global Ecology and Biogeography, 17,

585–599.

Bengtsson, J., Ahnstrom, J., & Weibull, A. C. (2005). The

effects of organic agriculture on biodiversity and abun-

ARTICLE IN PRESSF. Geiger et al. / Basic and Applied Ecology 11 (2010) 97–105 105

dance: A meta-analysis. Journal of Applied Ecology, 42,

261–269.

Benton, T. G., Vickery, J. A., & Wilson, J. D. (2003).

Farmland biodiversity: Is habitat heterogeneity the key?

Trends in Ecology & Evolution, 18, 182–188.

Berendse, F., Chamberlain, D., Kleijn, D., & Schekkerman, H.

(2004). Declining biodiversity in agricultural landscapes

and the effectiveness of agri-environment schemes. Ambio,

33, 499–502.

Bibby, C. J., Burgess, N. D., & Hill, D. A. (1992). Bird census

techniques. London: Academic Press Limited.

Biesmeijer, J. C., Roberts, S. P. M., Reemer, M., Ohlemuller,R., Edwards, M., Peeters, T. et al. (2006). Parallel declines

in pollinators and insect-pollinated plants in Britain and the

Netherlands. Science, 313, 351–354.

Billeter, R., Liira, J., Bailey, D., Bugter, R., Arens, P.,

Augenstein, I. et al. (2008). Indicators for biodiversity in

agricultural landscapes: A pan-European study. Journal of

Applied Ecology, 45, 141–150.

Breslow, N. E., & Clayton, D. G. (1993). Approximate

inference in generalized linear mixed models. Journal of

the American Statistical Association, 88, 9–25.

Brotons, L., Thuiller, W., Araujo, M. B., & Hirzel, A. H.

(2004). Presence–absence versus presence-only modelling

methods for predicting bird habitat suitability. Ecography,

27, 437–448.

Brussaard, L., Behan-Pelletier, V. M., Bignell, D. E., Brown,

V. K., Didden, W., Folgarait, P. et al. (1997). Biodiversity

and ecosystem functioning in soil. Ambio, 26, 563–570.

Buttner, G., Feranec, J., & Jaffrain, G. (2000). Corine land

cover update 2000: Technical guidelines. In: European

Environmental Agency (Ed.), Technical report.

Clough, Y., Holzschuh, A., Gabriel, D., Purtauf, T., Kleijn,

D., Kruess, A. et al. (2007). Alpha and beta diversity of

arthropods and plants in organically and conventionally

managed wheat fields. Journal of Applied Ecology.

Council Directive 91/414/EEC of 15 July 1991. Concerning the

placing of plant protection products on the market of the

European Communities. Official Journal of the European

Communities, 230, 1–32.

Council Regulation 2078/92/EEC of 30 June 1992. On

agricultural production methods compatible with the

requirements of the protection of the environment and

the maintenance of the countryside. Official Journal of the

European Communities, 215, 85–90.

Donald, P. F., Green, R. E., & Heath, M. F. (2001).

Agricultural intensification and the collapse of Europe’s

farmland bird populations. Proceedings of the Royal

Society of London Series B-Biological Sciences, 268,

25–29.

Hendrickx, F., Maelfait, J., Wingerden, W., Schweiger, O.,

Speelmans, M., Aviron, S. et al. (2007). How landscape

structure, land-use intensity and habitat diversity affect

components of total arthropod diversity in agricultural

landscapes. Journal of Applied Ecology, 1, 1–2.

Holzschuh, A., Steffan-Dewenter, I., & Tscharntke, T. (2008).

Agricultural landscapes with organic crops support higher

pollinator diversity. Oikos, 117, 354–361.

Hooper, D. U., Chapin Iii, F. S., Ewel, J. J., Hector, A.,

Inchausti, P., Lavorel, S. et al. (2005). Effects of

biodiversity on ecosystem functioning: A consensus of

current knowledge. Ecological Monographs, 75, 3–35.

Kleijn, D., Berendse, F., Smit, R., & Gilissen, N. (2001). Agri-

environment schemes do not effectively protect biodiversity

in Dutch agricultural landscapes. Nature, 413, 723–725.

Krebs, J. R., Wilson, J. D., Bradbury, R. B., & Siriwardena,

G. M. (1999). The second silent spring? Nature, 400,

611–612.

Marshall, E. J. P., Brown, V. K., Boatman, N. D., Lutman,

P. J. W., Squire, G. R., & Ward, L. K. (2003). The role of

weeds in supporting biological diversity within crop fields.

Weed Research, 43, 77–89.

Ostman, O. (2004). The relative effects of natural enemy

abundance and alternative prey abundance on aphid

predation rates. Biological Control, 30, 281–287.

Pain, D. J., & Pienkowski, M. W. (1997). Farming and birds in

Europe: The common agricultural policy and its implications

for bird conservation. London: Academic Press.

Payne, R. W., Harding, S. A., Murray, D. A., Soutar, D. M.,

Baird, D. B., Glaser, A. I. et al. (2008). The guide to genstat

release 11. In R. W. Payne (Ed.), Part 2: Statistics. Hemel

Hempstead: VSN International.

Robinson, R. A., & Sutherland, W. J. (2002). Post-war

changes in arable farming and biodiversity in Great Britain.

Journal of Applied Ecology, 39, 157–176.

Rundlof, M., Bengtsson, J., & Smith, H. G. (2008). Local and

landscape effects of organic farming on butterfly species

richness and abundance. Journal of Applied Ecology, 45,

813–820.

Stoate, C., Boatman, N. D., Borralho, R. J., Carvalho, C. R.,

De Snoo, G. R., & Eden, P. (2001). Ecological impacts of

arable intensification in Europe. Journal of Environmental

Management, 63, 337–365.

Ter Braak, C. J. F., & Smilauer, P. (2002). Canoco reference

manual and canodraw for windows user’s guide: Software for

canonical community ordination (version 4.5). Ithaca, NY:

USA Microcomputer Power.

Tilman, D., Cassman, K. G., Matson, P. A., Naylor, R., &

Polasky, S. (2002). Agricultural sustainability and intensive

production practices. Nature, 418, 671–677.

Tscharntke, T., Klein, A. M., Kruess, A., Steffan-Dewenter, I.,

& Thies, C. (2005). Landscape perspectives on agricultural

intensification and biodiversity – Ecosystem service man-

agement. Ecology Letters, 8, 857–874.

Weibull, A. C., Bengtsson, J., & Nohlgren, E. (2000). Diversity

of butterflies in the agricultural landscape: The role of

farming system and landscape heterogeneity. Ecography,

23, 743–750.

1

Table 1. List of farmland bird species (species nesting on the ground of arable fields,

pastures or field margins) and the corresponding requirements for assigning breeding

territories. Categories: A: at least two observations of birds displaying territorial behaviour

(foraging, calling, singing, conflicts indicating territory defence) at the same spot during

different survey rounds; B: one observation of territorial behaviour (species unlikely to be

present during all the three survey visits, for example because of long-distance migration,

or species considered difficult to observe); C: direct evidence of breeding activities.

2

English name Scientific name breeding category

Black-tailed Godwit Limosa limosa A

Corn Bunting Miliaria calandra A

Corncrake Crex crex B

Crested Lark Galerida cristata A

Curlew Numenius arquata B

Fan-tailed Warbler Cisticola juncidis A

Great Bustard Otis tarda B

Lapwing Vanellus vanellus A

Little Bustard Tetrax tetrax B

Mallard Anas platyrhynchos C

Montagu's Harrier Circus pygargus C

Meadow Pipit Anthus pratensis A

Marsh Harrier Circus aeruginosus C

Marsh Warbler Acrocephalus palustris B

Ortolan Bunting Emberiza hortulana B

Oystercatcher Haematopus ostralegus A

Grey Partridge Perdix perdix B

Pheasant Phasianus colchicus A

Quail Coturnix coturnix B

Red-legged Partridge Alectoris rufa B

Short-toed Lark Calandrella brachydactyla A

Stonechat Saxicola torquata A

Snipe Gallinago gallinago A

Tawny Pipit Anthus campestris B

Wheatear Oenanthe oenanthe A

Whinchat Saxicola rubetra A

Woodlark Lullula arborea A

Yellowhammer Emberiza citrinella A

Yellow Wagtail Motacilla flava A

3

Table 2. Results of the Generalized linear mixed models that we applied for the different

response variables: a) relationship with yield and its interaction with country; b) selected

models after forward selection considering 13 intensification variables (amount of a.i. =

amount of active ingredients); c) single effects of farm type and percentage agri-

environment schemes; d) relationships between biodiversity variables and the biodiversity

at the lower trophic level and between aphid survival and diversity and density of carabids

as possible predators (single effects). All analyses included the two surrounding landscape

variables (mean field size and percentage of land planted with arable crops); they are only

included in this table if the effect was significant. Note, that percentage agri-environment

scheme is included as part of the selected model in b) and as single effect in c).

4

response variable explanatory variable standardized effect χ2 p-value

number of plant species a yield -0.362 141.42 <0.001

yield*country 36.87 <0.001

b mean field size -0.094 6.09 0.014

% of land under AES 0.149 12.23 <0.001

herbicide applications -0.106 8.88 0.003

insecticide applications -0.105 6.15 0.013

fungicides (amounts a.i.) -0.262 31.45 <0.001

c farm type 0.403 164.96 <0.001

% of land under AES 0.326 51.97 <0.001

number of carabid species a yield -0.096 23.33 <0.001

yield*country 24.35 0.002

b % of land under AES 0.062 6.31 0.012

insecticides (amounts a.i.) -0.061 10.87 0.001

c farm type 0.040 3.98 0.046

% of land under AES 0.066 6.91 0.009

d number of plant species 0.010 15.45 <0.001

number of breeding bird species a yield -0.129 7.33 0.007

yield*country 7.46 0.383

b fungicide applications -0.127 5.71 0.017

c farm type 0.050 1.31 0.252

% of land under AES 0.067 1.56 0.211

d number of carabid species 0.036 5.05 0.025

number of carabid individuals 0.095 3.48 0.062

number of plant species 0.025 11.25 0.001

median survival time aphids a yield 0.108 6.85 0.009

yield*country 17.84 0.007

b % of land under AES -0.144 9.43 0.002

insecticides (amounts a.i.) 0.114 11.17 0.001

c farm type -0.156 2.93 0.087

% of land under AES -0.177 13.24 <0.001

d number of carabid species -0.020 6.20 0.013

number of carabid individuals 0.000 0.03 0.874

5

Table 3. Differences between study areas of the effects of yield on wild plant species

diversity, carabid species diversity and the median survival time of aphids. Estimates of the

effects of yield on the number of plant and carabid species and the survival time of aphids

are given for each study area. Different letters denote significant differences between

countries. 1 and 2 are the two different study areas in Germany: Göttingen and Jena,

respectively.

number of plant species

number of carabid species

median survival time aphids

estimate estimate estimate

Estonia -0.450 a b -0.105 a b 0.446 c

France -0.261 b c 0.106 b

Germany1 -0.511 a b -0.235 a 0.212 b c

Germany2 -0.532 a b -0.091 a b 0.052 a b

Ireland -0.129 c -0.046 b 0.200 a b c

Netherlands -0.583 a -0.023 b -0.091 a

Poland -0.299 b c -0.248 a -0.024 a b

Spain -0.180 c -0.054 b

Sweden -0.132 c -0.001 b 0.106 a b

6

Table 4. List of explanatory variables included in the analyses: explanatory variables,

variable description, units and sampling level (landscape, farm or field);

1 variables were log-transformed

2 percentage of land planted with arable crops was logit transformed, i.e. log{(percentage

arable crops + 5) / (105 - percentage arable crops)}

explanatory variable description unit sampling level

farm type conventional or organic farm

yield

standardized according to 14% moisture

content of the grains tons ha-1 field

number of crops number of crops cultivated in 2007 farm-1 farm

agri-environment scheme % area of farm with agri-environment

scheme % farm

field size1 size of sampled field ha field

pesticides (herbicides,

insecticides, fungicides) number of applications y-1 field

pesticides (herbicides,

insecticides, fungicides1) total amount of active ingredients kg ha-1 y-1 field

inorganic N fertilizer total amount of inorganic nitrogen fertilizer kg N ha-1 y-1 field

organic fertilizer total amount of organic fertilizer kg ha-1 y-1 field

ploughing ploughing (yes or no) field

farm

layo

ut a

nd

farm

ing

pra

ctic

es

mechanical weed control frequency mechanical weed control y-1 field

mean field size1 mean field size within 500m radius of

sampling points ha landscape

lan

dsc

ap

e

percentage arable crops2 % arable crop within 500m radius of

sampling points % landscape

7

Table 5. List of response variables included in the analyses: response variable, variable

description, sampling level (farm or sampling point) and distribution of the variables.

response variable description sampling level data distribution

number of plant species total number in the 3 plots sampling point Poisson

number of carabid species total number collected during 2

rounds

sampling point Poisson

number of breeding bird species only ground-nesting farmland

species

farm Poisson

median survival time of aphids time elapsed until half the aphids

had been removed

sampling point log-normal

8

Table 6 A, B. Correlation matrix (Pearson correlation coefficients) including all

intensification and landscape variables at sampling point level (A) and farm/bird plot level

(B): farm type, no. crops (number of crops) % AES (percentage land of a farm under an

agri-environment scheme), field size, herb appl (number of herbicide applications), herb ai

(amounts of active ingredients of herbicides), insect appl (number of insecticide

applications), insect ai (amounts of active ingredients of insecticides), fung appl (number of

fungicide applications), fung ai (amounts of active ingredients of fungicides), inorg fert

(amount of inorganic N fertilizer), org fert (amount of organic fertilizer), plough

(ploughing), weed control (frequency of mechanical weed control), field size 500

(landscape structure variable: mean field size within radius 500m), % arable crop 500

(landscape structure variable: percentage land under arable crops within radius 500m).

Pearson correlation coefficients higher than 0.7 are in grey (average n = 1398 for table A

and n = 251 for table B).

1 variables were log transformed

2 variable was logit transformed

9

Table 6A

farm

typ

e

yiel

d

no

. cro

ps

% A

ES

field

siz

e1

her

b a

ppl

her

b a

i

inse

ct a

ppl

inse

ct a

i

fung

app

l

fung

ai1

inor

g fe

rt

org

fert

plo

ugh

we

ed c

on

tro

l

field

siz

e 5

001

% a

rab

le c

rop

50

02

farm type 1.00

yield -0.52 1.00

no. crops 0.18 -0.03 1.00

% AES 0.46 -0.40 0.19 1.00

field size1 -0.03 0.12 0.13 0.18 1.00

herb appl -0.55 0.48 0.02 -0.27 0.15 1.00

herb ai -0.40 0.27 0.04 -0.13 0.07 0.51 1.00

insect appl -0.31 0.53 0.06 -0.18 0.10 0.38 0.29 1.00

insect ai -0.20 0.28 0.09 -0.02 0.13 0.08 0.22 0.60 1.00

fung appl -0.40 0.59 0.04 -0.23 0.15 0.54 0.35 0.69 0.46 1.00

fung ai1 -0.52 0.59 0.13 -0.24 0.13 0.53 0.41 0.66 0.44 0.83 1.00

inorg fert -0.69 0.66 0.01 -0.40 0.13 0.53 0.35 0.51 0.25 0.61 0.72 1.00

org fert 0.30 -0.07 0.11 0.03 0.05 -0.13 -0.10 -0.09 -0.11 -0.13 -0.10 -0.26 1.00

plough 0.19 -0.11 0.10 -0.09 -0.21 -0.14 0.10 0.01 0.12 0.01 0.05 -0.08 0.10 1.00

weed control 0.29 -0.39 -0.18 -0.01 -0.09 -0.36 -0.16 -0.29 -0.15 -0.38 -0.47 -0.44 0.12 -0.11 1.00

field size 5001 0.23 0.00 0.33 0.21 0.57 0.05 -0.02 -0.10 -0.02 -0.05 0.03 -0.03 0.23 -0.11 -0.06 1.00

% arable crop 5002 -0.01 0.00 0.26 -0.12 0.17 0.15 0.06 -0.11 -0.07 -0.06 0.03 -0.01 0.08 -0.01 -0.06 0.36 1.00

10

Table 6B

farm

type

yiel

d

no

. cro

ps

% A

ES

field

siz

e1

her

b a

ppl

her

b a

i

inse

ct a

ppl

inse

ct a

i

fun

g a

pp

l

fun

g a

i1

inor

g fe

rt

org

fert

plo

ugh

wee

d c

on

tro

l

field

siz

e 5

001

% a

rab

le c

rop

5002

farm type 1.00

yield -0.52 1.00

no. crops 0.19 -0.05 1.00

% AES 0.46 -0.45 0.20 1.00

field size1 -0.10 0.12 0.14 0.15 1.00

herb appl -0.60 0.48 0.02 -0.32 0.18 1.00

herb ai -0.45 0.28 0.04 -0.17 0.07 0.53 1.00

insect appl -0.35 0.55 0.07 -0.23 0.11 0.39 0.34 1.00

insect ai -0.24 0.29 0.10 -0.05 0.16 0.09 0.27 0.61 1.00

fung appl -0.43 0.59 0.04 -0.25 0.19 0.56 0.38 0.73 0.50 1.00

fung ai1 -0.51 0.59 0.14 -0.21 0.23 0.58 0.48 0.75 0.53 0.87 1.00

inorg fert -0.70 0.70 -0.01 -0.39 0.23 0.60 0.42 0.58 0.31 0.66 0.72 1.00

org fert 0.34 -0.08 0.11 0.06 0.00 -0.17 -0.12 -0.12 -0.15 -0.15 -0.11 -0.28 1.00

plough 0.22 -0.09 0.15 -0.02 -0.19 -0.13 0.10 0.00 0.15 -0.03 -0.03 -0.15 0.13 1.00

weed control 0.25 -0.41 -0.18 -0.04 -0.14 -0.39 -0.18 -0.33 -0.19 -0.40 -0.45 -0.42 0.08 -0.07 1.00

field size 5001 0.22 0.00 0.37 0.21 0.66 0.06 -0.03 -0.06 -0.01 -0.02 0.09 0.01 0.22 -0.07 -0.12 1.00

% arable crop 5002 0.00 -0.03 0.33 -0.06 0.26 0.14 0.03 -0.18 -0.10 -0.09 -0.01 -0.03 0.04 -0.08 -0.04 0.46 1.00

11

Table 7. Mean values and standard deviations (stdev) of all explanatory and all response

variables are given at the sampling point and at the farm level, i.e. the data used for the bird

analyses. Amount a.i. = amounts of active ingredients.

1 percentage conventional farms

2 percentage farms with a ploughing regime

farm sampling point

mean (stdev) mean (stdev)

farm type 77.51

yield [kg ha-1] 5401.8 (1929.1) 5398.9 (1878.9)

number of crops [farm-1] 5.0 (2.6) 5.0 (2.5)

agri-environment scheme [%] 26.9 (40.5) 23.9 (39.0)

field size [ha] 11.2 (12.1) 10.5 (13.4)

herbicide applications [y-1] 1.3 (1.1) 1.2 (1.1)

insecticide applications [y-1] 0.5 (0.8) 0.5 (0.8)

fungicide applications [y-1] 1.1 (1.4) 1.1 (1.3)

herbicides amount a.i. [kg ha-1 y-1] 0.8 (0.9) 0.7 (0.9)

insecticides amount a.i. [kg ha-1 y-1] 0.1 (0.1) 0.0 (0.1)

fungicides amount a.i. [kg ha-1 y-1] 0.5 (0.6) 0.5 (0.6)

inorganic N fertilizer [kg ha-1 y-1] 103.1 (79.2) 110.1 (78.5)

organic fertilizer [kg ha-1 y-1] 3585.5 (9332.2) 3076.1 (9110.6)

ploughing 75.32

farm

layo

ut a

nd

farm

ing

pra

ctic

es

mechanical weed control [y-1] 0.8 (1.4) 0.7 (1.3)

mean field size [ha] 17.1 (18.9) 14.1 (19.1)

lan

dsc

ape

percentage arable crops [%] 76.1 (18.9) 74.2 (19.4)

number of plant species 8.7 (6.3)

number of carabid species 6.8 (3.9)

number of breeding bird species 3.1 (2.0)

resp

on

se

vari

able

s

median survival time of aphids [h] 12.4 (7.2)

12

Fig. 1. Map of the study areas: The locations of the study areas are indicated by black dots:

Sweden, Estonia, Ireland, Netherlands, Germany (Göttingen), Germany (Jena), Poland,

France and Spain.

Fig. 2. Scatterplot of the principal component analysis (PCA) showing the 8 landscape

variables estimated around the sampling points within circles with radius 500 and 1000m:

Shannon habitat diversity (hab div), mean field size (mfs), standard deviation of mean field

size (stdev mfs), percentage of land planted with arable crops (% crops). Mean field size

(radius 500m) and percentage of land planted with arable crops had the highest correlation

with the plane defined by the two axes (longest PCA arrow).