Steps towards management on farms

CENTRALE LANOBOUWCATALOGUS

0000 0807 2635 Promotoren : Prof.Dr .Ir .E.A . Goewie Hoogleraar ind eMaatschappelijk e Aspecten van deBiologisch e Landbouw

Prof.Dr .Ir .A.H.C .va nBrugge n Hoogleraar ind eBiologisch e Bedrijfssystemen

Co-Promotor: Dr.W . Joenje Universitair Hoofddocent, leerstoelgroep Gewas-e n Onkruidecologie

Samenstelling Promotiecommissie: Prof.Dr .Ir .P.C . Struik (Wageningen Universiteit) Prof.Dr .Ir .L .Brussaar d (Wageningen Universiteit) Dr. G.R.d eSno o (Centrum voor Milieukunde, Universiteit Leiden) Prof.Dr . T.Tscharntk e (Georg-August Universitat, Gottingen) >,i^y .'.'•'..'- .' IC

Steps towards food web management on farms

FransW . Smeding

Proefschrift terverkrijgin g vand egraa dva n doctor opgeza gva n derecto r magnificus van Wageningen Universiteit, Prof.Dr.Ir. L. Speelman inhe t openbaar te verdedigen opwoensda g 6jun i 2001 desnamiddag so mhal ftwe e ind eAul a

1 > Smeding,F.W. , 2001

Stepstoward s food webmanagemen t on farms Ph.D.Thesis ,Wageninge n University, -With references-, With summary in English and Dutch ISBN 90-5808-414-0 Subject headings:agroecosystem , farming system, , food web,predator , organic farming, pestmanagement , ecological , field margin, Diptera, Carabidae, Aranae, Skylark,Alauda arvensis M/UatfZol.M^a

Stellingen

1. Biologische landbouwbedrijven hebbenvoo r organismen,hoge r ind evoedselketen ,ee n groteredraagkrach tvoo r instandhoudingva n debiodiversitei t dan conventionele landbouwbedrijven. Ditproefschrift.

Vergrotingva nhe tvoedselaanbo d aand ebasi sva n hetvoedselweb , isee nvoorwaard eo m deto p daarvant e vergroten,hoewe lterugkoppelingen , die dit effect tegenwerken ookt e verwachten zijn. Ditproefschrift.

3. Extrazor gva n boerenvoo r deecologisch e infrastructuur, ondermee rdoo r gericht beheer van sloten en slootkanten,heef t een sterk positiefefFec t op debiodiversitei t van het platteland.Dit proefschrift.

De ontrafeling van het menselijk genoom isbelangrijk , omdatdi t onderzoekd e veronderstelde bepalendero lva ngene n inlevend e systemen relativeert. StephenJay Gouldin NRC Hcmdelsblad dd 20maart 2001.

5. Deoorspronkelijk e ideeenva nd ebiologisch e landbouwzij n doorwetenschapper s eerst als'subjectie f afgedaan, maarworde ngestaa g ontdekt alsbruikbaa r en'objectie f .

Ecologisch onderzoek moetuitgaa nva nniet-begrepe n waarnemingeni nhe tvel d enmoe t deuitbreidin g van detheori e daaraandienstbaa r maken,i nplaat sva ndez evooro pt e stellen.

7. Wageningen Universiteit stoot studenten afdoo rretorisch e vragen entechnisc h vernuft, enzo u studenten aantrekken door hettone nva ntwijfe l enzor gove r actueleprobleme n in de landbouw.

8. Het opsteed sjonger e leeftijd aanspreken van rationele vermogens vankindere n leidtto t vervreemdingva nd elevend eomgevin ge nasociaa l gedrag.

Stellingen bij het proefschrift vanFran sW . Smeding,Steps towards food webmanagement on farms, 6juni 2000 Tom y father andmothe r whonurture dm yinteres tan dlov e ofnatur e andth ecountrysid e

andi nmemor yo f Peter A.va n derWerf f (1949-1996) whogreatl y encouraged mei nthi swor k Preface

"Our attention isdraw nb y ablackbir d screaming. It isJanuar y 28th and Ia m writing this thesis. Iloo k through thebedroo m window with mywif e and four-year-old son andw ewatc h a sparrow hawk with ablackbir d in hisclaws .An d Igues s that his victim is "vlek"("spot "i n English),th e female blackbird with onewhit e feather onto p of each shoulder, that lives around ourhouse . Impulsively Ira n downstairs, through the front door in anattemp t to rescue theblackbird ; but the sparrow hawk flies away carrying, of course, its prey".

That evening Ifel t abi t confused about this event. How could abiologis t studying food webs, interfere with thehun t of animpressiv e bird ofprey ? Perhaps ithelpe d met o realise more profoundly thattrophi c interactions areno t figures buthav e to dowit h life and death. Then I remembered an intriguing story from theHind u epicMahabharata , which tells of arighteou s kingwh oprotecte d a frightened dovefro m apursuin g falcon. The falcon beggedth ekin gt o lethi m eat and warned thekin g that his interference with 'Dharma' (perhaps 'natural order', in this thesis) would create fate. However, the king stuck to hisprinciple s andpromise d the falcon aquantit y ofhi sow n flesh that equalled the weight ofth e dove. Eventually thekin g was eaten to thebone s andbot h the falcon andth e dove laughed atth ekin g and flew away.

Although theMahabharat a story is mysterious tome ,i traise s an association that Ihav e wanted topu t into theprefac e of thisthesis .I tconsider s thatther e isn o doubt that man,a s part ofnature , should interfere in to feed himself.However , ifhi s interference goesbeyon d his proper needs, and are, sot o say, 'none ofhi sbusiness' , hema y not really know what he isdoin g andma y therefore unwittingly disturb thenatura l order. Thusth e work ofmoder n biologists like Rupert Sheldrake convinced me that current has alimite d comprehension of living systems. A'righteou s ecologist'(wh o is for example interfering with food webs) should, therefore, remain modest, whilst keeping anope n mind and anope neye , both when looking atnatur e and athi s ownmotives . This might help topreven t his worst errors.Fortunatel y that afternoon my attempt to interfere with thehuntin g sparrow hawk failed, because Iwoul d not haveknow n how to care for aseverel y wounded blackbird.

The small maps ofth eFlevolan d polder inthi s thesis have aspecia l meaning tome ,becaus e they echoth elarg eembroidere d map ofth eNoordoostpolde r thathun g onth ewal l ofth e dining room when Iwa s achild . My grandfather devoted hisprofessiona l life topolde r reclamation. Hispolde r service employed outstanding ecologists like W.Feeke s andD . Bakker, who were engaged onbot h applied and fundamental questions,whic hwa s much appreciated by my grandfather (som y father told me).Howeve r the aims of reclamation concentrated onth emateria l concerns of food and labour, reflecting needso f society attha t time andresultin g in large scaled rational . Following this tradition my father contributed to the development of agricultural education. It ism y sincere and modest wish that mycurren t work may contribute to the implementation of awide r scope of agricultural production, and address thephysica l aswel l asth e emotional and spiritual needs.

Without the support of many people, this thesiswoul d never havebee n accomplished. Ia m grateful to: Mypromote r Eric Goewie who gavem e space and confidence tobegi n theproject , andwh o strengthened mybackbon e and inspired me inmoment s ofhesitation . My promoter Ariena van Bruggen who, after her arrival in Wageningen,neede d only ahal f word to understand the contents ofth ewor k andhelpe d me toproces s thepiled-u p field data. My co-promoter Wouter Joenje, withwho m Ishar e common interests in ecological theories and field biology, particularly ofpionee r successional stages andth epolders . The funding organisation (LNV-DWK) for supporting asmal l research group sotha t they could embark upon aprojec t with anuncertai n outcome.Howeve r thisresearc h groupwa s much strengthened thanks toth e invaluable co-operation and friendship ofth eresearcher s of Plant Research International (formerly IPO-DLO):Kee s Booij, Clasien Lock and Loesde n Nijs.

The supervision committee ofth eprojec t 'Food web management on farms', that enthusiastically guided theresearc h by it's reflections onou rplan s and results. The committee included: Geertd e Snoo (CML-Universiteit Leiden),Fran k vanBell e (Vereniging Natuurmonumenten), Paul Aukes (IKCNatuur) ,Pau l vanHa m (IKCLandbouw) , Tibbe Breimer (LNV-DWK), aswel l asKee s Booij, Wouter Joenje (the chairman) and EricGoewie .

Myresearc h assistant, AndreMaassen , whowa s apilla r of strength with all our laborious activities,th e designer ofinstrument s and forms, companion ofth ehundred s of kilometres walked whiledat a collecting, and theon ewh o could, if asked, always help meremembe r the aimso fmy work .

TheMS c students who didreconnaissanc e work for the thesis: ElkeBoesewinkel , Jacinta de Huu,Nathali e Reijers, Julien Cothenet, Edwin Coomans, HongNan , and inparticula r Bart Venhorst who also supported the field work andinsec t identification.

People with whom Iexchange d ideas orwh oprovide d valuable information that contributed tothi sthesis :Gerar d Oomen,Wille mBeekman , JanDie kva nMansvelt , Sjaak Wolfert, DorineDekker , Derk Jan Stobbelaar, Karina Hendriks,Dark o Znaor, Prof.R.A.A . Oldeman, Ruurdtje Boersma, Manolis Kabourakis, AnorFiorini ,Han s Esselink (who named 'Voedselwebbeheer'), Joop Schaminee, John Vandermeer, Yvette Perfecto, Julian Park, Nigel Boatman, Flip vanKoesveld , JanJaa p van Almenkerk, Wolter van der Kooij, Rob Boeringa, HenkKloen ,Yvonn e van den Hork,Moniqu e Bulle, members ofth e PE-Phd discussion group 'Plant and Crop ', Paula Westerman, Eefje den Belder, Walter Rossing.

The farmers who allowed me access tothei r farms and who gave their timet o answerm y questions; inparticular , especial thanks aredu et o Maart en Piet van Andel, Carl enRin i Ferket, Jan enMarlee n van Woerden-Zeelenberg, Lex enJanni e Kruit, Henk Leenstra, Evert- JanRienks ,Jansj e Timmermans,Douw eMonsma ,He m and Marjon Cuppen, staff and residents of'Thedinghsweert ' for their hospitality and interest. Joop Overvest, Johan Jorink, Harry Alting andcolleague s (staff andworker s ofth eA.P . Minderhoudhoeve), Swifterbant, and Egbert Lantinga (chairman ofth e research team) for discussing myresearc h andimplementin g recommendations onth e ecological farm.

Allth epeopl e who did the organisation behind theproject , particularly Gijsbertje Berkhout. Thestaf f ofUnifar m whower e alwayswillin g to support the project. Colleagues ofth e Biological Farming Systems groupwh o accepted my absence, whilst encouraging met o finish mythesi s research.

NickyCampbel l ofCombpyne , Devon, UK for correcting the English. Wampieva n Schouwenburg for help with the final editing.

Heleen Boerswh o helped met o collect myself. And last but not least my dear friends and family who supported and encouraged me, andwit h some ofwho m Icoul d share my aspirations: Annelies, Francis,Philip ,m ybrothe r Guido,m y uncle Kees Cleveringa; but most of all, Elisean dTim owh o gavem ewarmt h atth e coldest moments ofm ypersona l 'Winterreise'. Abstract

This paper isth e report of four years ofresearc h onth e functional group composition ofth e animal inrelatio n to farm and ecological infrastructure (E.I.)managemen t on organic arable farms. Theresult s aremainl y based on data of ground dwelling arthropods obtained bypitfal l trapping, density data ofvegetatio n dwelling arthropodsb y vacuum sampling and density datao finsectivorou sbird sb yterritor y mapping. Arthropods were collected inwhea t crops (representing the crop area) and on the adjacent canalban k (representing the E.I.);th ebird , farm and E.I. variables were measured atth e farm level. Study areasinclude d intota l 18farm s with varying extentso forgani c duration, croprotatio n intensity, and quantity and quality ofE.I .

Thehypothesi s ofth eresearc h wastha tth e food web structure of anorgani c arable farm with long organic duration aswel l aswit h an improved E.I. (i.e.enlarged , latemown) ,woul d show ahighe r abundance of meso- and macrofauna ofbot hherbivorou s and detritivorous functional groups. These enhancedprimar y groupswer e expected to carry ahig hpredato r abundance at both secondary (i.e.invertebrates ) aswel l astertiar y (i.e.birds ) levels.Wit h regard to the crop areas itwa s found, in contradiction toth ehypothesis ,tha t weremos t abundant in crop areas of recently converted farms and oforgani c farms withintensiv e croprotation ; this abundance was associated with invertebrate predator abundance and diversity. In accordance with thehypothesis , some evidence was found for increased and related epigeicpredato r abundancerelate d to extensive cropmanagemen t on the farms of long organic duration. Whilst studying the E.I., an increased abundance of dwelling predators and also was found in improved E.I. However K- herbivore numbers did not increase in the improved E.I.whe nthe y werecompare d to the traditionally managed E.I. The summer abundance of epigeicpredator s was also not related to an improved E.I. Field studiesprovide d some evidence for the dispersal of functional groups, abundant inth e E.I., towards the crop area. However, the effects of crop conditions onth e arthropod abundance inth e crop areawer e observed to offset the influence ofth e E.I.Bir d studies atth e farm level revealed positive correlation between bird functional groups anda combination ofcro p area and E.I. characteristics. Bird density was found tob e positively associated with high arthropod abundance inth e E.I. vegetation canopy. Observations also suggested positive correlation to anincrease d herbivory inth e crop area ofth e long duration organic farms that had an intensive crop rotation.

Aproposa l for adescriptiv e ortopologica l farm food web is drawn from field observations as well asfrom reference s in literature.Prediction s aremad e for four different farm food web structures that express four extremes oftw o environmental gradients, which correspond toth e length oforgani c duration and the amount/quality ofth e E.I.Wit hreferenc e tofield observations important themes inth e food webtheor y are discussed, including the indirect effects of subsidised detrital food chains onherbivor e abundance and consequently onbir d abundance, aswel l asth epossibl e effects of intra guildpredatio n on arthropod functional group composition.

The implications ofth e study aretha t organic duration andth e amount/quality of theE.I .ma y contribute to improving services andt o aimsbase d onnatur e conservation. However an optimisation ofth e farm food webwit h regard to ecosystem servicesma yno t necessarily improve nature conservation values. Iti s argued that increased understanding of the farm food web and itsmanagemen t is likely to support the development of multi-species agroecosystems that integrate improved ecosystem services andnatur e conservation goals. Notes

Thisprojec t was funded byth e Ministry ofAgriculture ,Natur e Management andFisheries , Department of Science andKnowledg e Dissemination.

Chapters 2,3, 4 and 6hav ebee n submitted for publication in internationaljournals . of contents

Chapter 1: General introduction 1

Chapter 2: Functional group compositions ofvegetatio n dwelling arthropods inrelatio n to ecological infrastructure and time sinceconversio n to organic farming Smeding, F.W., A.H.T.M. Maassen& A.H.C. vanBruggen 5

Chapter 3: Shifts in functional group composition of ground dwelling arthropods in relation to ecological infrastructure and conversion to organic farming Smeding, F.W., CAM. Lock &C.J.H. Booij 31

Chapter 4: Insectivorous breeding birds onorgani c farms inrelatio n to crop area, ecological infrastructure and arthropods Smeding, F.W. &C.J.H. Booij 51

Chapter 5: Effect offield margi n management oninsectivorou s birds, aphidsan d their predators in different landscapes Smeding, F.W. &C.J.H. Booij, 1999. Aspectsof Applied Biology 54: 367-374. 75

Chapter 6: Aconcep t of food web structure inorgani c arable farming systems Smeding, F.W. & G.R. deSnoo 83

Chapter 7: Farm-Nature Plan: ecology based farm planning Smeding, F.W. &W. Joenje, 1999. Landscapeand Urban Planning46: 109-115. 101

Chapter 8: General discussion 109

References 117

Summary 131

Samenvatting 133

Curriculum vitae 137 Chapter 1:Genera l introduction

This thesis ismotivate db y concerns about ,du e to agricultural intensification (Altieri, 1994,1999; Paoletti, 1999;Woo d &Lenne , 1999;Vandermee r etal, 1998) andi s primarily concerned withplan t and animal species indigenous to agricultural landscapes. Biodiversity loss has been thought to result in artificial ecosystems, requiring constant human intervention and costly external inputs.Fro m thisperspectiv e biodiversity has economical value,relate d to its ecological services,e.g. pest control andbioti c regulation of fertility (Altieri, 1994, 1999;Pimente l etal, 1995).Natur e conservation worthy speciesma y haven o evident significance for the functioning of agroecosystem (Duelli etal, 1999).Howeve r it canb e argued that has,t o acertai n extent, responsibility for all species and communities which co-evolved with farming during around 10,000year s (Wood &Lenne , 1999),irrespectiv e ofthei r utility.

One important solution for thereversa l ofbiodiversit y loss in conventional agriculture of industrialised countries isth e development of farming systems that are economically based on utilisation ofbiodiversit y andtha t alsoharbou r conservation worthy species.Thi s idea is compatible with theconcep t ofmulti-specie s agroecosystems (Vandermeer etal, 1998; Altieri, 1994, 1999;Almekinder s etal, 1995;Swif t &Anderson , 1993).Developmen t of organic farming systems is apossibl e implementation ofthi s concept (Stobbelaar &Va n Mansvelt, 2000; Vandermeer, 1995; Van derWerff , 1991).Th e adoption of thisconcep t in the crop area may be complemented by an appropriate management of the semi-natural onth e farm, which isdefine d asth e ecological infrastructure (E.I.) (Smeding & Booij, 1999).

A food web approach is anappropriat e method for investigations into the higher integration levels ofecosystem s (Polis &Winemiller , 1996;Pimm , 1991; Gezondheidsraad, 1997).Thi s approach istherefor e suitable for fanning systems research. Progress in scientific understanding of food webs iscurrentl y expanding into as isdemonstrate d by articles on interactions between detrital andherbivor e subsystems (Brussaard, 1998;Wis eet al, 1999)an d among predator functional groups (Tscharntke, 1997).Thes einteraction s may affect the abundance of functional groups,cro p performance aswel l asdecompositio n rates. Complementary to functional analysis,a foo d web approachmigh t also offer a comprehensive framework for the scattered information on farmland species andhabitat s (De Snoo& Chaney, 1999).

A simplified set ofhypothese s with regard to food web structure onorgani c farms served as the starting point for thefield researc h of thisthesis :i twa s expected that the duration of organicmanagement , extensive croprotatio n aswel l asa nimprove d E.I. (i.e.enlarge d area and latemowin g date),woul d relatepositivel y to the abundance ofnon-pes t primary arthropod functional groups (Kromp& Meindl , 1997):th edetrita l webwoul d beenhance d by the increased input of organicmatter , consisting of organic manure, compost and crop residues (ofcereals , leypasture) ;non-pes t herbivore numbers would be enhanced bycro p diversity aswel l aswee d diversity (Hald &Reddersen , 1990)an d inparticula r would be boosted by thevarie d plant tissues (Curry, 1994;Tscharntk e &Greiler , 1995)whic h occuri n improved E.I. Consequently predator andparasitoi d numbers onth e farm would be supported. An accumulated high invertebrate density on the farm would support insectivorous vertebrates, particularly birds (e.g. Poulsen etal, 1998). The objectives ofthi s thesiswere : - To analyse and conceptualise food webs onorgani c arable farms that encompass the important above- andbelo w ground functional groups and that canb erelate d to farmers' decisions about, for example, crop rotation and manure management, farm lay-out and field margin management; - To linkknowledg e ofth ebiolog y of groups of individual species,cro p and vegetation management to food webtheories , andt o identify possible food webmediate d factors that might influence the abundance of functional groups on the farm; - To suggest strategies for development of farming practices that strivet o promote ecological services and conservation ofspecies .

Tomee t the objectives, field research and analysis of literature datawer e undertaken. The study area involved organic arable farms that weremainl y situated in Flevoland. This region was chosenbecaus e ofit suniformit y intopograph y andhistor y andbecaus e ofth e occurrence of around 75organi c farms including farms with improved ecological infrastructure, resulting from aprototypin g research project (Vereijken, 1997, 1998).

Empirical investigations were concentrated onparticula r sections of thefarmlan d community. Functional group compositions (e.g. detritivores, herbivores and predators) within these sections were assessed. All three empirical chapters start from the same initial set of hypotheses explained above,an d reflect to acertai n extent on each other inthei r discussions. Observations onvegetatio n dwelling arthropods onthirtee n farms (Chapter 2) arepresente d first, because this section included largeproportion s of allthre emajo r functional groups, involving especially herbivores and connected predators. Observations ongroun d dwellers on eight farms (Chapter 3) included numerous epigeic predators that are supposed tohol d ake y position inth e farm food web.Howeve r this group appears tob emor e strongly related toth e detrital subweb than toth eherbivor e subweb. Densities ofbir d territories onte n farms (Chapter 4) are related to variables ofbot h vegetation dwelling and ground dwelling arthropods aswel l ast o farm and ecological infrastructure traits.Bird s represent vertebrates,tha t might be supported by an increased abundance ofpre y from the herbivore subweb aswel l asth e detrital subweb.

Chapter 5include s alandscap e comparison between ayoun g polder landscape (i.e.Flevoland ) and alandscap e that hasbee n farmed for many centuries (i.e.rive r region).Thi s explorative studyplace s the observations inFlevolan d into abroade r geographical context.

Aproposa l for adescriptiv e ortopologica l farm food web (Chapter 6) is drawn fromfield observations aswel l as from references inth e literature. Important themes inth e food web theory are tentatively applied tothi spreliminar y model, explaining differences between local farm food web structures andho w they arerelate d to farm and/or E.I. management. The initial set ofhypothese s onwhic h thefieldwork wa s based, ismodifie d into more elaborate hypotheses.

An additional chapter (Chapter 7) deals with plans for the coexistence of farm and natural life. Itpresent s apragmati c advisory instrument that structures expertjudgement . The instrument isbase d on acombinatio n of ecological theory andfield biology . However results ofth e food web study areno t included since the instrument (Smeding, 1995)wa s developed earlier and was therefore anincentiv e to the food web studies ofthi sthesis .

Thediscussio n chapter (Chapter 8)consider s briefly theimplication s ofth eresearc h for the development ofmultispecie s farming systems. Itmus t benote d that compared to the chronology ofth e investigations, chapters 2-5 arei n reverse order. Theprojec t started with an assessment of farms in different landscapes (1997). Howeverth e study areainclude d toomuc hvariatio n andto o few sitest ob e analysed statistically. Therefore investigations inth e following year (1998) were confined to amor e uniform area inFlevoland . However therelation s between ground dwelling functional groups and the farm and E.I.variable s were, atfirst glance ,difficul t to interpret and showed little coherence with thebir d data.Th eherbivor e subweb ofth e farm food web was expected, and wasals ofound , to givebette r distinctions, withregar d toinvestigate d variables. Consequently vegetation dwelling arthropods were assessed in 1999. Chapter 2:Functiona l group compositions ofvegetatio n dwelling arthropods in relation to ecological infrastructure and timesinc e conversion to organic farming

F.W. Smeding,A.H.T.M . Maassen &A.H.C.va n Bruggen Department of Plant Sciences, Biological Farming Systems group, Wageningen University, Wageningen,The Netherlands

Abstract Hypotheses on the relation of farm management and ecological infrastructure (E.I.) characteristics to the functional group composition of vegetation dwelling arthropods in the crop and E.I. were tested on organic arable farms. The study area included 13farm s that varied with regard to the studied traits. Arthropods were collected by vacuum trapping in wheat crops and adjacent boundaries. Observations in the crop area suggested that both organic duration and farm 'intensity' affect the arthropods. In contrast to expectations from the literature, K-herbivores and associated predators were most abundant in the luxurious crops of recently converted farms. Long duration organic farms with an intensive management, seemed tob e most prone to r- herbivore outbreaks, associated with species-diverse but not numerous predators. In support of the expectations, some evidence was collected for an increase of detritivores on long duration farms with an extensive crop rotation. Observations in the E.I. showed that, according to the expectations, improved management (i.e. large area and late mowing) enhanced predators and also detritivores. However, herbivore abundance was similarly high in traditionally managed E.I. Effects ofhig h abundance of arthropod functional groups in the E.I. on dispersal into the crop, were largely offset by crop conditions. Possible indirect effects of subsidized detrital food chains on herbivore abundance, as well as effects of intraguild on arthropod functional group composition, are discussed with reference to field observations.

1. Introduction

1.1. Agricultureand biodiversity

Thebiodiversit y of farms and agricultural landscapes in industrialised countries has decreased dramatically (e.g. Andreassen etal, 1996;Fulle r etal, 1995;Aebischer , 1991).Thi s biodiversity-loss is amatte r of great concern for ecologists, agriculturists aswel l aspolitician s (Paoletti, 1999;Woo d &Lenne , 1999).T o reverse this trend ofbiodiversity-loss , farming system innovation isprobabl y needed (Altieri 1994;Vandermee r etal, 1998;Almekinder set al, 1995;Swif t &Anderson , 1993).On eo fth e optionst o enhance biodiversity ist o promote on-farm natural areas that couldb e valuable for bothproductio n aswel l as for nature conservation aims.

The total area of semi-natural habitat onth e farm is also called the 'ecological infrastructure' (E.I.). Field margins are amajo r constituent ofth eE.I . (Smeding &Booij , 1999). Agricultural production could be enhanced by various ecological functions ofthi s E.I., such as regulation ofpest sb ypredator s andparasitoids ,nutrien t cycling by soil organisms, andpollinatio n by various arthropods. For example,biodiversit y of arable field margins significantly contributed to densities onth e farm ofpredator y invertebrates,butterflies , birds and mammals(e.g. Boatman etal, 1999;Joenj e etal, 1997;LaSalle , 1999). 1.2. Farmmanagement andon-farm communities

Despite the increasing amount of information on farmland species and vegetation, there is still fragmentary knowledge of therelationship s between farm management practices and farm communities (De Snoo &Chaney , 1999).Fo r example, crop rotation, fertility level,wee d control, field size, anddistributio n andmanagemen t of field margins may all affect farm community composition.

Management practices onorgani c farms corresponding to production guidelines 2092/91i n the ECcountrie s (Anonymous, 1991)ar e essentially different from those on conventional farms. Inparticular , maintenance andpromotio n ofbiodiversit y is inherent to thephilosoph y oforgani c farming. Therefore, organic farms could function as a stepping stone onth eroa d to developing rational,biodivers e farming systems (Stobbelaar &Va n Mansvelt, 2000). However, information onth etrophi c structure of animal communities at the farm level is still scattered andincomplet e for conventional aswel l asorgani c farming systems (e.g. Kromp& Meindl, 1997;Isar t &Llerena , 1995).Bot h empirical and experimental research atth e farm level isrequire d tofill th e gaps inth eknowledg e of effects of conversion from conventional to organic farming and offield margi nmanagemen t practices on agroecosystem community composition.

Research addressing thewhol e 'biocoenosis' isneede d to understand the effects of management practices on agroecosystem functioning (Biichset ai, 1997).Recen t advances in research of food webs (Polis &Winemiller , 1996),includin g some studies in agricultural (DeRuite r etai, 1997;Tscharntke , 1997;Wis e etal, 1999),ar eprovidin g inspiring examples ofho w (agro)ecosystems canb e approached at integration levelshighe r than the species-level. Some ofthi sresearc h indicates that food webs canmediat e the effects of environmental stress factors on species (Winemiller &Polis , 1996).

1.3. Hypotheses andobjectives

Detritivorenumber s areenhance db yth eincrease d organicmatte r input, involving organic manure, compost and crop residues (of cereals, and leypasture)(e.g . Pfiffner &Mader , 1997; Weber etal, 1997; Idinger etal, 1996;Heimbac h &Garbe , 1996).Non-pes t herbivores are enhanced by crop andwee d diversity (Hald &Reddersen , 1990;Moreb y &Sotherton , 1997; Andow, 1988).Moreover , the effects oforgani cduratio n may continue towor k cumulatively overyears ,wit h regard tobot h detritivores (Idinger &Kromp , 1997)a swel l as herbivores (Hald &Redderson , 1990).Th e abundance ofprimar y functional groups would support predacious functional groups ofe.g. epigeicpredator s (Kromp, 1999; Wise etal, 1999), predacious flies andparasitoid s (Idinger &Kromp , 1997). The increase of predator abundance may subsequently depress pest-herbivores (Altieri, 1994),provide d that crops areno t too much affected by herbivores (Van Emden, 1988).Base d onthes e observations, we hypothesise that farms that have been organic for many years,hav ehighe r densities ofnon - pestprimar y functional groups (both detritivores andnon-pes t herbivores) and their predators inth e crop area than farms that were converted recently.

Additionally wehypothesis e that appropriately managed E.I. will support largenumber s and diversity ofnon-pes t herbivores,predator s andparasitoid s both in field margins and crop areas. An increased variety inplan t tissue quality and amor e varied vegetation structure are assumed tob eke y factors (Curry, 1994;Febe r etal, 1999).Als o mown grass strips, less complex than most otherboundar y habitats, contribute to resources for these groups (Barker etal, 1999;D e Snoo &Chaney , 1999).A n increased amount and spatial density ofE.I . 6 affects the inth e adjacent crop area, involving both herbivores (Holland & Fahrig, 2000) and predators (Lys &Nentwig , 1992;Sunderlan d etal, 1996a). In addition to organic duration, crop rotation intensity and E.I. characteristics (with regard to quantity and diversity), other farm management factors are expected to influence the functional group composition ofvegetatio n dwelling arthropods. For example the effects of the studied factors mayb e offset or enhanced by cropperformance . Totak ethes e influences into account some crop variables, indirectly related tovariable s central inth e hypotheses, are also included inth e study.

However, the above-stated hypotheses about the effects oforgani c farming duration andE.I . management on vegetation dwelling arthropod functional groups have notbee n tested so far. Thus, the objectives ofthi s studywere : 1. to determine the density anddiversit y ofvegetatio n dwelling arthropod functional groups inth e crop area and semi-naturalfield margin s onorgani c farms, and relate the density and diversity of these groups to crop area and ecological infrastructure characteristics; 2. to determine ifther e are distinct species compositions ofvegetation-dwellin g arthropod functional groups among farms differing incro p area and ecological infrastructure traits; 3. torelat e abundance and diversity of vegetation arthropod functional groups toth e distance from the ecological infrastructure.

Our study was focused onvegetatio n dwelling arthropods because this group comprises both detritivores, herbivores,predator s andparasitoids . Inman y studies the herbivore subweb is not well represented, particularly whenpitfall s areused , which primarily collect epigeic predators (Duelli etai, 1999).

N

W -*4^" E

S

10 20 Kilometers s

Fig. 1.Study area. Theinvestigated organic arablefarms are represented by closed circles. 2. Materials and methods

2.1. Studyarea

The study was done on organic farms andon e integrated farm inth eprovinc e of Flevoland, TheNetherlands . The farms are located within a28 0km 2 areai nth epolde r Oostelijk Flevoland (Fig.l). Thispolde r wasreclaime d in 1954an d isdominate d by agricultural landuse. The soil is acalcareou s clay ofmarin e origin. Flevoland was chosen as astud y area because itha s aconcentratio n of 75organi c arable farms that are similar with regard to their history and topography.

Inth e study area 13farm s were selected that differed invariou s farm structure and ecological infrastructure (E.I.) traits (Table 1).Togethe r the farms comprised 656h a arable .Th e selection included eleven commercial organic farms andtw o experimental farms. Three ofth e selected arable commercial farms reared livestock which were fed onhom eproduce d fodder inth ewinter , but weregraze d elsewhere during theres t ofth e year (farms T, Kan d S).Th e experimental farms ofWageninge n University (farms E and I) aremixe d dairy and arable farms; farm Ei sorgani c and farm Ii sintegrated . Farm Ii sdivide d intotw opart sb y aroa d and only the northern part was included inth e study area. Three commercial farms (T, Lan d S)wer e Ecological ArableFarmin g Systems (EAFS)prototyp e farms during 1992-1997 (Vereijken, 1997, 1998).

2.1.1. Crop area characteristics The organic duration ofth e selected farms ranged from onet o sixteen years.Ploughin g is generally at25-3 0 cmdepth . On allth e farms a6- 7 yearcro p rotation ismaintained , including per farm 11-41% cereals, 0-18% leypasture , 12-43%traditiona l lifted crops (potatoes, sugar beet) and 25-67%vegetabl e crops.Th evegetable s grown weremainl y onions,peas , sweet corn,variou s cabbages, andrunne r beans. The farms with themos t intensive crop rotations had up to 60%vegetable s and therefore alowe r percentage (<20%) of gramineous crops (cereals,an d leypasture) .Becaus e the input of nutrients (N,P,K)ha st o be adjusted to crop requirements, and vegetables need morenutrient s than cereals, intensity of croprotatio n is assumed torelat et o the amount ofN application atth e farm level. Compared with other crops, gramineous crops provide alarg e amount of carbon in crop residues andrequir e few soil disturbing operations.

2.1.2.Ecological Infrastructure characteristics TheEcologica l Infrastructure (E.I.) of arable farms inFlevolan d ismainl y determined by canals andbank s (together c. 3-4 mwidth) ,tha t are laid out as linear herbaceous boundaries adjacent toth e crops. On five farms, boundaries arewide r because of adjacent (2-3 mwide ) grass strips,ofte n including clovers (Trifolium spp.),o non e orbot h sides of thecanals .Far m E also has grass strips between crop plots.Percentage s of E.I.highe r than c. 2%(Tabl e 1) were mainly due toth epresenc e ofthes egras s strips.Onl y farms Lan d Epossesse d a boundary planted with ahedgerow . Spatial density ofE.I .i spurposefull y increased on farm E (Smeding &Joenje , 1999) and farm L. Thespatia l arrangement ofE.I . on the other 6 farms mainly reflects landscape scale.

Extreme types ofvegetatio n management are: - 'traditional management' involving three or four more cuts peryea r with aflai l mower leaving the shredded cuttings in situ; - 'latemowin g management' involving one or two cutspe ryea r after June 21st with a finger- barmowe r with hiab grab for removal ofcuttings . Thegras s strips along thecanal s in'lat e mowing management' are required for transport. The E.I. area is,therefore , related toth e mowing date.I nthi s article theE.I .wit h enlarged area and late mowing isdefine d as'improve d E.I.'. The improved E.I. onth e four farms A,T ,L , and Dwa s created during the EAFS-prototyping research program in 1992-1997 (Vereijken, 1997,1998). The improved E.I. onth e experimental farm E started in 1995 simultaneously with conversion ofth e conventional dairy farm to an organicmixe d farm (Smeding & Joenje, 1999).

Table1 Thecharacteristics of thefarm, the wheat crop and the ecological infrastructure (E.I.) on the thirteen selected farms. 00 Q c/1 op . & & _; ^ _; O u o 3 ui w 00 ui o en 1™ rt 3 c c J3 T3 o o .3 3 o rt l I ed <4-l 00 a O g 00 "3 o t-H l Vi .2- o I-* 3 o > > 1 > <4- o V3 3 00 c a Oo o VI o o 'o > 3 &Q O. Q o Q C/3 O Variable: dur area int ccov ucov uspn mow eicov eispn unit: year % % % % n/m2 month % n/m2 Farm: level/habitat: farm farm farm crop crop crop E.I. E.I. E.I. K organic, arable 16 1.8 30 71 28.6 1.8 5 64 2.1 J organic, arable 11 4.4 26 64 7.1 4.0 6 92 4.4 S organic, arable 11 3.2 35 76 7.5 2.7 7 89 5.8 L organic, arable 11 3.1 18 66 7.8 3.1 7 88 5.1 O organic, arable 11 1.1 50 73 5.1 4.4 6 89 4.9 T organic, arable 9 2.3 31 69 9.4 2.7 9 92 4.2 F organic, arable 8 1.4 35 28 2.3 2.6 5 73 2.7 E organic, mixed 5 4.5 55 86 16.4 3.6 11 97 5.1 B organic, arable 2 1.4 17 86 0.3 0.4 5 91 4.4 R organic, arable 2 1.2 11 63 25.6 3.6 5 76 3.2 H organic, arable 1 1.5 27 78 5.7 1.3 5 96 4.0 Z organic, arable 1 1.2 17 79 0.0 0.2 5 90 3.4 M integrated, mixed 0 2.0 65 87 1.7 1.6 5 68 6.2

Dominant plant species in canalban k swards were:couc h (Elymus repens),perennia l meadow grasses (Festuca rubra,Poa trivilais, Loliumperenne, Agrostis stolonifera), commo n reed (Phragmitis australis)an d afe w perennial herbs:dandelio n {Taraxacum officinale) and stinging nettle (Urticadioica). Canal s with traditional management have alow , relatively open, speciespoo r vegetation (<10 species/4 m2), dominated by a few grass species. Late mowing particularly related to increased vegetation inJun e andJul y asexpresse d by vegetation height (upt o 1-3 m) and cover (upt o 90-100%).Howeve r latemowin g doesno t necessarily relatet o increased higher plant for two reasons: a)tal l grass species may dominate the sward;b ) swards oftraditiona l management may locallyb e cutt o ground level and therefore be invaded by ruderals and annuals (increased speciesnumbe ru p to 15species/ 4 m ),e.g. thistles (Cirsium arvense, Sonchus arvensis),annua l weeds(e.g. Sonchusasper, Polygonum aviculare) an d annuals that arerarel y pernicious weeds in Flevoland (e.g. Myosotisarvensis, Veronicapersica, Cardamine hirsuta). Highperennia l plant diversities found in some locations may relate to various interrelated factors: reduced vegetation , removal ofth ehay , nutrient buffering by the grass strip, and long organic duration. However, also,artificia l species introduction is involved; in the EAFS-project, around 90differen t nativeperennia l dicotswer e artificially seeded inth e canalbank st o obtain higher plant diversity and flower abundance ofplant s with limited dispersal capacity (Vereijken, 1998).Althoug h few species were able to settle well(e.g . Seneciojacobea, Crepisbiennis, Heracleum sphondylium), the introduction clearly affected species diversity: vegetation including siteswit h >15 species/4 m2 were confined to prototype farms.

2.2.Sampling and measurements

2.2.1.Arthropods Sampling ofarthropod s was done inwhea t crops and in adjacent canalsbank s (E.I.). Spring wheat (Triticum aestivumL. ) was chosen asa representativ e crop asthi s crop is common on organic arable farms inFlevolan d and arthropod numbers in cereals are large compared to numbers inothe r common crops likepotatoe s and onions (Booij &Noorlander , 1992). Within onewhea tfield o n each ofth e 13farm s one 50x30 m sampling areawa s selected; this area was 50mete r longborderin g acana lb y 30m perpendicula r totha t canal.Nin eplot s of 1 m2a t distances of 0m (E.I.), 10m (wheat) and 30m (wheat) from the canal bank were selected for measurement of arthropods, crop,wee d and E.I.-vegetation variables.

Distance from theto p ofth e canalban k to the crop edge ofth ewhea t could vary between farms, duet oth eoccurrenc e of grassy strips (farms T,L , S,E ,I ,J) , asteril e strip (H),a recently drilled flower strip (F) or atranspor t track (O). Sampling was done once, inth e periodfrom th e 7th ofJun eunti l the 6th ofJuly , following themethod s of Reddersen (1997) andMoreb y &Sotherto n (1997). Vegetation dwelling arthropods were collected by means of avacuu m sampler (ES2100 , Echo LakeZurich , U.S.A.) for 120s i n each 1 m sampling area. The air flow ofthi s sampler was 8.5 m3i n a sampling area of0.0 1m 2, which corresponds to the air flow of the apparatus recommended by MacLeod etal. (1995) .A ne t (pore size:0.8x0. 8 mm2) was inserted inth e suction tube.A squar e of 1 m2 area and 0.4 mhigh , made of four multiplex boards,wa s placed inth evegetatio n before switching onth e sampler.

The sampleswer epu t in aplasti c bag, transported in acoo lbo x and stored in adee pfreeze (-20 °C).Sample swer e cleaned inth e laboratory. Litter was removed and soil waswashe d awaywit hwate r in asiev e (pore size: 1x 1 mm2). Slugs and snails (Gastropoda) and springtails (Collembola)wer e discarded because vacuum sampling isno t effective for quantification ofthes e groups (Potts &Vickermann , 1974).Arthropod s were stored in95 % alcohol. Individuals were sorted to taxa at order- or family-level andplace d into five different arthropod functional groups: - Detritivores, including mainly adult diptera thathav e detritivorous larvae, and woodlice (Isopoda); - K- or senescense feeding herbivores (White, 1978) that feed onmatur e plant tissues with high C/N ratios andten d tohav e aslowe r development, largerbod y size (Crawley, 1983) and defensive traits (Power etal., 1996).Herbivore s which are an important constituent of 'chickfood' for farmland birds (Moreby etal, 1994), like sawflies, lepidopteran larvae, hoppers,bugs , leafbeetles ,weevil s and grasshoppers belong to thisgroup ; - r- or flushfeeding herbivores (White, 1978)tha t feed on youngtissue s with lowC/ N ratios and lowfibre conten t (White, 1978).The y generally have asmal lbod y size, fast development and good dispersal abilities. For reasons of simplicity, r-herbivores were 10 represented by onlyAphididae; in few available studieso norgani c crop communities (Kromp &Meindl , 1997), aphids corresponded most clearly to the definition; - Predators including both polyphagous and oligophagous species; - Parasitoids, almost exclusively represented byParasitica (Hymenoptera).

Extra efforts were madet o identify Diptera toth e family-level (key:Unwin , 1981)becaus e Diptera occurred in largenumber s and involved largeproportion s of allthre e detritivores, herbivores andpredator s (Frouz, 1999;Webe r etal., 1997) .

2.2.2.Crop andE.I.-vegetation In each 1 m2plot"inwhea t fields, visual estimations weremad e of cropcove r {i.e.% cove ro f soil surface by vertical projection) and under story vegetation cover. The under story included arableweed s and/or undersown grass (Loliumperenne) and/or clovers (Trifoliumpratense or T. repens).Th e species ofth eunde r storywer e listed. In each 1 m2plot inth e E.I., the vegetation cover was estimated visually and species were listed. Plant species number of under story and E.I. vegetation were calculated for eachplot .

2.2.3.Farm management and lay out A 1:5000ma pwa s made of each farm, depicting the different crops andth e ecological infrastructure (E.I.)- Onthes emap s the area of semi-natural and crop habitats were measured. The 'E.I.-area' was calculated asth epercentag e of semi-natural habitat of the farm area (without the farm yard).Far m rotation intensity was defined asth eproportio n of gramineous crops (cereals and leypasture ) ofth etota l crop area. The spatial arrangement ofth e E.I.o r 'E.I. density', was expressed by thepercentag e of the farm which lied within 100m distance ofth eE.I . In case the canalban k vegetation was cutbot h after September 15th aswel l as before May 15th, adistanc e of 70m wa s considered. Theinfluenc e offield margin s onth e abundance of largecarabi d beetles and spiders probably does not exceed 70-100m (Booij , unpublished data). Data onth eyea r of conversion ofth ewhol e farm to organichusbandr y practices inth ewhea t crop in 1998,an dmowin g ofcana l banks and grass strips were obtained using a questionnaire. Themowin g management isrepresente d by themont h ofth e first cut, because this cut largely determines thephenologica l stage ofth e vegetation inJun e and July.

2.3.Statistical analyses All measurements were tested for normality and '°log-transformed. All subsequent analyses were conducted using Statistical Analysis System (SAS Institute, Cary,Nort h Carolina).

2.3.1.Discriminant analyses Individual farms were classified into one ofthre egroup s onth ebasi s of corresponding characteristics (Table 1).Fo r analysis ofdat a from crop area samples (234 m2),th e classifications werebase d on relevant farm structure characteristics (duration, E.I.-area and rotation intensity) or on cropperformanc e characteristics (crop cover, under story cover and under story species number).Fo r analysis of data from E.I. samples (117m ),th e classifications werebase d onrelevan t farm structure characteristics (E.I.-area and duration) or on vegetation performance characteristics (month of first cut, E.I.-vegetation cover orplan t species number perm 2). Farms were classified according to two splits:th efirst spli t based on one ofth e above mentioned characteristics separated acategor y of 3-7 farms, andth e second split based on another ofthes e characteristics divided the remaining 6-10 farms in asecon d and athir d category. Theminimu m category size was 3 farms. For all obtained classifications canonical discriminant analyses wereperforme d onth enumber s of arthropods in each functional group. 11 Observations inth ewhea t and inth e E.I.wer e analysed separately for crop area and E.I. classifications, respectively. Classifications with the highest Wilks'Lambd a F-valuewer e selected; this selection yielded the four best classifications. Stepwise discriminant analyses wereperforme d onth enumber s of arthropods in each functional group to determine themos t important groups that could classify the farms intoth e threecategories . Arthropod observations inth e crop area and inth eE.I . were again analysed separately. Canonical discriminant analysis was used to determine themagnitud e and direction ofth e association ofindividua l variables with indicator variables. Standardised canonical coefficients larger than 0.3 divided by the square root ofth e eigenvalue ofth e canonical function (Afifi &Clark , 1984)wer e considered large enough to contribute significantly to the classification.

2.3.2.Taxa differentiatingamong farm categories anddistance from E.I. Differences in species composition were expressed by the taxa that had significantly highero r lower numbers in one category ascompare d to the other categories. Differences in abundance ofvariou stax abetwee n thethre e categories inth e four selected classifications were testedb y ANOVA (P<0.05),includin g aDuncan' s multiple rangetest . In this test non-transformed data were used, and taxa that occurred with less than 20 individuals inth e whole data-set were omitted. Differences in abundance oftax a among the sampling distances from the E.I. of 0 m (inth eE.I.) , 10m (inwheat ) and 30m (i nwheat ) were also tested with ANOVA (P<0.05) including Duncan's multiple range test.

2.3.3.Differences in species diversity Differences in species diversity were also analysed. Taxatha t occurred with less than 20 individuals inth ewhol e data set were omitted. Parasitoids and r-herbivores wereno t considered becausebot h groups included onetaxon . Observations inth e crop area (234m 2; pooled samples of 10an d 30m from the E.I.) and inth e E.I. (117m 2) were analysed separately. Calculations weremad e ofth enumbe r oftax ape r arthropod functional group per 2 m (species density: S)an d theMargale f diversity index per m (Dm= (S-l)/lnN: S= numbe r of species, andN = numbe r of individuals) (Magurran, 1988).Difference s in San d Dm among thethre e categories in each ofth e four selected classifications were tested by ANOVA (P<0.01),includin g Duncan's multiple rangetest . Differences in San d Dm among the sampling distances (0, 10,an d 30m from the E.I.) were also tested by ANOVA (P<0.01), including Duncan's multiple rangetest , using thewhol e data-set (351m 2).

2.3.4.Gradient analysis Thepoole d data-set ofwhea t and E.I. samples (351m 2)wa s used to analyse the distribution of arthropod functional groups alongth e gradient from E.I. to crop centre, represented by three sampling distances 0m (E.I.), 10m (wheat) en 30m (wheat), for each farm category. Linear regressions (GLM;P<0. 1 )wer eperforme d for the four selected classifications using 10log-transformed data. Effects of distance, category, and distance by category interaction were tested. Crop area observations (234m 2) and E.I. observations (117m 2)wer e also tested separately for differences between categories inth e four selected classifications (GLM; P<0.1), including Duncan's multiple range test.

12 3. Results

3.1. Observednumbers ofepigeic arthropodfunctional groups

Atota l of 26,722 organisms were caught in 351m 2 sampling area including 12,346 aphids. On average, 63%an d 13% of vegetation dwelling arthropod numbers were aphids inth ecro p area and inth e ecological infrastructure, respectively. Average arthropod density inth ecro p area, excluding aphids,wa s 28organisms/m 2 ofwhic h themos t abundant taxa (>2%)were : herbivores: leafbeetle s (Chrysomelidae)(6%) andhopper s (Cicadellidae)(5%); detritivores: chironomid midges (12%)an d sciarid midges (4%); predators: spiders(10%) ,empidi d flies (9%), anddolichopi d flies (3%);an dparasitoid s (32%) (Table 2a).Th e average density in the E.I., excluding aphids,wa s 67organisms/ m with themos t abundant taxa (>2%)being : herbivores: hoppers (36%), chloropid flies (7%),an dbug s (3%);detritivores :woo d lice (7%) and chironimid midges (5%);predators : spiders (9%) and ants (3%);an dparasitoid s (7%) (Table2b) .

Table2a Thetotal number/18 m~perform of arthropod taxa in the wheat crop captured by vacuum sampling.

Taxa Funct. Farm group T F K Z R L H B O S J E M Isopoda detr 0 0 0 0 0 0 0 1 0 0 0 0 1 Chironomidae dea­ 22 8 34 26 67 92 11 60 20 380 10 52 11 Mycetophilidae den- 2 0 1 3 9 2 1 20 7 1 2 7 10 Sciaridae detr 4 12 10 10 18 6 22 65 61 4 23 12 2 Lonchopteridae dea­ 1 0 43 2 3 0 1 5 0 3 0 14 0 Phoridae den 0 1 0 2 1 0 2 3 0 0 0 1 2 Borboridae dea­ 1 1 1 3 0 0 2 7 5 1 3 20 10 Drosophilidae den- 30 1 20 4 20 1 3 9 3 0 1 18 0 Sepsidae detr 0 0 1 1 0 0 0 0 1 1 0 8 23 Cicadelidae K-herb 0 2 18 5 8 5 30 41 11 8 59 30 80 Heteroptera, herbivores K-herb 2 0 1 0 1 1 0 1 0 0 0 0 0 Chrysomelidae K-herb 26 6 30 13 29 58 13 62 51 65 8 32 8 Curculionidae K-herb 0 0 0 0 0 1 0 1 0 0 0 51 4 Tipulidae K-herb 0 0 6 1 1 0 0 5 0 0 0 0 1 Chloropidae K-herb 5 1 2 2 23 4 2 14 2 3 0 8 9 Opomyzidae K-herb 1 0 2 0 0 1 1 0 1 0 0 1 1 Ephidridae K-herb 6 1 6 6 23 7 1 0 0 0 2 1 0 Symphyta K-herb 1 0 0 1 4 2 1 0 3 0 0 0 2 Heteroptera, predator pred 1 0 0 0 1 1 1 1 2 0 0 0 2 Coccinellidae pred 14 1 7 0 30 8 3 3 11 3 10 1 5 Cantharidae pred 12 18 12 18 12 10 14 4 5 0 3 4 2 Chrysopidae pred 2 0 2 0 2 0 0 4 1 6 7 11 6 Syrphidae pred 20 0 28 3 36 26 1 1 4 3 7 18 8 Empididae pred 23 7 25 22 27 12 13 68 44 29 17 173 139 Dolochopodidae pred 1 0 6 7 1 0 3 12 14 5 0 103 30 Aranae pred 7 10 62 21 72 27 15 189 20 57 42 81 61 Carabidae pred 0 0 1 0 0 0 1 16 1 3 0 5 10 Staphylinidae pred 1 2 2 0 3 5 2 6 5 4 3 3 2 Formicida e pred 0 0 0 2 0 0 0 0 0 1 0 2 0 Parasitica par 117 68 140 183 311 336 28 118 251 238 67 143 58 Aphididae r-herb 4693 270 1284 1437 746 701 82 397 290 188 236 367 465

13 Table2b Totalnumber/9 m~perform of arthropod taxa in the ecological infrastructure captured by vacuum sampling.

Taxa Funct. Farm group T F K Z R L H B O S J E M Isopoda detr 28 61 12 13 0 364 1 1 0 13 12 26 0 Chironomidae detr 220 0 2 18 2 32 0 3 3 126 0 14 1 Mycetophilidae detr 1 0 0 1 0 1 0 1 1 2 0 0 0 Sciaridae dea­ 16 0 2 13 5 5 2 18 47 4 1 5 2 Lonchopteridae den- 16 0 15 1 2 9 1 1 0 6 2 17 0 Phoridae detr 4 0 0 12 7 0 3 7 6 1 5 1 0 Borboridae detr 48 3 3 5 16 16 4 4 2 10 1 26 2 Drosophilidae detr 34 1 6 12 7 8 6 6 10 5 0 1 0 Sepsidae detr 17 1 2 3 3 2 1 1 3 2 12 4 6 Cicadelidae K-herb 388 58 8 211 772 190 105 311 496 76 94 66 32 Heteroptera K-herb 2 2 0 54 1 142 1 0 0 1 7 36 1 Chrysomelidae K-herb 1 0 3 0 0 1 0 1 0 0 0 0 0 Curculionidae K-herb 1 0 0 0 0 3 0 0 0 1 2 0 0 Tipulidae K-herb 6 4 1 3 0 4 0 1 1 1 0 0 0 Chloropidae K-herb 16 3 2 29 36 12 125 55 104 27 72 35 14 Opomyzidae K-herb 25 0 2 11 9 5 0 3 1 5 1 1 0 Ephidridae K-herb 7 0 2 14 7 29 0 1 2 3 1 1 1 Symphyta K-herb 4 1 0 13 1 6 1 6 1 1 0 2 0 Heteroptera pred 9 0 2 22 3 91 0 13 0 11 9 5 6 Coccinellidae pred 4 0 2 0 0 0 0 0 1 0 5 1 6 Cantharidae pred 17 4 4 26 0 10 0 0 1 0 0 1 0 Chrysopidae pred 4 1 0 0 0 1 0 0 1 1 0 5 0 Syrphidae pred 2 0 0 0 0 0 1 0 0 1 0 1 1 Empididae pred 41 4 21 1 4 9 1 1 1 13 7 21 12 Dolochopodidae pred 1 0 1 3 6 4 0 11 1 10 2 2 3 Aranae pred 76 34 31 86 83 116 8 32 31 82 57 43 15 Carabidae pred 19 6 11 6 4 14 3 1 1 16 4 2 0 Staphylinidae pred 7 2 6 2 1 29 5 0 0 3 6 1 1 Formicidae pred 49 49 65 7 19 36 6 0 5 6 5 2 1 Parasitica par 88 14 27 95 49 47 8 48 59 68 30 35 6 Aphididae r-herb 488 11 45 106 127 59 4 35 98 10 82 35 90

3.2.Crop areaarthropod functional groupsand species diversity

3.2.1.Arthropod functional groupsaffecting distinctions amongfarm structure categories Theclassificatio n 'dur/int'base d onorgani c farming duration (first split) androtatio n intensity (second split) gaveth ebes t separation ofth e observations (canonical discriminant analysis, Wilks'Lambda , F-value=18.7, PO.0001; Table 3).Categor y 1include d the recently converted organic farms; category 2 included the long duration (>6years ) organic farms with an intensive crop rotation (<33% gramineous crops);categor y 3include d long duration organic farms with an extensive crop rotation. Classifications based on difference in percentage E.I. ('area') showed less distinction than classifications based on duration and rotation intensity (Table3) .

The separation into duration-intensity categories,base d on arthropod functional group numbers inwheat , is illustrated in aplo t ofcanonica l variable two versus canonical variable one (Fig. 2).Al l arthropod functional groups were selected as important factors by stepwise discriminant analysis for discrimination among the three categories (Table 4).Th efirst 14 canonical function containing all arthropod functional groups, separated the organic extensive crop rotation category from both other categories (Fig.2) .r-Herbivores , predators and K- herbivores werepositivel y associated with recent conversion or organic intensive crop rotation, whereas detritivores andparasitoid s werepositivel y associated with organic extensive crop rotation (Table 4). The second canonical function, determined mainlyb y predators andr-herbivore s separated longduratio n intensive farms from recently converted farms; predators were associated with recently converted farms and r-herbivores with intensive farms.

K-herbivores andpredator s were >25%mor e abundant on average, inrecentl y converted farms (category 1)tha n inth e othertw o categories. Detritivores andparasitoid s were respectively >40%an d >10%mor e abundant in organic extensive farms (category 3)tha ni n the other the two categories.r-Herbivore s were most abundant in intensive organic farms (category 2) and had intermediate numbers inrecentl y converted farms ascompare d to organic extensive farms (Table4) .

Table3 Examples of classifications of the crop area and the ecological infrastructure, including the the Wilks'Lambda F-value and thesquared canonical correlation obtained by canonical discriminant analysis. Abbreviations of category characteristics are explained in Table I.

Crop area Category 1 Category 2 Category 3 F-value Sq. Can. Farm structure category dur<6 dur>6 & int<33 dur>6 & int>33 18.7 0.37 categories farms HZBREM TKLJ FOS category dur<6 dur>6&area<2.1 dur>6& area>2.1 8.66 0.23 farms HZBREM FKO TLSJ category int>33 int<33 & dur>6 int<33 & dur<6 7.54 0.26 farms FOSEM TKU HZBR

Crop performance category ccov>85 ccov<85 & ucov>7.6 ccov<85 & ucov<7.6 38.32 0.56 categories farms BME TKRL FOHZSJ category ccov>85 ccov<85 & usp>3.5 ccov<85 & usp<3.5 25.1 0.54 farms BME ORJ TFKHZLS category usp>3.5 usp<3.5 & ucov>7.6 usp<3.5 & ucov<7.6 14.09 0.34 farms OREJ TKL FHZBMS

Ecological Infrastructure Category 1 Category 2 Category 3 Farm structure category area>2.1 area<2.1 & dur>6 area<2.1 & dur<6 7.84 0.39 categories farms TLSEJ FKO HZBRM category dur<6 dur>6 & area<2.1 dur>6 & area>2.1 6.53 0.34 farms HBZRME FKO TLSJ

E.I. vegetation category mow>7 mow<7 & eicov>80 mow<7 & eicov<80 15.94 0.48 performance farms TLSE HZBOJ FKRM categories category mow>7 mow<7 & eisp>4,3 mow<7 & eisp<4,3 10.53 0.51 farms TLSE BOMJ FKHZR category Eicov<80 eicov>80 & eisp<4.3 eicov>80 & eisp>4.3 6.8 0.37 farms FKRM THZ BOLSEJ

15 3 H

2

1

0 • no -1

-2 H

-3 •1 0 can1

Fig. 2. Plot of thesecond and first canonicalfunctions discriminating amongfarm categories of the classification 'int/area'.Squares representfarms with >-40%gramineous crops; circles representfarms with < 40%gramineous crops and >2.2%E.I. area; triangles representfarms with >40%gramineous crops and <2.2%E.I.

-1 0 can1

Fig. 3.Plot of thesecond and first canonicalfunctions discriminating amongfarm categories of the classification 'int/area'. Circles representfarms with dense crops (>85%); triangles representfarms with sparse crops (<85%) including an understory >7.6%; squares representfarms with asparse crop including a poor under story (<7.6%).

16 Table4 Arthropod variables that contributed significantly to the classification of the crop area observations in the three farm categories bystepwise and canonical discriminant analyses, standardised canonical coefficients, and mean valuesper crop area category. Classification in 'dur/int'-categories isbased onfarm structure characteristics and the classification 'ccov/ucov'-categories isbased on cropperformance characteristics. Theabbreviations of the category characteristics are explained in Table 1. Comparison of arthropod mean numbers in the categories according toDuncan's multiple range test.

Canlb Mean values and multiple range comparison Farm structure (dur/int) Standardised Category 1 Category 2 Category 3 coefficient dur<6 dur>6 & dur>6& int<33 int>33 Variablesa mean mean mean r-Herbivores 0.95 c 32 b 96 a 14 c (PO.001) Predators 0.46 13 a 6 b 5 b (PO.001) Detritivores -0.62 5 ab 4 b 10 a (P<0.07) K-herbivores 0.67 5 a 4 a 3 b (PO.001) Parasitoids -0.49 8 b 9 ab 10 a (P<0.05)

Canld Mean values and multiple range comparison Crop performance Standardised Category 1 Category 2 Category 3 (ccov/ucov) coefficient ccov>85 ccov<85 & ccov<85 & ucov>7.6 ucov<7.6 Variables * mean mean mean Predators 1.15 e 20 a 7 b 5 c (PO.001) r-Herbivores 0.54 23 b 103 a 23 c (PO.001) Detritivores -0.15 7 a 5 a 6 b (PO.05) K-herbivores 0.52 7 a 4 b 3 c (PO.001) Parasitoids -0.44 6 b 13 a 8 b (PO.001)

*Variable s are listed in order of selection by stepwise discriminant analysis. Canonical function 1;responsibl e for 71% of the variation. cStandardize d coefficients >0.39 were considered large enough for interpretation. Canonical function 1;responsibl e for 73%o f the variation. eStandardize d coefficients >0.26 were considered large enough for interpretation.

3.2.2. Differencesin species diversity amongfarm structure categories Recently converted and intensive longduratio n farms had similar speciesrichnes s inal l functional groups considered (detritivores, K-herbivores, andpredators) . The species diversities wereonl y similar for K-herbivores and totaltax a (Table 5).Detritivore swer emos t diverse inrecentl y converted farms. Therewer etw o differentiating taxa:Phoridae (detritivorous flies) preferred recently converted farms andherbivorou s bugs (Heteroptera) preferred long duration intensive farms.

3.2.3. Arthropodfunctional groupsaffecting distinctions amongcrop performance categories It was envisaged that cropperformanc e factors could offset farm structure factors in determining arthropod functional group assemblages. Thisprove d tob e correct: the classification 'ccov/ucov'base d on crop cover (first split) and weed cover (second split) provided thebes t separation ofobservation s (Wilks' Lambda, F-value=38.3, P<0.0001; Table 3).Cro p performance category 1 included dense crop stands (cover >85%);cro p performance category 2 included relatively sparse crop stands (cover <85%),accompanie d by

17 anunde r story vegetation (>7.6%cover) ;cro p performance category 3include d relatively sparse crop stands which had apoo r under story vegetation.

The separation of crop-weed cover categories,base d on arthropod functional group numbers inwheat , isillustrate d ina plo t of canonical variable two versus canonical variable one (Fig. 3). All arthropod functional groupswer e selected as important factors by stepwise discriminant analysis for discrimination amongth e threecategorie s (Table 4). Extremes were dense crops and sparse crops with poor under story (Fig. 3).Significan t contributions toth e first canonical function were thepositiv e associations ofpredators , r- and K-herbivores with dense crops, and thenegativ e association ofparasitoid s with dense crops (Table 4).Th e relation between under story vegetation and arthropod groups was not aspronounced : the category of sparse cropswit h under story was separated from both other categories by the second canonical function, determined significantly by predators (negative association) and r- herbivores (positive association).

K-herbivores and predators indens e crops (category 1)wer e respectively >65%an d>40 % more abundant on average,tha n inth e other categories.r-Herbivore s andparasitoid s in sparse cropswit hunde r story wererespectivel y >75% and>40 %mor e abundant than in thetw o other categories (Table4) .

3.2.4. Differencesin speciesdiversity among crop performance categories was generally highest when crop cover wasmor etha n 85%. Diversity of senescense-feeding herbivores,predator s and detritivores was lowest in the category of sparse crops withpoo r under story ascompare d toth e other categories. Dense crops had thehighes t diversity of detritivores.Dens ecrop s and sparse cropswit h under story had asimila r diversity ofK-herbivores . Sparse cropswit h under story had the highest diversity ofpredator s (Table 5).

Thirteen taxaha d significantly higherdensitie s in one ortw o ofth e crop performance categories (ANOVA Duncan, P<0.05): indens e crops thehighes t densities wereobserve d for woodlice (Isopoda, detritivore), Phoridae, Borboridae, Sepsidae (all detritivorous diptera), Cicadellidae (herbivore), Aranae, Carabidae (polyphagous predators), and Dolichopodidae, Empididae (predacious flies); in sparse crops with under story the highest densities occurred among Drosophilidae (detritivorous fly), herbivorous bugs (Heteroptera), Ephydridae (herbivorous fly), andhooverfl y larvae(Syrphidae ; predacious fly).

3.3.Ecological Infrastructure arthropodfunctional groupsand species diversity

3.3.1.Arthropod functional groupsaffecting distinctions amongfarm structure categories Oftw o considered possibilities the classification area/dur,base d on E.I. area as first split and farm organic duration as second split gave thebes t separation of observations (canonical discriminant analysis, Wilks' Lambda, F-value=7.84, P<0.0001; Table 3).Far m structure category 1 included farms with an enlarged E.I. (>2.1% ofth e farm arable area) and corresponded to 'improved E.I.'; the othertw o categories are farms with a small E.I.whic h could be long duration organic farms that did not enlargeth e E.I. area (category 2)o r recently converted organic farms (category3) .

The separation into area-duration categories,base d on arthropod functional group numbers in theE.I. ,i sillustrate d in aplo t ofcanonica l variable twoversu scanonica l variable one(Fig . 4). Detritivores and K-herbivores were selected asimportan t factors by stepwise discriminant Table5 Comparison between the threecategories inall four selected classifications, with regard to their species richness (S) and Margate/diversity index (DJ (GLM;significance level *=P<0.1; **P<0.01; *** P<0.001); detr = detritivores; K-herb= senescensefeeding folivores; pred =predators; total =detritivores, K-herbivores andpredators. Abbreviations of category characteristics are explained in Table I.

Crop area

S Dm S Dm S Dm S Dm Categories detr detr K-herb K-herb pred pred Total Total Farm structure dur<6 2.3 a 1.0 a 2.1 a 0.8 a 3.4 a 1.2 b 7.8 a 2.3 a categories dur>6 & int<33 1.9 a 0.7 b 1.8 a 0.8 a 3.1 a 1.3 a 6.8 b 2.3 a dur>6 & int>33 1.5 b 0.6 b 1.1 b 0.4 b 2.4 b 1.1 b 5.0 c 1.7 b Significance *** *** *** *** *** * *** **# Crop ccov>85 2.7 a 1.0 a 2.4 a 0.8 a 7.0 a 1.2 b 9.1 a 2.4 a performance ccov<85 &ucov>7. 6 2.1 b 0.7 b 2.0 b 0.8 a 3.4 b 1.4 a 7.5 b 2.4 a ccov<85 &uc ov<7. 6 1.5 c 0.7 b 1.2 c 0.5 b 2.4 c 1.1 b 5.0 c 1.8 b Significance *** ** #** *** *** *** *** ***

Ecological Infrastructure

S Dm S Dm S Dm S Dm Categories detr detr K-herb K-herb pred pred Total Total Farm structure area>2.1 4.1 a 1.2 3.8 a 0.9 a 4.8 a 1.4 a 12.8 a 2.9 a area<2.1 & dur>6 2.5 b 1.1 2.0 c 0.7 b 3.2 b 1.0 b 7.7 b 2.0 b area<2.1 & dur<6 2.4 b 1.3 3.0 b 0.6 ab 2.8 b 1.0 b 8.2 b 2.0 b Significance *** NS *** * *** ** *** *** E.I. vegetation mow>7 4.7 a 1.2 4.0 a 0.9 a 5.1 a 1.4 a 13.8 a 3.0 a performance mow<7 & eicov>80 2.7 b 1.3 3.2 b 0.6 b 2.6 c 0.9 c 8.5 b 2.0 b mow<7 & eicov<80 2.0 b 1.1 2.0 c 0.7 b 3.6 b 1.2 b 7.2 b 2.0 b Significance *** NS *** ** *** *** *** ***

analysis for discrimination among the three categories (Table 6).Th e largest differences in arthropod functional group composition werebetwee n the enlarged E.I. category and recently converted farm category (Fig. 4).I nth e first canonical function detritivores were positively associated and K-herbivores negatively associated with enlarged E.I. (Table 6).Althoug h the predator number was significantly higher inth e enlarged E.I. category ascompare d to the other categories (Table 6),i tha d little canonical power inth e first canonical function because of ahig h correlation (Pearson correlation coefficient =0.79 ;P<0.01 )betwee n detritivore and predator abundance. The second canonical function, determined by K-herbivores (positive association), separatedbot hextrem e categories from thethir d category oflon g organic duration farms with small E.I.

Detritivores andpredator s were >70%an d>40 %mor e abundant on average in the enlarged E.I.categor y (1) as compared to the small E.I. categories.K-herbivore s were >20%mor e abundant on average in recently converted farms (category 3)tha n inth e other categories (Table6) .

3.3.2. Differencesin species diversity amongfarm structure categories Species richness of K-herbivores and detritivores and diversity ofpredator s and total taxa were significantly higher in the enlarged E.I.categor y than in other categories (Table 5).Th e only differentiating taxawer e weevils (Curculionidae; K-herbivore) which preferred enlarged E.I.

19 3.3.3. Arthropodfunctional groupsaffecting distinctions amongE.I. vegetationperformance categories The classification 'mow/eicov'base d on first mowing date (first split) and E.I. vegetation cover (second split) gaveth ebes t separation ofobservation s (canonical discriminant analysis, Wilks'Lambda , F-value=15.94 , PO.0001; Table 3).Vegetatio n performance category1 included E.I.whic h was cut late (after July 1st) and included the 'improved E.I.'o fth ethre e prototype farms andth e organic experimental farm (E).Ecologica l infrastructure which was cutbefor e July could have ahig h vegetation cover (>80%)(category 2) or an 'open vegetation' i.e.a lo w cover (<80%) (category 3).W edefine d category 2 as'grass y E.I.'becaus e early mowing in May, combined with ahighe r mowing frequency peryea r and/or careful mowing resulted ina stable,dens e sward dominated byperennia l grass species.

Table6 Arthropod variables that contributed significantly to the classification of the E.I. observations in the threefarm categories bystepwise and canonical discriminant analyses, standardised canonical coefficients, and mean valuesper crop area category. Classification in 'area/dur'-categories isbased onfarm structure characteristics and the classification 'mow/eicov'-categories isbased on vegetationperformance characteristics. The abbreviations of the category characteristics are explained in Table I. Comparison of arthropod mean numbers in the categories according toDuncan's multiple rangetest.

Canlb Mean values and multiple range comparison Farm structure (area/dur) Standardised Category 1 Category 2 Category 3 coefficient area>2.1 rea<2.1 & area<2.l & dur>6 dur<6 Variables a mean mean mean Detritivores 1.16c 26 a 7 b 4 c (PO.001) K-herbivores -0.49 32 a 26 b 41 a (PO.01) Predators -0.18 21 a 11 b 9 b (PO.001) Parasitoids 0.33 6 a 4 b 5 b (PO.05) r-Herbivores 0.03 13 a 6 a 8 a NS

Canld Mean values and multiple range comparison E.I. vegetation performance Standardised Category 1 Category 2 Category 3 (mow/eicov) coefficient mow>7 mow<7 & mow<7& eicov>80 eicov<80 Variables * mean mean mean Detritivores 1.34e 31 a 6 b 5 b (PO.001) K-herbivores -0.26 35 a 40 a 27 b (PO.001) Predators 0.16 23 a 9 c 12 b (PO.001) Parasitoids 0.11 7 a 5 a 3 b (PO.001) r-Herbivores -0.22 14 a 7 a 8 a NS

"Variables are listed in order of selection by stepwise discriminant analysis. The italicised variables were not selected by stepwise discriminant analysis. bCanonica l function 1:responsibl e for 85%o f the variation. cStandardise d coefficients >0.37 were considered large enough for interpretation. Canonical function 1:responsibl e for 63%o f the variation. "Standardise d coefficients >0.31 were considered large enough for interpretation.

20 3

2

l H A 4A

% ,-,•*§* • AA% H&*A DA A D a ^» A A " ** -,• o *,°A &* A A /A 0 AflD • 1 A D o 2° 0 -2 oo

-3

-4 -2 -1 can1

Fig. 4.Plot of the second and first canonicalfunctions discriminating among E.I.categories of the classification 'area/dur'. Triangles represent enlarged E.I. (>2.1%); circles represent small E.I. onfarms with a long organic duration; squares represent small E.I.on farms with a short organic duration.

3 1 0 o

2 - 0o° °o 8 0 0 o 1 - A° O°T> A o n o o 0$ tP<* 0 - o

1 0A & A $> O A A D D A AA A A A A -2 - A A A A A A A 1 ! -3 - i • -r — ^ 1 -1 0 can1

Fig. 5.Plot of thesecond and first canonicalfunctions discriminating among E.I. categories of the classification 'eispn/eicov'.Squares represent late mown E.I.; circles represent early mown E.I. with dense vegetation; triangles represent early mown E.I.with sparse vegetation.

21 The separation into 'mow-eicov'categories ,base d onarthropo d functional group numbers in theE.I. , isillustrate d in aplo t ofcanonica l variable twoversu s canonical variable one(Fig . 5).Al l arthropod functional groups were selected as important factors by stepwise discriminant analysis for discrimination among thethre e categories (Table 6).Th efirst canonical function containing only detritivores, separated latemowin g category from grassy E.I. and openvegetatio n categories (Fig. 5).Detritivore s are associated positively with late mowing (Table 6).Althoug h thepredato rnumbe rwa ssignificantl y higheri nth e latemow n E.I. category ascompare d to theothe r categories (Table 6),i tha d little canonical power inth e firstcanonica l function because of ahig h correlation between detritivore and predator abundance inth e E.I. The second canonical function, determined mainly by K-herbivores and parasitoids (bothpositivel y associated) andpredator s (negatively associated) showed a separation between grassy E.I. (category 2) and open vegetation (category 3).Al l functional groups,excep t theK-herbivore s were on averagemor e abundant inth e late mown E.I. category (1) as compared tobot h other categories: detritivores and predators were respectively >80%an d >40%mor e abundant. K-herbivores were >10%mor e abundant on average ingrass y E.I. (category 2)tha n inbot hth e other categories (Table6) .

3.3.4. Differencesin species diversity among E.I. vegetationperformance categories Speciesrichness an d diversity of allarthropo d functional groups, except diversity ofdetri ­ tivores,wer e significantly higher in latemow n E.I.tha n in other categories. The category of grassy E.I.ha d the lowest diversity ofpredator s ascompare d toth e other categories (Table5) . Ofth etota l of eight differentiating species, sixpreferre d latemow n E.I.: three detritivorous taxa (Chironomidae, Lonchopteridae, Borboridae) andthre epredaciou s taxa (Chrysopidae, Carabidae, and Empididae). The high abundance of chloropid flies (herbivorous) andphori d flies(detritivorous ) wastypica l for the category ofgrass y E.I. vegetation.

3.4. Arthropodfunctional groupsalong the gradient from E.I.to crop

3.4.1.Functional group composition alongE.I. - cropgradients Itwa s expected that the relations between abundance ofvariou s functional groups and distance from the E.I. would be similar for different farm categories. However, there were significant interactions between categories in all four selected classifications and distance from the E.I., with regard to abundance ofdetritivore s andpredators . Therewer e also significant interactions between categories and distance with regard to K-herbivores for some classifications but not for others. The interactions were strongest for predator distributions, particularly for crop area classifications but also for 'mow/eicov' classifications. These interactions arebes t illustrated for the E.I. classification 'mow/eicov' (Fig. 6).Althoug h in most farms abundance ofpredator s and detritivores were lowest at 30m distanc e from the E.I., farms with an early mowing date andrelativel y littleplan t cover in the E.I. (category 3) had morepredator s and detritivores at 30m tha n at 10m distanc efrom th e E.I. (Fig. 6aan d c). Thus,hig h densities inth eE.I . dono t always gohand-in-han d with high densities inth e adjacent crop.

Despite the significant interactions,predators , K-herbivores, and detritivores were generally most abundant in the E.I. andwer e lower inth ewhea t field (Fig. 6 a,b,c). Although abundancewer e generally decreasing or similar with increasing distance from the E.I.,i n category 3the y were sometimes higher at 30m tha n at 10m insid e thewhea t field (Fig.6 a,b,c). Onth e other hand, abundance ofparasitoid s and r-herbivores were generally lowest in theE.I . andhighes t at3 0m insid e thewhea tfield (Fig . 6d,e) ,wit hth eexceptio n ofth e parasitoids in category 2 (grassy E.I.) that werehighes t at 10m insid e thewhea t field and lower inth e E.I. and at 30m insid e thewhea tfield (Fig .6d) . 22 3.4.2. Differencesin diversity between E.I. andcrop area Speciesrichnes s ofdetritivores ,K-herbivore s andpredator swa s significantly higher inth e E.I.,bu t diversity (Margalef) was only larger for detritivores. Ten of 30considere d taxa of predators,K-herbivore s and detritivores showed apreferenc e for E.I.: Phoridae, Borboridae, Cicadellidae, Chloropidae, Opomyzidae, Symphyta,predaciou s Heteroptera, Aranae, Carabidae, andFormicidae . Two taxa showed apreferenc e for the crop area: Chrysomelidae (herbivorous) and Syrphidae (predacious).Non e ofth econsidere d taxaha d a significantly different abundance between 10m and 30m sampling stations.

A: detritivores B: K-herbivores 1.5 2

1.5

1

0.5

0 10 m 30 m 0 m 10 m 30 m

C: predators D: parasitoids 1.5

0.5

0m 10m 30m

E: r-herbivores 2

1.5

£ 1 o

0.5

0 10m 30 m

Fig. 6.Abundance of vegetation dwelling arthropodfunctional groups along thegradient from E.I. (0 m) to the crop area (10and 30 m). E.I.vegetation performance categories of selected classification 'mow/eicov': category I (squares)= late mowing; category 2 (triangles)— early mowing, dense vegetation; category 3 (circles)= early mowing and sparse vegetation. Thelines between the averages of arthropodfunctional group numbers('"log- transformed) are drawn to illustrate differentpatterns among the three E.I.categories and do not represent estimatedpoints in between thesampling stations, (a)detritivores; (b)K-herbivores; (c)predators; (d) parasitoids; (e) r-herbivores.

23 4. Discussion

4.1. Observations comparedto the hypotheses

Itwa s expected that duration oforgani cmanagemen t and extensive crop rotation would relate positively to the abundance ofprimar y arthropod functional groups andtha t consequently, predator andparasitoi d numbers wouldb e supported. Itwa s also expected that improved E.I would promote the abundance ofK-herbivore s and predators inbot h E.I. and the adjacent crop area. Observations thatma y confirm ourhypothese s include therelativel y high detritivore andparasitoi d numbers on farms which had both a long organic duration aswel l as an extensive crop rotation, and increased densities ofpredator s in improved (i.e.lat emown ) E.I.However , several observations contradicted thehypothesis : dense crops on recently converted farms had thehighes t densities ofK-herbivore s andpredators ; several long duration farms had increased numbers ofr-herbivores ; K-herbivores andparasitoid s were equally abundant inimprove d E.I. and 'grassy E.I.'; increased densities ofpredator s and K-herbivores inth e crops areawere ,i n several cases,no trelate d to increased densities ofthes e groupsi n improved E.I.

4.2. Vegetation dwellingarthropods community inthe crop area

Observations suggest theoccurrenc e of four distinctive arthropod functional group compositions. Outlining these compositions may explainth eobserve d contradictions with the hypotheses and contribute to improved hypotheses. The four functional group compositions represent ashif t along the gradient of organicduratio n (involving e.g. carbon flow) and production intensity (involving e.g. nitrogen input).The ywil l be discussed in ascending order ofintensity : a) short duration with aspars e crop,b ) short duration with adens e crop,c )lon g duration with intensive croprotation , andd ) long duration with extensive crop rotation. a)Recentl y converted farms which did not have aluxuriou s crop growth andwhic hha d a poor under story (corresponding to farms Han dZ) . Crop areas werecharacterise d by relatively lowdensitie s off all functional groups. Theselo w densities areofte n encoutered in field centres of conventionally managed crop habitats that include almost no arableweeds ,particularl y when r-herbivores areno t considered (Kromp& Meindl, 1997).Establishmen t of several taxama yrequir e time (Hald &Reddersen , 1990). This 'unsaturated'arthropo d community may develop, during at least one crop rotation, towards acommunit y oftyp e d) or c),dependin g onth e farm's (rotation) intensity. However recently converted farms represent relatively unstable farming systems (oral communication G. Oomen) inwhic h stillunbalance d N-availability and/or apparent crop susceptibility may rapidly change acommunit y oftyp e a) into acommunit y oftyp eb) . b) Short organic duration with adens e crop represents farms inth econversio n process (corresponding to farms B and E),involvin g the adaptation by soil life to changed manure management, and creating asite-specifi c balance between N-application and crop requirements. In thisunstabl e situation luxurious growthma y occur andprovid e aric h food for an array ofK-herbivorou s taxa. In the dense canopy with damp microclimate, several detritivorous diptera taxa are enhanced by decaying leaves (Weber etah, 1997).Th e large K- herbivore populations inthes e cropsrelate d to highpredato r densities (Pearson correlation between predators andK-herbivore s incro p area: 0.81,P<0.001) . But also detritivorous flies mayhav e supported predators like, for example, empidid flies. All groups ofherbivores , detritivores andpredator s had high speciesrichness ,probabl y related to theric h food 24 resource. High predator densities mayhav e affected r-herbivores andparasitoids , which were less abundant than in typec) . c) Long duration farms with an intensive crop rotation (corresponding to farms T,L , Kan dJ ) may support the detrital subweb to amuc h lesser extent than long duration farms witha n extensive crop rotation; low densities ofdetritivore s were observed here. Therefore polyphagous predators mayb e less abundant and therefore have less effect onth e herbivore subweb arthropods. Also cropsma y bemor e susceptible for herbivory duet o thepresenc eo f moreN , andcrop s needing moretillag e inth e crop rotation. Density of (polyphagous) predators and resistant cropphysiolog y are complementary factors controlling r-herbivore outbreaks (Embden, 1988);system swit h 'moderate' levels ofbot h factors might alreadyb e vulnerable. The long duration farms had amor e diverse arthropod community ascompare d to type d),involvin g acoexistenc e of several predacious taxa and K-herbivores in lowdensities . Howeverth epredator s could apparently not inhibit outbreaks ofr-herbivores . The r-herbivore outbreaks mayhav e supported the observed highparasitoi d numbers. d)Lon gduratio n farms with an extensive crop rotation (corresponding to farms Oan d S)ha d increased numbers ofdetritivore s whichma yhav erelate d to anincrease d detrital subweb supported by cropresidue s and organicmanur e (e.g. Heimbach &Garbe , 1997;Webe r et al, 1997).Polyphagou s epigeicpredator s mightb e supportedb y anincrease d quantity of detritivorous prey (Wise etal., 1999)an dsubsequentl y depress theherbivor e subweb includingr - and K-herbivores andoligophagou spredators . Also lessnutritiv e and less susceptible crop tissuesma y explain the lownumber s ofherbivore s in alon g duration, with extensive crop rotation (Phelan, 1997;Theunissen , 1994;Va n Emden, 1988).Parasitoid s were themai n predacious functional group inth e canopy,possibl y due toreduce d interference by vegetation dwellingpredator s (cf.Tscharntke , 1997).Host s ofparasitoid smigh thav ebee nr - herbivores occurring inlo w densities and also detritivorous diptera (Potts &Vickermann , 1974).Parasitoi d abundance on all studied farms appeared tob e significantly correlated to detritivore abundance (Pearson correlation coefficient=0.52; P<0.06) and not to r-herbivore abundance,whic hma y expressbot hdirec t (e.g.food ) and indirect (e.g. predator release) factors affecting this functional group.

Three ofth e studied farms corresponded only slightly to thepresente d typology. Farm Fwa s most similar to type a).However , low arthropod densities onthi s farm were duet over y sparse crop standsnea r the crop edge andcro p centre,relate d to mechanical weed control. Farm Mwa s similar totyp e c),bu t onthi s integrated farm estimated crop N-requirements werepartiall y supplied by artificial fertilisers. Farm Rwa srecentl y converted and corresponded therefore totyp ed) ;however , duet owe t harvest conditions in thepre-crop ,th e investigated crop had aspars e stand including arelativel y dense and diversewee d vegetation whichrelate d to increased arthropod abundance.

Itmus tb e noted that the extensive organic arable farms inFlevolan d areprobabl y more intensive than the organic farms studied by Moreby &Sotherto n (1997) and Reddersen (1997).Relativel y extensive organic farms inFlevolan d have crop rotations including potatoes andvegetables , and are situated in intensively farmed open landscape. Thismigh t be illustrated by thetota l densities ofvegetatio n dwelling arthropods (collected bymean s of suction sampling); averagedensitie s found inorgani c cereal fields inFlevolan d were9-15 % of the average densities found by Moreby &Sotherto n (1997), Reddersen (1997) andPott s& Vickermann (1974).

25 4.3. Vegetation dwellingarthropod community inthe ecological infrastructure

Effects ofE.I .characteristic s andparticularl y of'improved E.I.1 will be explained by discussing theresponse s of thedifferen t functional groups.

4.3.1. K-herbivores K-herbivoreswer e equally abundant in improved (i.e.enlarge d and latemown ) andi n traditionally managed 'grassy' E.I. However species richness and diversity were greater in improved E.I. Weevils were theonl y differentiating K-herbivorous species for improved E.I., andma y have indicated clovers inth e grass strips of enlarged E.I. The arthropod community in 'grassy E.I.'wa spredominate d (55%o f individuals) byherbivores :hopper s and chloropid flies. This wasprobabl y duet o theprolonge d vegetative development of frequently mown vegetation. Accordingly, Haugthon etal. (1999 )observe d higher numbers of arthropods (mainly heteroptera) under afiel d marginmanagemen t of cutting twice ascompare d to cutting once.I n their experiment, one cutpe ryea rwa s found tob e devastatingbecaus ei t inhibited the life cycles ofhopper s andbugs .Althoug h this effect could not be assessed in late mown E.I.prio r to itsmowin g date,observation s inth e 'intermediate' management type seemed to confirm the conclusion ofHaugtho n etal. (1999) .Thi s management is often applied by farmers who are convinced that there isn o need for frequent mowing,bu t who are not ablet o doth e extra labour (e.g.finger-ba r mowing) required for improved management. Vegetation ofthi skin d ofboundarie s lost itsadaptatio n to frequent mowing and became colonisedb yrudera l dicots,e.g. dandelion; bothplan t growth andherbivor e community seemed tob emuc h more disturbed following afirst cu t after May 15thwit h a flail mower set at alo wcu t level than following traditional management.

4.3.2. Predators andparasitoids Predators werepromote d by 'improved E.I.1wit h regard to bothnumber s and diversity. Accordingly, Wakeham-Dawson etal. (1999 ) found atw otime s increase of spider numbers in longgraze d ascompare d to shortgraze d vegetation and argued thatvegetatio n structure is ake ypropert y (Curry, 1994).Th e observed predator numbers wereparticularl y correlated to detritivore numbers andno t toherbivor e numbers. Predators were less abundant in 'grassy E.I.'(wit h abundant herbivores) ascompare d to open swards.I n open swards, spiders,th e generally predominant species inth epredato r functional group,wer e accompanied by ahig h proportion of ants. High parasitoid numbers were also related to improved E.I., although thesenumber s were much smaller thannumber s inth ecrop ,suggestin gtha tth egrou p ismor e affected by crop area factors. However the increased proportion ofparasitoid s in the arthropod community of 'grassy E.I.' may suggest arelatio n to the abundance of homopteran hosts.

4.3.3. Detritivores andr-herbivores These groups were not included in the hypotheses because they were expected to have little importance.Howeve r the detritivorous component ofth e E.I.vegetatio n dwellers showed to be very significant, particularly in improved E.I. Possibly detritivorous diptera and woodlice were favoured by the increased amount of fibrous litter and stalks and ahighe r humidity in tall vegetation. The earlier mown vegetation of'grassy E.I.'ma yprovid e another litter quality that supports other, not sampled, taxa, e.g. fungivorous staphylinids (Smeding etal, 2001a). r-Herbivores had variable but generally low abundance inE.I . and no significant differences between management typeswer e found. Numbers of aphids in E.I. were highly correlated withnumber s in the crop (Pearson correlation coefficient=0.96, P<0.0001). However, distinctive cropcommunitie s with many or few aphids weremor e or less evenly distributed overth e distinctive E.I.managemen t categories.

26 4.4. Differences and interactions between crop area and ecological infrastructure

4.4.1. Differences According to the hypotheses predators, K-herbivores were more abundant and species rich in the E.I. as compared to the crop area. Detritivores were more abundant and both more species rich and diverse in the E.I. The vegetation dwelling arthropod community included several differentiating taxa compared to the crop area. These observations clearly indicate the increased resource of various plant tissues and structures in perennial vegetation as compared to annual crops (Curry, 1994). Parasitoids and r-herbivores were more abundant in the crop area. Only two taxa were significantly more abundant in the crop than in the E.I.; according to observations of Venhorst et al. (in prep.), aphidophagous syrphid abundance increased towards the crop centre, probably due to the increase of prey availability.

The ratio between E.I. and crop arthropod numbers (excluding aphids) on the 13 studied farms, was on average 2.8. However, including aphids, seven of 13 studied farms had more arthropod individuals per m in the crop area than in the E.I. Accordingly Remmelzwaal & Voslamber (1996) found in Zuid Flevoland, the highest arthropod numbers in the E.I. in spring and early summer, whereas numbers (of particularly crop herbivores) became higher in the crop area later in the season.

4.4.2. Interactions Visual assessment of E.I.-crop area gradients suggested that higher average densities in the E.I. of predator, K-herbivore and detritivore functional groups might relate to increased average densities in the adjacent crop area. However crop area conditions could have a much stronger effect on the distribution of these groups. In particular, crop canopy density supporting K-herbivores, and factors supporting detritivorous diptera, may offset potential E.I. influence on the crop community.

Difference between the 10 m and 30 m sampling station with regard to abundance, species richness and diversity of K-herbivores, detritivores and predators were generally insignificant, indicating that the influence of E.I. on the crop community in the summer, may stop before 10 m or might reach further than 30 m.

Aphids in the crop area may be transported and accumulated into emergent structures like ditch slopes and tall vegetation (Hradetzky & Kromp, 1997). This might explain the correlation of E.I. predator abundance to aphid abundance in the crop (Pearson correlation coefficients. 69, P<0.01). Intermediate numbers of parasitoids and aphids at the 10 m sampling station may indicate the effects on these groups of arthropods (e.g. predators) in the field margin. Detritivorous diptera taxa, emerged in the crop area, may also accumulate in the E.I.; however an increased diversity and occurrence of differentiating species in the E.I. suggested that this group is at least partly autochthonous in the field margin habitat.

4.5. Conclusions

Comparison of vegetation dwelling arthropod functional groups in the wheat crops on thirteen organic arable farms suggested that both organic duration and 'intensity' are determining factors. Intensity relates to crop rotation (i.e. percentage gramineous crops) but also to soil nutrient level. Observations provided no support for the general hypothesis that organic duration and extensive crop rotation would promote abundance of both primary arthropod functional groups and consequently, support predator and parasitoid numbers. 27 Results indicated ashif t or succession in arthropod functional group compositions from recently converted farms to long duration farms with extensive crop rotation. In recently converted farms pest susceptible crops,e.g. due to luxurious growth, may attract various herbivores andtherefor e support predator abundance. In longduratio n farms, arthropod communities areno tnumerou s but more diverse. However in thesecommunities , related also to crop condition, particularly r-herbivores (aphids) mayno t be controlled. However, in extensive crop rotation a'subsidise d increased detrital subweb',relate d to increased total organic matter input, may support polyphagous predators. Thesepredator s possibly depress theherbivor e subweb (i.e.herbivore s and specific predators) bybot h predation and . Inthi s community, aswel l asi nothe r observed communities, effects of 'predator interference' could play arole ,e.g. inreleasin g parasitoids from predation or competition by (oligophagous) vegetation dwelling predators.

According to ourhypothese s improved E.I. (i.e.larg e area and latemowing ) promoted the predacious functional groups. However herbivore abundance may also be high inth e traditional management. Thedetrita l functional group (particularly diptera) contributed strongly toth edistinctio n between improved E.I. and other E.I.type s and was therefore a moreimportan t component inth e E.I. arthropod community thanw e expected. E.I.ha d generally much higher densities ofnon-pes t arthropods than the crop area and these arthropods might therefore disperse into the crop area.Howeve r it was difficult to detect these E.I. effects because crop conditions werevariabl e between farms and crop area conditions had a strong effect on functional group abundance which mayhav e offset the E.I. effect.

4.6. Recommendations

Further research related tothi s study mayconcer n the impact of detrital subsidy (Wise etal., 1999;Poli s& Hurd , 1996)i n arable farming systems, andmigh t specifically address predation pressure in crops where epigeic predators find abundant detritivorous prey. Hereth e conceptual issue is indirect control of theherbivor e subweb, including pests,b y management ofth e detritivore subweb,i.e. the soilsystem (De Ruiter etal, 1993;Brussaard , 1998). Connected with the theme ofdetrita l subsidy, are studies onhighe r trophic levels inth e farm food web,whic h are supported by both detrital andherbivor e subwebs. Relations between both primary subwebsma y affect food availability for insectivorous birds that havevalu e for nature conservation.

Interference between predators might be another important research topic.Informatio n on and subsequent releases oftax a at lowertrophi c levels,migh t explain features that arepossibl y common in arthropod communities on farms (Tscharntke, 1997).

Appropriate study areas for research with regard to detrital subsidy,predato r interference and food web structure should include,preferably , acombinatio n ofrelativel y similar (commercial) farms, and additionally experimental fields for testing extreme management options. Practice farms are important study objects, because effects may only occur ina farming system, and such farm level experiments arever y expensive; another reason istha t experimental manipulation on farms must necessarily remain within theboundarie s of organic agricultural production systems which may safeguard the later applicability of research outcomes.

With regard to functional group assessment, improved definition of the assemblages is needed. Inparticular , herbivorous taxa might be aggregated in amor e sophisticated way. For

28 example Hald &Redderso n (1990) distinguished crop-related and non crop-related herbivores, and Tscharntke (1997)distinguishe d endo-an d ectophagous species.

Foroptimisatio n of E.I.wit h regard to arthropod functional groups, further comparative research including experiments ofmowin g regimes mightb erequire d (e.g. Boatman et ai, 1999; Remmelzwaal &Voslamber , 1996).Thes e studies should also take into account detritivorous and root herbivorous arthropods. Additional attention in arablefield margin s is required for abundant molluscs and earthworms andth e effects of ant nests onth e arthropod community.

With regard topractice ,result s suggest that improving the ecological infrastructure is worthwhile: inparticular , late mowing is important to enhance arthropod numbers and diversity. However, frequently mown margins with intact swards can also harbour high insect densities.I t might bebette r to avoid an'intermediate ' type ofmowin g management, involving few cutspe ryea r and alat e (June) first cutt o ground levelwit h aflai l mower. This management isprobabl y deleterious for arthropod abundance. With regard to the promotion ofbeneficial s inth ecrops ,mor eresearc h is stillneede d to find soundly based practical recommendations relating to crop and ecological infrastructure management.

Acknowledgements

The authors would like tothan k the farmers who allowed us access to their land, and Kees Booij and Clasien Lock(PRI-WUR )wh ogav evaluabl e suggestions andprovide d necessary equipments.Th ewor k at the Wageningen University is funded by theMinistr y ofAgriculture , Nature Management and , Department of Science and Knowledge Dissemination.

29 Chapter 3: Shifts in functional group composition of ground dwelling arthropods in relation toecologica l infrastructure and conversion to organic farming

F.W. Smedinga, C.A.M. Lockb &C.J.H . Booijb "Departmentof Plant Sciences, Biological Farming Systems group, Wageningen University, Wageningen, TheNetherlands ''Plant Research International, Wageningen, The Netherlands

Abstract Hypotheses on the relation of farm management and ecological infrastructure (E.I.) characteristics to the functional group composition of ground dwelling arthropods inth e crop and E.I. were tested on organic arable farms. The study area included 8 farms that varied with regard to the studied traits. Arthropods were collected by pitfall traps in wheat crops and adjacent canal banks. Observations in the crop area suggested that the arthropods were affected by farm 'intensity', as defined by crop rotation, manuring and soil cultivation. Broadly in accordance with the hypotheses, relatively high abundance ofpredaciou s beetles, lycosid spiders, granivorous beetles and folivores were found in crop areas with a comparatively low 'intensity'. In contrast to the hypotheses, small staphylinid beetles, as representing the detritivores, seemed tob e associated with a high 'intensity', and the linyphiid spiders suggested an association to large continuous fields. Contrary to improved E.I., traditionally managed E.I. enhanced specific groups like lycosid spiders, small staphylinids as well as ground dwelling herbivores. Improved E.I. did not support ahig h density of ground dwellers in the summer. Comparison of crop area and E.I. observations suggested that the most numerous epigeic groups (predacious beetles and linyphiid spiders) inbot h crop area and E.I. were affected by crop area conditions. Possible interactions between epigeic arthropod predators are discussed with reference to field observations.

1. Introduction

1.1.Biodiversity-loss in agriculture

Populations ofvariou s animal andplan t species on farmlands in industrialised countries are seriously declining duet o agricultural intensification (e.g. Paoletti, 1999; Fuller etal., 1995; Andreasen etal, 1996).Thi sbiodiversity-los s is amatte r of great concern for ecologists, agriculturists aswel l asth e general public (Paoletti, 1999;Woo d &Lenne , 1999),wh o are increasingly awareo fbot h the intrinsic aswel l asth e economic value of biodiversity. Torevers e thistren d ofbiodiversity-loss , development ofmulti-specie s or sustainable farming systems isprobabl y needed (Vandermeer etal, 1998;Altieri , 1994, 1999;Almekinder set al, 1995).Thes e farming systems utilise ecosystem servicesprovide d bybiodiversit y (e.g. pest control, soil fertility) and may also giveroo m for nature conservation-worthy biodiversity of whichth eutilit y isno tclea r (Duelli etal., 1999) . Several examples ofmulti-specie s farming system practices are elaborated by Altieri (1994),Woo d &Lenn e (1999), and other authors.

Development ofmulti-specie s farming-systems mayrequir e specific research approaches, particularly with regard to the study area and studied variables.Firs t theresearc h should take site-specificity ofbiodiversit y into account (Vandermeer etal, 1998),an d should preferably start from 'nearly natural systems'whic h exhibit theprocesse s onwhic h ecosystem services arebase d (Brown, 1999a). Second the variables, should specifically addressth e farm-, or higher levels of aggregation (Almekinders etal, 1995), andtherefor e address thewhol e biocoenosis (Biichset al, 1997). This may involve recognition of theholisti c concept that not 31 all structures atth e system level could bepredicte db y component knowledge (Stobbelaar & Van Mansvelt, 2000;Bockemuhl , 1986;Sheldrake , 1988, 1991).Thir d thepredicto r variables should remain clearly related to decisions in the farm management, sooutcome s canb e translated to farming practice andpolicy , including management of semi-natural habitatso n the farm (Vereijken, 1998;Smedin g& Joenje , 1999;Par k &Cousins , 1995).

1.2.Organic farms andecological infrastructure

Among appropriate objects inEurop e for empirical system-oriented research on farm biodiversity areprobabl y organic farms, corresponding toproductio n guidelines 2092/91 in the EC countries (Anonymous, 1991);th e maintenance andutilisatio n ofbiodiversit y is inherent toth ephilosoph y of organic farming. Research provides evidence that organic farming ascompare d to conventional farming involves increased numbers and species diversity ofbird s (Chamberlain etah, 1999;Braa e etat, 1988),vegetatio n dwelling arthropods (e.g. Moreby &Sotherton , 1997;Hal d &Reddersen , 1990), epigeic arthropods (reviewed by Kromp, 1999),earthworm s (Pfiffner &Mader , 1997),arabl e weeds (e.g. Van Elsen,2000 ; Smeding, 1993)an dhabitat s(e.g. Stobbelaar &Va nMansvelt , 2000; Chamberlain etal, 1999).

Ofparticula r interest,becaus e ofthei r increased similarity to 'nearlynatura l systems' (Brown, 1999a),migh tb e farms that combine organic cropping with apurposefull y enlarged, spatially distributed and diversified ecological infrastructure. In this article ecological infrastructure (E.I.) isdefine d asth e total of semi-natural (aquatic, woody, tall herb and grassy) habitats on the farm including its spatial arrangement (Smeding &Booij , 1999).I nvariou s farming systems, enlarged and diversified E.I. enhances thenumber s and diversity of indigenous plants and animals atth e farm level, including the crop area (e.g. Joenje etal., 1997;LaSalle , 1999;Boatma n etal., 1999).I n organic farms this may contribute topes t control (Kromp& Meindl, 1997;Theunisse n &Kohl , 1999) and to soil fertility (Brown, 1999b).

1.3. Foodweb approach

An appropriate methodology for biodiversity research, addressing variables athighe r levelso f aggregation, isoffere d by recent advances inresearc h of food webs (Polis& Winemiller, 1996;Pimm , 1991).Thes e advances, including some studies in agricultural habitats (e.g. De Ruiter etal, 1995;Tscharntke , 1997),ar eprovidin g inspiring examples ofho w (agro) ecosystems canb e approached athighe r integration levels than the species-level. Someo f this research indicates that farm food webs canmediat e the effects of environmental factors on species (Wiseet al, 1999;Tscharntke , 1997).

Our generalhypothesi s about the farm food web on organic arable farms with improved E.I. isbase d onthre e sub-hypotheses: First that the detrital web is enhanced by the increased organic matter input, involving organic manure, compost and crop residues (of cereals and leypasture) , and relates to increased numbers of invertebrate meso- andmacrofaun a (e.g. Pfiffner &Mader , 1997; Weber et al, 1997;Idinge r etal, 1996; Heimbach &Garbe , 1996). Second that numbers of variousnon - pest herbivorous functional groups are enhanced by improved E.I. (Holland &Fahrig ,2000 ; Boatman etal, 1999)bu t alsob y organic crop characteristics (Hald &Redderson , 1990; Morbeby &Sotherton , 1997).Third , combined increase ofdetrita l and herbivore invertebrates relates to acumulativ e positive effect on the abundance ofpredator s which feed onpre y from bothdetrita l and herbivore subsystems (Wiseet al, 1999;Idinge r etal, 1996;Sunderlan d et 32 al, 1996a;Kromp , 1999).Thi s effect is accelerated by the frequently occurring food shortage on farms, limiting the abundance ofbot hpredaciou s invertebrates (Booij etal., 1996)a swel l asvertebrate s (Poulsen etal, 1998).

Additionally we expect that succession of food web structure requires time (Idinger &Kromp , 1997;Idinge r etal, 1996),possibl y upt o atleas ton eo r two crop rotations. Former application ofpesticide s and depletion oforgani cmatte r and duet o reliance on artificial fertilizers may negatively influence therat eo fthi s succession {e.g. Paoletti, 1999).

1.4. Hypotheses andobjectives

Ourfield stud ywa s focussed on functional group composition ofgroun d dwellingarthropods , comprising detritivore,herbivor e andpredato r groups.Particularl y the epigeicpredator s have major quantitative and functional importance inth efar m food web, linking thedetrita l and grazer food chains (e.g. Kromp, 1999;Sunderland , 1991).Complementar y studies,publishe d elsewhere, weredevote d tovegetatio n dwelling arthropods (Smeding etal., 2001b ) and insectivorous birds (Smeding &Booij , 2001).

Applying the farm food web hypotheses to ground dwelling arthropod functional groupsw e expected to find more epigeicpredators , detritivores andherbivore s incrop so forgani c farms with alon gduratio n and/or extensive crop rotation (i.e.hig hproportio n of gramineous crops). Weals o expected to observepositiv e effects of improved E.I. (i.e.larg e areaan d appropriate management) onth e abundance ofpredaciou s andherbivorou s functional groups inbot h crops andE.I .Althoug hpart so fthi shypothese s are alreadyteste d (references mentioned above and e.g. Ryszkowski etal. (1993 ) andRyszkowsk i &Kar g (1991),wh o compared above-ground insect biomass invariou sagricultura l landscapes),hypothese s integrating the effects ofduratio n of organicmanagemen t and effects of E.I.managemen t on functional groupcompositio n haveno tbee nteste d so far.

Thus,th eobjective s ofthi s studywere : 1. to determine the abundance of ground dwelling arthropod functional groups inth ecro p area and inth e ecological infrastructure of organic farms, and relateth e abundance of these groups to crop area and ecological infrastructure characteristics; 2. to relateth e abundance ofgroun d dwelling arthropod functional groups inth e crop areat o their abundance inth e ecological infrastructure.

2. Materials and methods

2.1. Study area

The study was done on organic farms inth eprovinc e ofFlevoland , TheNetherlands . The farms are located inth ethre epolder sNoordoostpolder , Oostelijk Flevoland and Zuidelijk Flevoland (Fig.l). These threepolder swer e reclaimed around 1940, 1957 and 1968 respectively, and are dominated by agricultural landuse. The soil is acalcareou s clay-sandy clay ofmarin e origin. Flevoland was chosen as astud y areabecaus e itha s aconcentratio n of organic farms that are similar with respect to theirhistor y andtopography . There are currently 75 economically viableorgani c arable farms inFlevoland , which are generally large (c. 30-70 ha) and intensive relative to other organic arable farms in .

33 Eight farms were selected based on differences in crop area and E.I. characteristics (Table 1). The selection included seven commercial organic farms andon e experimental farm. The selected commercial farms were all arable, although two (farms N and T)possesse d astabl e with livestock, which received fodder crops from the same farm but grazed elsewhere during the summer. The experimental farm of Wageningen University (farm E) is amixe d dairyan d arable farm. Four commercial farms (farms A, T, L, and D)wer e Ecological Arable Farming Systems (EAFS)prototyp e farms during 1992-1997 (Vereijken, 1997);amon g these farms, farm Ai sdifferen t because it is situated inth e youngest polder Zuidelijk Flevoland andjoine d theEAFS-projec t a few years latertha nth eothers .Farm sA an dN started in 1985o na n experimental reclamation area, which hadneve r received pesticides.

w

Fig. 1.Study area. Theinvestigated organic arablefarms are represented by closed circles.

2.1.1. Crop area characteristics Organic duration of theselecte d farms ranged from onet o twelve years.Ploughin g isdon e generally at25-3 0 cm depth. On allth e farms a6- 7 year crop rotation ismaintained , with 6-11differen t cropspe r farm. Together the farms comprised 542 ha arable land withpe r farm 15-33% cereals,0-36 % leypasture , 10-31% traditional lifted crops (potatoes, sugarbeet ) and 19-69%vegetabl e crops.Vegetable s involve mainly onions,runne r beans, various cabbages, peas, and sweet corn.Th efarm s withmos t extensive croprotation s had ahig h percentage (>30%) of gramineous crops (cereals and leypasture ) andtherefor e alo wpercentag e (<50%) ofvegetabl e crops.A scompare d to other crops,gramineou s cropsprovid e alarg e amount of carbon incro p residues andrequir e few soil disturbing operations. Because input of nutrients (N,P ,K )ha s tob e adjusted to crop requirements, andvegetable s need more nutrients than cereals, intensity ofcro p rotation isassume d torelat e tobot h the amount of cropresidue s and the amount ofN-applicatio n at the farm level.

34 Table1 Characteristics of thefarms and ecological infrastructure and total numbersperform of arthropod functional groups captured in weekly trappings over an eight weekperiod inthe wheat crop (30 mfrom the field margin) and in the ecological infrastructure.

Level Variables D T F R L E A N Farm n Yearo forgani c farming (n) dur 8 8 7 1 10 4 12 12 Farm Cereals and leypastur e (%) int 35 38 24 18 11 61 38 59 Farm Cropswithi n 100m rang e ofE.I . (%) ei 100 72 74 74 65 80 90 47 40 Farm Semi-natural habitat (E.I.) (%) area 3.5 2.3 1.4 1.6 3.1 4.5 2.1 0.8 E.I. Mowing date (farm level) (month) mow 7 9 5 5 6 6 7 5 E.I. Mowing frequency (n/year) mowfrq 2 1 2 3 2 2 2 3 E.I. Height vegetation (m) eihgt 1.4 2.4 1.1 0.7 0.8 1.5 0.4 0.5 E.I. Cover vegetation (%) eicov 95 95 88 77 85 100 98 67 E.I. Species diversity vegetation (n/4m 2) eispn 16.0 12.0 9.5 7.5 13.2 11.8 10.5 9.3 E.I. Dominant species diversity (n/4m 2) eidom 3.8 3.4 3.1 2.3 3.0 3.4 3.4 2.8 Agepolde r (year) lsage 57 41 41 41 41 41 30 30

Crop Predacious beetles 1731 1333 3765 1011 1035 2889 952 1064 Crop Lycosid spiders 8 57 38 18 83 35 53 146 Crop Linyphiid spiders 1944 2979 2622 2515 2055 1687 3061 3311 Crop Small staphylinid beetles 541 893 425 640 772 205 1308 329 Crop Granivorous carabids 15 7 4 3 19 504 5 83 Crop Folivores 5 7 21 13 6 8 4 3

E.I. Predacious beetles 926 1205 2808 832 1706 2603 1365 1319 E.I. Lycosid spiders 64 208 302 46 168 226 87 535 E.I. Linyphiid spiders 2554 880 4235 1191 1607 1930 2508 2577 E.I. Small staphylinid beetles 709 444 334 365 217 411 739 703 E.I. Granivorous carabids 66 83 79 253 291 279 52 149 E.I. Folivores 100 150 33 38 61 35 104 275

2.1.2.Ecological infrastructure characteristics TheEcologica l Infrastructure (E.I.) of arable farms inFlevolan d ismainl y determined by canals andbank s (together c. 3-4m width )tha t are laid out as linearherbaceou s boundaries adjacent to the crops.O n five farms, boundaries arewide rbecaus e of adjacent (2-3 mwide ) grass strips,ofte n including clovers (Trifolium spp.),o n oneo rbot h sideso fth e canals.Far m E also has grass stripsbetwee n crop plots.Percentage s of E.I.highe r than c. 2%(Tabl e 1) weremainl y due to thepresenc e ofthes egras s strips.Onl y farms Lan d Epossesse d a boundary planted with ahedgerow . Spatial density ofE.I .i spurposefull y increased on farm E (Smeding &Joenje , 1999)an d farm L.Th e spatial arrangement of E.I. onth e other 6 farms mainly reflect landscape scale. Extreme types of vegetation management are: - 'traditional management' involving three or four more cutspe r year with a flail mower leaving the shredded cuttings in situ; - 'latemowin g management' involving oneo rtw o cutspe ryea r after June 21st witha finger- barmowe r with ahia b grab for removal ofcuttings .

The grass strips along the canals arerequire d forfield transpor t in 'latemowin gmanagement' . The E.I. area is,therefore , related to the mowing date. Inthi s article the E.I. with enlarged area and latemowin g isdefine d as'improve d E.I.'. The improved E.I. onth e four farms (A,T , L, andD )wa s created during the EAFS-prototyping research program in 1992-1997

35 (Vereijken, 1997, 1998).Th e improved E.I. onth e experimental farm Estarte d in 1995 simultaneously with conversion of theconventiona l dairy farm to an organic mixed farm (Smeding &Joenje , 1999).

Dominant plant species incana lban k swards are:couc h (Elymus repens),perennia l meadow grasses (Festuca rubra, Poa trivialis, Loliumperenne, Agrostis stolonifera), commo n reed (Phragmitis australis)an d afe w perennial herbs: dandelion {Taraxacum officinale) and stinging nettle (Urtica dioica).Canal s with traditional management have a low, relatively open, speciespoo r vegetation (<10species/ 4 m2),dominate d by a few grass species. Late mowingparticularl yrelate st oincrease d vegetation biomassi nJun ean dJul y asexpresse db y vegetation height (upt o 1-3 m) and cover (up to 90-100%). However, latemowin g doesno t necessarily relatet o increased higherplan t species diversity for two reasons: a)tal l grass speciesma y dominate the sward; b) swards oftraditiona l management may locally be cut to ground level and therefore be invaded by ruderals and annuals (increased species numberu p to 15species/ 4 m2 ),e.g. thistles (Cirsium arvense, Sonchusarvensis), annua l weeds(e.g. Sonchusasper, Polygonumaviculare) an d annuals that are rarelyperniciou s weeds in Flevoland (e.g. Myosotisarvensis, Veronicapersica, Cardamine hirsuta).

Highperennia l plant diversities found insom e locations may relate tovariou s interrelated factors (Kleijn, 1997):reduce d vegetation productivity, removal ofhay , nutrient buffering by thegras s strip, and long organicduration .However , also, artificial species introduction is involved; inth eEAFS-project , around 90 different nativeperennia l dicots were artificially seeded inth e canal banks toobtai n higher plant diversity and flower abundance ofplant s with limited dispersal capacity (Vereijken, 1998).Althoug h few species were able to settle well (e.g. Seneciojacobea, Crepis biennis, Heracleum sphondylium), the introduction clearly affected speciesdiversity : vegetation including siteswit h>1 5 species/4 m wereconfine d to prototype farms.

2.2.Sampling and measurements

2.2.1.Arthropod sampling Sampling was done inwhea t crops and in adjacent canals banks (E.I.) which were situated in the interior of theorgani c farm area. Wheat (Triticum aestivumL. ) was chosen asa representative crop asthi s crop is common onorgani c arable farms inFlevolan d and arthropod numbers incereal s are large compared to numbers inothe r common crops like potatoes and onions(Booi j &Noorlander , 1992).Withi n onewhea t field on each ofth e eight farms one 60x30 m2 sampling areawa s selected; this areawa s 60m longborderin g acana l by 30m perpendicula r to that canal. Sixpitfal l traps (10c m diameter) wereplace d in aro wwit h intervals of 10m a t distances of 0m (E.I.) and 30m (wheat) from the field boundary. Traps contained apreservativ e of 4% formaline plus detergent. Weekly samples were taken during8 weeks,startin g from 6th ofJun eunti l 28th ofJul y 1998.Th eobservation s ofth esi xtrap s and eight sampling dates were treated separately, giving 48measurement s per habitat (cropo r E.I.)pe r farm. Springtails (Collembola), and ants (Formicidae)wer e discarded because pitfall traps dono t giverepresentativ e measurements ofthes egroups . Arthropods were stored in 95% alcohol. Individuals were sorted to taxa at order- or family-level andplace d into six different arthropod functional groups: - Predacious polyphagous beetles including carnivorous Carabidae and large (>6mm ) staphylinid beetles; - Lycosid spiders which arerelativel y large polyphagous predators andrepresen t diurnal wandering spiders (Marc etal, 1999; Sunderland, 1991);

36 - Linyphiid spiders which are relatively small species and represent highly dispersive sheet- web spiders (Marc etai, 1999;Sunderland , 1991); - Small staphylinid beetles ofth e epigeobiontic lifeform (Bohac, 1999)whic h aremainl y fungivorous andrepresen t the detritivore functional group among epigeic arthropods; - Granivorous carabid beetles (generaAmara an dHarpalus) representin g the invertebrate granivore functional group; although these carabids may feed, to some extent, on invertebrates aswel l (Westerman etal, inprep.) ; - Folivorous herbivores represented by weevils (Curculionidae) and leaf hoppers (Cicadellidae).

2.2.2. Vegetation measurements Vegetation studies included the E.I. wherepitfall s were located. Theinvestigatio n was done from Mayt oAugust : latemow n vegetation was sampled inJune-Jul ybefor e thefirst cut . Early ando r frequently mown vegetation was sampled before thefirst cu t (May) or later (until August) when species composition could be identified adequately. The herb layerwa s sampled in sevenplot s of 4m 2are awit h intervals of 50m betwee n samples. Plots included the ditchban k vegetation from theto p toth ehelophyti c zone;vegetatio n affected by flooding was excluded. Parameters were:cover , maximum height, plant species and dominant plant species (totalling 80%cover) .Average s were calculated for each farm.

2.2.3.Farm management andlay out A 1:5000ma p wasmad e of each farm, depicting the different crops and the ecological infrastructure (E.I.). Onthes emap s the area of semi-natural and crop habitats were measured. The 'E.I.-area' is calculated asth epercentag e of semi-natural habitat of the farm area (without the farm yard).Far m rotation intensity isdefine d asth eproportio n of gramineous crops (cereals and leypasture ) ofth e total crop area. The spatial arrangement of theE.I . or'E.I . density', was expressed asth epercentag e ofth e farm within a 100m distanc e ofth eE.I .I n case the canalban k vegetation was cutbot h after September 15th andbefor e May 15th,a distance of 70m wa s considered. The influence offield margin s onth e abundance of large carabid beetles and spidersprobabl y doesno t exceed 70-100m (Booij,unpublishe d data).

Datao nth eyea r ofconversio n of allo fth e farm toorganic ,husbandr ypractice s inth ewhea t crop in 1998,an dmowin g of canalsbank s and grass stripswer e obtained using a questionnaire. The mowing management isrepresente d by themont h ofth e first cut, because this cut largely determines thephenologica l stage ofth evegetatio n inJun e and July.

2.3. Statisticalanalyses

Data from all sampling periodswer e combined, tested for normality and log-transformed. All subsequent analyses were conducted using Statistical Analysis System (SAS Institute, Cary, North Carolina).

Individual farms were classified into one ofthre egroup s on thebasi s of corresponding farm and E.I. characteristics (Table 1).Wit hregar d to the crop area samples,th e classifications werebase d onrelevan t farm level characteristics (duration, rotation intensity, area ofE.I. , and E.I. density).Wit h regard toth e E.I. samples,th e classifications werebase d onbot h farm level characteristics aswel l asE.I.-vegetatio n characteristics (duration, E.I.-area, month of first mowing, vegetation cover, vegetation height, plant speciesnumbe r or dominant plant species numberpe r 4m 2). The classifications were determined by two splits:th efirst spli t based onon eo fth e abovementione d characteristics, separated acategor y of2- 3 farms and the second splitbase d on another ofthes e characteristics, divided theremainin g 5-6 farms ina 37 second and athir d category. Theminimu m category sizewa s2 farms. For all obtained classifications canonical discriminant analyses wereperforme d on thenumber s ofth e arthropod functional groups.Dat a ofth e crop area ando fth e E.I.wer e analysed separately for crop area andE.I . classifications respectively. Theclassification s with thehighes t Wilks' Lambda F-valuewer e selected.

Stepwise andcanonica l discriminant analyses wereperforme d on thenumber s of arthropod functional groupst o determine themos t important groupstha t could classify the farms into three categories. Arthropod observations inth ewhea t and inth e E.I.wer e again analysed separately. Stepwise discriminant analysiswa s also used to identify variables that contributed mostt o the selected classification. Canonical discriminant analysiswa s used to determine the magnitude and direction ofth e association of individual variables with indicator variables. Standardised canonical coefficients larger than 0.3 divided by the squareroo t ofth e eigenvalue ofth e canonical function (Afifi &Clark , 1984)wer e considered large enough to contribute significantly to the classification.

Differences in the abundance of arthropod functional groupsbetwee n the three categories wereteste d by linear regression (GLM) usinu g the 10log-transformed measurements and including aDuncan' s multiple range test.

Thecombine d data-set ofwhea t and E.I. sampleswa s used to analyse the abundance of arthropod functional groups along the gradientfrom E.I .t o crop centre, represented bytw o sampling distances 0m (E.I.) and 30m (wheat)from th efield margin . ANOVA was performed onth etw o selected classifications to determine interactions between distance and category.

3. Results

3.1. Observednumbers ofepigeic arthropodfunctional groups

During eight weeks 77,784 organisms werecaugh t in 96pitfal l traps including 40,233 organisms inth e crop area and 37,551organism s inth e Ecological Infrastructure. Crop area samples comprised 33% predacious carabid beetles, 1%larg e (>6mm )predaciou s staphylinid beetles, 1%lycosi d spiders, 50%linyphii d spiders, 13% small (<6 mm) detritivorous staphylinid beetles, 0% folivores and 2%granivorou s carabid beetles. E.I. samples comprised 33% predacious carabid beetles, 1%larg epredaciou s staphylinid beetles,4 % lycosid spiders, 10%smal l detritivorous staphylinid beetles,47 %linyphii d spiders,2 % folivores (Curculionide and Cicadellidae) and2 %granivorou s carabid beetles.

3.2. Arthropodfunctional groupsin the crop area, affecting distinctions amongfarm categories

Theclassificatio n 'int/area'base d on crop rotation intensity (first split) and area ofth e ecological infrastructure (second split) gave thebes t separation ofcro p area arthropod observations (canonical discriminant analysis, Wilks' Lambda, F-value=30.1,PO.0001 ; Table 2). The second best classification 'eilOO/dur' (Wilks' Lambda, F-value=28.1, PO.0001) was interesting because the distinction was affected by the the most abundant group predacious beetles,wherea sth e distinction ofth ebes t classification was only affected by less abundant groups.Therefor e both classifications arepresente d below.

38 Table2 Examples of classifications of the crop area and the ecological infrastructure, including the the Wilks'Lambda F-value and thesquared canonical correlation obtained by canonical discriminant analysis. The abbreviations of category characteristics are explained in Table1.

Crop area: Category 1 Category 2 Category 3 F-value Sq. Can. Category int>40 int<40 & area >2.2 int<40 & area <2.2 30.1 0.52 Farms NE TLD AFR Category eil00<70 eil00>70 & dur<8 eil00>70&dur>8 28.5 0.46 Farms ANR FE TLD Category int<20 int>20 & eil00<50 int>20 & eil00<50 22.8 0.34 Farms RL AN TFED Category int<30 int>30 & area<2.2 int>30 & area>2.2 19.7 0.32 Farms ARL FN TED Category dur<6 dur>6 & area<2.2 dur>6 & area>2.2 12.0 0.19 Farms RE ANF TLD

Ecologica 1 infrastru :ture: Category 1 Category 2 Category 3 F-value Sq. Can. Category eispn>12 eispn<12 & eicov>80 eispn<12 & eicov<80 15.9 0.32 Farms TLD AFE NR Category eispn>12 eispn<12 & eihgt>l eispn<12 & eihgt<1 15.7 0.33 Farms TLD FE ANR Category eihgt<0.5 eihgt>0.5 & mow<7 eihgt>0.5 & mow>7 14.2 0.27 Farms AN FRL TED Category mow>7 mow<7 & eidom>3 mow<7 & eidom<3 12.7 0.25 Farms TED FL ANR Category eicov<80 eicov>80 & mow<7 eicov>80 & mow>7 11.9 0.23 Farms NR AFL TED

Table3 Arthropod variables that contributed significantly to classification of the crop area observations in'int/area' farm categories bystepwise and canonical discriminant analyses, standardised canonical coefficients, and mean valuesper crop area category. Comparison of arthropod numbers in the categories by Duncan's multiple range test.

Canlb Mean values and multiple rang e comparison

Standardised Category 1: Category 2: Category 3: coefficient int>40 int<40 & int<40 & area>2.2 area<2.2 Variables a mean mean mean Granivorous carabids 0.89 c 6.1 a 0.3 b 0.1 b (PO.0001) Lycosid spiders 0.43 1.9 a 1.0 b 0.8 b (P<0.0001) Small staphylinids -0.39 5.6 b 15.3 a 16.5 a (P<0.0001) Linyphiid spiders 0.15 52.1 b 48.5 b 56.9 a (PO.01) Folivorous insects -0.05 0.1 b 0.1 b 0.3 a (P<0.05) Predacious beetles -0.04 41.3 a 28.5 b 39.8 b (P<0.05)

"Variable s are listed in order of selection by stepwise discriminant analysis. Italicised variable was not selected by stepwise discriminant analysis. bCanonica l function 1;responsibl e for 95%o f the variation. cStandardise d coefficients >0.29 were considered large enough for interpretation.

39 3.2.1. Classification'int/area' Category 1 included farms with an extensive crop rotation (>40%cereal s and leypasture) ; category 2include d farms with an intensive crop rotation and alarg eE.I . area (which were thethre e former prototype farms in Oostelijk Flevoland);categor y 3include d farms with an intensive crop rotation and asmal l E.I. area.

All arthropod functional groups except predacious beetles,wer e selected asimportan t factors by stepwise discriminant analysis for discrimination among thethre e categories (Table3) . The first canonical function containing granivorous carabid beetles, lycosid spiders,an d small staphylinid beetles separated category 1 (extensive crop rotation) farms from both other categories with intensive croprotatio n (Fig. 2);granivorou s carabids and lycosid spiders were positively associated, and small staphylinid beetleswer e negatively associated with this first axis (Table 3).I n the second canonical function noneo f thearthropo d groups had sufficient canonical loading; however, linyphiid spiders contributed most to the separation of categories 2 and 3,an dwer epositivel y associated with the small E.I. area of category 3.

A

D

- cJ&Acg nDQn °

•

1 -2 4 10 can 1

Fig. 2. Plot of the second andfirst canonicalfunctions discriminating amongfarm categories of the classification 'int/area'.Squares representfarms with >40%gramineous crops; circles representfarms with <40%gramineous crops and >2.2%E.I. area; triangles representfarms with >40%gramineous crops and <2.2%E.I.

On farms with extensive crop rotation (category 1),granivorou s carabid beetles were>2 0 times more abundant on average, lycosid spiderswer e>90 %mor e abundant on average and small staphylinid beetles were >60%les s abundant onaverage ,tha n onth e othertw o categories. Linyphiid spiders were 15%mor e abundant on average in category 3(intensiv e croprotatio n with small E.I.) than incategor y 2(intensiv e crop rotation with large E.I.) (Table3) .

3.2.2.Classification 'eilOO/dur' Category 1 included farms with large scale E.I./landscape; category 2include d farms witha spatially denseE.I . andmoderatel y long (4-7years )organi c duration; category 3farm s had a spatially dense E.I. and alon g organic duration (8-10 years) (these wereth e three former prototype farms in Oostelijk Flevoland). 40 The separation into 'eilOO/dur'categories ,base d on arthropod functional group numbers in wheat, isillustrate d in aplo t ofcanonica l variable two versus canonical variable one (Fig.2) . All arthropod functional groupswer e selected asimportan t factors by stepwise discriminant analysis for discrimination among thethre e categories (Table 4). The first canonical function containing predacious beetles,smal l staphylinid beetles and granivorous carabids separated the farms with dense E.I. andmoderat e duration (category 2) from both other categories (Fig. 2). Predacious beetles andgranivorou s carabidswer epositivel y associated and small staphylinid beetles werenegativel y associated with thefirst axi s (Table 4).I n the second canonical function none ofth e arthropod groupsha d sufficient canonical loading; however, linyphiid spiders contributed most the separation of category 3an d 1,an d were positively associated with alarg e scale landscape (category 1).

In extensive farms (category 1),predaciou s beetleswer e >2 times more abundant on average, granivorous beetles >8time s more abundant and small staphylinid beetles were >2 times less abundant on average, than in the other two categories. Linyphiid spiderswer e >20%mor e abundant on average incategor y 3(larg e scale landscape farms) thani nbot hothe r categories (Table4) .

Table4 Arthropod variables that contributed significantly to classification of the crop area observations in 'eilOO/dur' farm categories bystepwise and canonical discriminant analyses, standardised canonical coefficients, and mean valuesper crop area category. Comparison of arthropod numbers in the categories by Duncan's multiple range test.

Canlb Mean values and multiple range comparison

Standardised Category 1: Category 2: Category 3: coefficient eil00<70 eiI00>70& eiI00>70& dur<8 dur>8 Variablesa mean mean mean Predacious beetles 0.86c 21.1 c 69.3 a 28.5 b (PO.0001) Small staphylinid beetles -0.42 15.8 a 6.6 b 15.3 a (PO.0001) Lycosid spiders -0.26 1.5 a 0.8 b 1.0 b (P<0.0001) Granivorous carabid beetles 0.28 0.6 b 5.3 a 0.3 b (P<0.0001) Linyphiid spiders -0.06 61.7 a 44.9 b 48.5 b (P<0.0001) Folivorous insects 0.16 0.1 b 0.3 a 0.1 b (PO.05) aVariable s are listed in order of selection by stepwise discriminant analysis. bCanonica l function 1;responsibl e for 86% of the variation. cStandardise d coefficients >0.32 were considered large enough for interpretation.

3.3. Arthropodfunctional groups inthe Ecological Infrastructure affectingdistinctions among E.I. categories

Theclassificatio n 'eispn/eicov', based onE.I.-specie s number (first split) and E.I.-vegetation cover as second split gave the best separation of observations (canonical discriminant analysis, Wilks' Lambda, F-value=15.9 (P<0.0001);Tabl e 2).Classification s based on organic duration, E.I. area, date offirst mowing , E.I.-vegetation height and dominant species number showed less distinction than the classification 'eispn/eicov' (Table 2).Categor y 1 included species rich E.I. onprototyp e farms in Oost Flevoland; category 2 included ales s speciesrich E.I. with adens e vegetation; category 3include d aspecie s poor E.I.wit h open vegetation related totraditiona l mowing management.

41 The separation into 'eispn/eicov' categories,base d on arthropod functional group numbers in theE.I. , isillustrate d in aplo t of canonical variable twoversu s canonical variable one(Fig.3) . All arthropod functional groups were selected asimportan t factors by stepwise discriminant analysis for discrimination among the threecategorie s (Table 5).Th e first canonical function separated the speciespoo r categories 2 and 3 from each other (Fig. 3);predaciou s beetles and linyphiid spiders werepositivel y associated with densevegetatio n and granivorous carabid beetleswer enegativel y associated with densevegetatio n (Table 5).I nth e second canonical function none ofth e arthropod groupsha d sufficient canonical loading; however, linyphiid spiders contributed most to the separation of category 1 from both categories 2 and 3,an d werenegativel y associated with species rich vegetation.

can 1

Fig. 3. Plot of thesecond and first canonicalfunctions discriminating among E.I. categories of the classification 'eispn/eicov'. Circlesrepresent theE.I. withplant species number >I2; squares represent E.I. withplant species number <12 and vegetation cover >80%; triangles represent E.I.with plant species number

Predacious carabid beetles and linyphiid spiders were>70 %an d >50%respectivel y more abundant inth eE.I . with few plant species anddens e vegetation (category 2) as compared to other catagories. Granivorous carabids were >30%mor e abundant inE.I . with species poor and open vegetation (category 3;'traditiona l management'); also folivores were more abundant herebu t did not contribute toth e distinctions between the categories because of their lownumber s (Table5) .

42 Table5 Arthropod variables that contributed significantly to classification of E.I. observations in 'eispn/eicov'categories bystepwise and canonical discriminant analyses, standardised canonical coefficients, and mean valuesper crop area category. Comparison of arthropod numbers in the categories byDuncan's multiple range test.

Canlb Mean values and multiple -ange comparison

Standardised Category 1: Category 2: Categorj '3: coefficient eispn>12 eispn<12& eispn<12 & eicov>80 eicov<80 Variablesa mean mean mean Predacious beetles -0.92c 26.7 b 47.2 a 22.6 b (P<0.001) Granivorous carabids 0.52 3.1 b 2.9 b 4.2 a (PO.001) Linyphiid spiders -0.51 35.0 b 60.2 a 39.3 b (P<0.001) Small staphylinids 0.43 9.5 b 10.3 b 11.1 a (P<0.1) Folivorous insects 0.32 1.1 b 0.8 b 1.7 a (PO.001) Lycosid spiders 0.27 3.1 b 4.3 b 6.1 a (P<0.05)

"Variables are listed in order of selection by stepwise discriminant analysis. bCanonica l function 1;responsibl e for 89%o f the variation. cStandardise d coefficients >0.43 were considered large enough for interpretation.

3.4.Interactions between distanceto E.I. andfarm category

Abundance of lycosid spiders, granivorous carabid beetles and folivores were significantly higher in the E.I. than in the wheat (Table 6).Th e distribution of linyphiid spiders showed a preference for the crop area.Abundanc e ofpredaciou s beetles and small staphylinid beetles did not significantly differ between crop area and E.I. (Table6) .

Table6 Average numbers of arthropodfunctional groups in theE.I. and inthe wheat crop (30 mfrom thefield margin). Comparison of log-transformed values byDuncan's multiple range test.

Arthropod functional group E.I. Crop area Significance Predacious beetles 12764 a 13780 a (NS) Lycosid spiders 1636 a 438 b (PO.001) Linyphiid spiders 17482 b 20174 a (PO.001) Small staphylinids 3922 a 5113 a (NS) Granivorous carabids 1252 a 640 b (PO.001) Folivorous insects 446 a 67 b (PO.001)

High numbers ofpredaciou s beetles and lycosid spiders inth e E.I. coincided with high numbers in the crop area (Fig. 4a and 4b).Th e differences in linyphiid spiders (Fig. 4c)an d granivorous carabid beetles (Fig. 4d)wer e similar for the E.I. categories 1('prototyp e farms') and 3('traditiona l management'). However, the difference of these species was in the opposite direction in E.I.-category 2 (speciespoo r and dense vegetation): high linyphiid abundance in the E.I. decreased towards the crop area, while it increased inbot h other categories (Fig.4c) ; granivorous carabid beetles incategor y 2wer emos t numerous in the crop area and least numerous in the E.I. (Fig. 4e). Increased numbers of small staphylinid beetles and folivores in the E.I. did not always coincide with increased numbers in the crop area (Fig. 4e and 4f): distribution patterns of small staphylinids were very variable; folivore abundance inth eE.I . was different between categories but was generally low inth e crop area.

43 A: predacious beetles B: lycosid spiders 0.6 2

00 1.5 60 o O 2 0.2 1 Om 30 m Om 30 m

C: linyphiid spiders D: small staphylinids 2 i 1.1

0.8 0 m 30 m Om 30 m

E: granivorous carabids F: folivorous insects 0.8 0.3 0.6 0.2 0.4 o 0.2 - 0.1 0 0m 30m 0 m 30 m

Fig. 4. Theabundance of arthropodjunctional groups in the E.I. (0m) and in the crop area (30 m). E.I. categories of selected classification 'eispn/eicov': Category I (squares)= species rich E.I. vegetation; category 2 (triangles)= species poor and dense vegetation; category 3= (circles) species poor and sparse vegetation. Lines between the averages of arthropodfunctional group numbers('" log-transformed) are drawn to illustrate the differentpatterns among the threeE.I. categories and do not represent estimatedpoints in between the sampling stations, (a)predacious beetles (b) lycosid spiders; (c) linyphiid spiders; (d)small staphylinid beetles; (e) granivorous carabid beetles; (f)folivorous insects.

4. Discussion

4.1.Epigeic arthropods in the crop area

4.1.1. Observations comparedto the hypotheses Itwa s expected that organic duration, extensive crop rotation, area ofE.I. , and high spatial density of E.I.woul d relatepositivel y with the abundance ofpredaciou s and herbivorous ground dwelling arthropods. Also,tha t organic duration and extensive crop rotation would relatepositivel y toth eabundanc e ofdetritivorou s epigeic arthropods. Allmentione d factors indeed related to differences inth e abundance of thearthropo d functional groups involved in

44 the study. However only afe w ofth e observations ofth e direction ofchange s of abundance, were in accordance with thehypothesis : lycosid spiders (which were the least abundant predacious functional group) and granivorous carabidbeetle s showed affinity for extensive crop rotation; predacious beetles were associated with the farms with dense E.I.

Most observations didno t correspondwit h ourhypotheses . Extensive croprotation s (farm N andE )ha d alo w abundance of (epigeic) detritivores, represented by small staphylinids. Predacious beetles and linyphiids,representin g 84%o f allobserve d epigeic arthropods inth e crop area, didno tprefe r extensive croprotation , neither alarg eE.I . areano r alon gorgani c duration. Contrary to expectation, prototype farms which had alon g duration, alarg e E.I. area and adens e spatial arrangement ofE.I. ,harboure d comparatively low abundance of all five predacious andherbivorou s functional groups.

4.1.2.Small staphylinid beetles Contrary to ourhypothesis , an abundance of detritivorous staphylinids seemed tob e associated with 'intensity' including not onlypercentag e gramineous crops but also interrelated management factors like manure application, soil cultivation, weed control and pest control by entomopathogens. Ambiguous behaviour of staphylinids with regard to intensity, ascompare d to other large epigeic groups,i s found in studieswhic h relatet o tillage treatment (Heimbach &Garbe , 1996)an d farming systems comparisons (Booij &Noorlander , 1992;Reddersen , 1997;Biich set al, 1997).However , positive effects on staphylinids were reported with regard tomanur e (Bohac, 1999)an dhous ewast e compost (Idinger &Kromp , 1997).I n acompariso n offour , long-duration organic farms ina nol d culture landscape and polder landscape, more staphylinids were found inth epolde r which represents an earlier successional stage of soil community (Smeding &Booij , 1999).

Our observations and data from literature may suggest that small staphylinid abundance is positively related to acertai n quantity of organicmanur e and frequent soil cultivation. Small staphylinidsma ytherefor e notb e anappropriat e indicator for the aggregate functional group of saprovorous arthropods (including also large Collembola and Diptera larvae); lowC/ N ratio andfiber conten t of organicmatte r input could relatet o within this functional groupb y small staphylinids.

4.1.3. Predacious beetles The observed highest densities ofpredaciou s beetles (mainly carabids) wereno t associated with long organic duration or extensive crop rotation. This observation iscontradicte d by several field studies in which carabids were found tob e associated with less intensive or organic crops (e.g. Booij &Noorlander , 1992;Hollan d etal, 1996;Basedow , 1991). Several authors reported that the underlying factors whichpositivel y affected carabids were crop cover and weeds,manur e and less frequent soil cultivation (DenNij s etal., 1996;Hollan det al., 1996; Booij &Noorlander , 1992;Idinge r etal, 1996).Therefor e it couldb epossibl e that the farm category 2wit h moderately long-duration farms might haveha d a lower 'intensity' (according to theabov e givenwide r definition) thancategor y 3wit h long-duration farms; category 3include d former prototype farms which had proceeded further in optimalisation of intensive vegetable production. Wheat crops in 1998 of farm category 2ha d relatively high weed densities, including Capsella bursa-pastoris on farm Ean dPolygonum aviculare o n farm F,an d therewa s an apparent abundance of springtails on farm F, although wedi dno t measure these features. Onbot h farms, therobus t carabid speciesPterostichus melanarius dominated thecarabi d community, whichwa sals o found inth emos t extensive systems of comparative studiesb y Booij &Noorlande r (1992) and Holland etal. (1996) .

45 4.1.4. Spiders Superior dispersive abitilies of linyphiid spiders are stressed by several literature sources {e.g. Sunderland, 1991; Halley et al, 1996). This could explain the observed positive association with farms that have a low E.I. density. Both farms in Zuidelijk Flevoland had a long organic duration but the polder is young (1968) which implies larger fields and also a possibly earlier successional stage of the soil biocoenosis. Farm R is the most recently converted farm in the study area and is situated centrally in a very open landscape part of Oostelijk Flevoland. Abundance of linyphiid spiders might also indicate an earlier successional stage of the epigeic community (that could be sustained in large scale landscape with less reinvasion possibility). Accordingly, Idinger et al. (1997) observed most linyphiids in the inorganic as compared to organic treatments.

In all the epigeic arthropod data there was a negative correlation between large organisms (mainly carabids) and small organisms (mainly linyphiids). This suggested that intraguild predation might influence functional group composition of ground dwellers. According to Sunderland et al. (1996b) linyphiid abundance might be affected by predation. Lycosid spiders were positively associated with extensive crop rotation. This was in accordance with other studies which found that lycosids prefer organic above conventional farms (Idinger et al, 1997) and belong to later stages of succession (Biichs et al, 1997). This suggests that lycosids are the most vulnerable functional group among predacious ground dwellers.

4.1.5. Ground dwelling herbivores With regard to granivorous carabids, extensive crop rotation or lower 'intensity' was probably correlated with (weed) seed availability. Several studies reported the relation between mobile granivorous carabids and presence of weeds (De Snoo, 1995; Kromp, 1999). Folivores seemed to be an insignificant group in the crop area and were slightly affected by 'intensity' {e.g. weed control). The higher mid-field densities in less intensive systems observed by Hald & Reddersen (1990) and Biichs et al. (1997) were collected by vacuum sampling or traps respectively, which might be more appropriate sampling methods for this group.

4.2. Epigeic arthropods in the Ecological Infrastructure

4.2.1. Observations compared to the hypotheses Our hypothesis was that improved E.I. would promote the densities of predacious and herbivorous epigeic arthropod functional groups. However, observations did not provide evidence for this hypothesis. The E.I. category based on high plant species number, included the most succesful examples of improved E.I. This catogory involved the three former prototype farms in Oostelijk Flevoland. In this E.I. all six investigated functional groups had decreased densities as compared to other categories. Highest densities of ground dwelling detritivore and herbivore functional groups were found in the E.I. category with presumed adverse management ('traditional management' including a damaged sward).

4.2.2. Small staphylinid beetles Although detritivores were not considered in our hypothesis with regard to the E.I., observations suggested that small staphylinids might be an important functional group in field margins. Small staphylinid numbers were positively associated with 'traditional management'. This observation suggests that staphylinids could be supported by the shredded and easily decomposable litter of frequent cuts. In a field margin study in Zuidelijk Flevoland, staphylinid density increased in the grass margins with 4 cuts per year as compared to crop 46 edges (Remmelzwaal & Voslamber, 1996). Accordingly, Canters et al. (1997) found that staphylinids related positively to mowing frequency. Preference of small staphylinids to crop as compared to E.I., observed in the 'prototypes' category (late mowing) was in accordance with observations of Holland & Thomas (1996).

4.2.3. Predacious beetles In improved E.I. in early summer, predacious beetles may be a relatively small component of the arthropod community, because potential prey is or resting higher in the vegetation canopy. High densities of vegetation dwelling arthropods were found in improved E.I. by means of vacuum sampling (Smeding et al., 2001b). Similar observations with regard to predacious beetles were done by Haysom et al. (1999) in grassed headland; lowest densities of carabids were observed in the uncut and 1cu t per year sites as compared to 3 cuts per year, field centre and unmanaged boundary sites. However, the highest predacious beetles abundance were observed in tall and species poor vegetation. Interpretation of this abundance may require a consideration of crop area influence.

4.2.4. Spiders Linyphiid spider abundance in E.I. was difficult to interpret in relation to E.I. characteristics and probably requires consideration of interaction with the crop area density. Lycosid spiders prefer sites with a warm microclimate (Remmelzwaal & Voslamber, 1996); this may explain the observed positive relation to 'traditional management'.

4.2.5. Ground dwelling herbivores Granivorous carabids and folivores showed positive association with the 'traditional management' of frequent mowing without litter removal. With regard to the carabids this might be explained by the occurence of annual species in open locally disturbed swards; these plants have short life cycles, may have ascendant phenotypes and can provide seeds early in the summer. With regard to folivores more frequent cutting may improve the food availability near the soil surface, because the vegetative stage of the vegetation is sustained by the inhibition of flowering. Observations by Haughton et al. (1999), involving around 85%bug s and hoppers (Hemiptera), showed a devastating effect on arthropod numbers by 1cu t per year in the summer as compared to 2 cuts per year in spring and late summer. In our study such a devastating effect was not found, however folivore numbers captured by pitfalls indeed showed a positive association with frequent mowing.

4.3. Comparison of Ecological Infrastructure and the crop area

4.3.1. Differences between E.I. and crop area According to the hypothesis, lycosid spiders, granivorous carabids and folivorous insects showed a higher abundance in the E.I. than in the crop area. However, abundance of the most numerous functional groups (predacious beetles, linyphiid spiders and small staphylinids) did not significantly decrease towards the crop. In the case of linyphiid spiders the opposite was found, which corresponds to the observations of Holland & Thomas (1996) and could relate to the dispersive abilities discussed above.

The E.I. crop area distribution pattern of linyphiid spiders, granivorous carabids and folivorous insects on the farm-type with species poor and tall E.I. vegetation was significantly different from distribution patterns found on the farm types with other E.I. traits. Mean linyphiid numbers decreased towards the crop centre on this farm type, and mean numbers of granivorous carabids and folivorous insects decreased less on this farm type than in other 47 farm types.Anothe r apparent characteristic of farms with speciespoo r and tall E.I. vegetation was the low abundance of small staphylinids inbot h E.I. and crop area. Apossibl e explanation ofthes e observations on linyphiid spiders,granivorou s carabids, folivorous insects and small staphylinid beetles may involve the abundance of predacious beetles. Predacious beetlesha d thehighes t abundance inth e crop area of farms with species poor and tall E.I.vegetation . This highpredaciou s beetle abundance in the crop areacoul d have interfered with the abundance of small-organism epigeic groups. Accordingly, inth e wholedat a set amarginall y significant negative correlation was found in the crop area between total number/farm of carabid beetles andtota l number/farm ofpoole d linyphiids and small staphylinids (Pearson correlation coefficient =-0.61 , PO.0105). Interaction between predacious beetles and linyphiid spidersma y involve 'intraguild predation' (Sunderland et al, 1996b).Th e abovementione d increased abundance ofherbivore s (i.e.granivorou s carabids and folivorous insects) in cropswit h abundant predacious beetles, could indicate acaus ea s well as aneffec t ofpredaciou s beetle abundance: - herbivorous functional groups indicate the suitability ofth e crop habitat (weeds) for herbivory which involves increased prey availability for the ground dwelling beetles (Booij &Noorlander , 1992); - predacious beetles reducepredatio n pressure onherbivore s by interfering with predators that canmor e effectively hunt inth ecanop y (e.g. spiders) whereas carabids prefer tohun t atth e soil surface (Kromp, 1999).

4.3.2. Indicationsfor dispersive movements between E.I. andcrop area Only lycosid spider distribution suggested that theE.I .coul db e aresourc e for crop area populations,becaus e E.I. s were generally much larger than crop area populations andth e lycosid abundance inE.I .wa s correlated to lycosid abundance inth ecro p area.

High numbers ofpredaciou s beetles inth e E.I.wer e also associated with highnumber s inth e crop area.Accordingly , inth ewhol e data set asignifican t positive correlation is found between totalnumbe r per farm of carabid beetles in the crop area andtota l numberpe r farm inth e E.I. (Pearson correlation coefficient = 0.72, P<0.05).However , highnumber s of predacious beetles inbot h E.I. and crop area, could bebette r explained by crop areatrait stha n by E.I.traits :th egrou p of farms (farms A,F ,E )wit h species poor and tall E.I. vegetation overlapped withth e group of farms with arelativel y 'extensive production' inth ecro p area (farms F,E) .Fo r this latter category it couldb e arguedtha t predacious beetles were enhanced inth e crop area. Whereas for the former category explanation was difficult because theE.I . represented an intermediate E.I.typ epossessin g avegetatio n cover orplan t species number similar toothe r E.I. types.A n additional argument for the assertion that crop areatrait s and not E.I.trait sdetermine d thehig hpredaciou s beetle number, is that crop area size exceeds the E.I. size andma y therefore include a larger amount of food for epigeic predators than theE.I . Sincepredaciou s beetles aremobil e species (Thomas etal, 1997),the y may easily dwell from crop area to E.I. andvice versa.

Observations of linyphiid spiders,a mobil e species,als o suggested that abundance inth eE.I . might be determined by crop area abundance. High abundance in the crop area was associated with high abundance in the E.I. However, thehighes t abundance inth e E.I. (speciespoo r with tallvegetation ) wasno t related to ahig h crop area abundance. This divergent situation might havebee n due tohig h carabid abundance in the crop area, as explained above.

With regard to the distribution of small staphylinids inE.I . and crop area, observations didno t indicate dispersive movement. Apparently high levels inbot h E.I. and in crop areawer e associated with low levels in adjacent habitat (crop area or E.I. respectively).Hig h abundance

48 in each habitat werepossibl y related to localhabita t factors, as suggested earlier inthi s paper: flailmowin g in E.I. and intensive crop management inth e crop area. Sinceth eimprove d E.I.vegetatio n didno tharbou rhig h abundance ofepigei c functional groups,i t could notb e identified as aresourc e for epigeic arthropods in the crop area.Thi s observation and also indications of effects on abundance inth eE.I .b ypopulation s inth ecro p area (withregar d topredaciou sbeetle s andlinyphii d spiders),ar e seemingly contradicting reports on enhancing effects on arthropods of (improved) E.I. vegetation. However, these reports areofte n dealing with winter shelter and subsequent dispersion of arthropods in spring (references reviewed inD e Snoo &Chaney , 1999).I tmigh t bepossibl e that enhancing effects of E.I.canno t be observed inth e summer, because epigeic arthropods have already left their shelter. Subsequently theirnumbe r atth efarm-leve l might bedetermine d by crop area resources of initially detrital and latermixe d detrital/herbivore origin. However availability of shelterprovide d by E.I.durin gwinte r and spring may still affect thetota l farm-level numbers (Lys, 1994;Ly s &Nentwig , 1992).Thi s expectation was supportedb y observations (in the discriminant analyses above) ofth epositiv e effect of spatial density ofE.I . on predacious beetles. However, this effect may notb e observable in acompariso n ofE.I . and crop area numbers.

4.4. Conclusions

Comparison ofepigei c arthropod functional groupso n eightorgani c arable farms suggests that the 'intensity' ofth ecro p areai s amajo r determining factor. The concept of'intensity' should not only addresspercentag e gramineous cropsbu t also, for example,manurin g and soilcultivation . Length oforgani c duration by itself mayprovid e no clear basis for distinctions between the farms because ofdifference s in 'intensity' among the long-duration farms. Interpretation offield observation s deliniates threemor eo rles sdistinc t functional group compositions inth e crop area: - relatively high abundance ofpredaciou s beetles (mainly carabids), lycosid spiders, granivorousbeetle s and folivores incro p areaswit hcomparativel y low 'intensity'; - relatively high abundance of small (fungivorous) staphylinids in crop areas with a comparatively high 'intensity'; - relatively high abundance of linyphiid spiders in crop areaswhic hrepresen t an earlier successional stage duet o recent conversion ort o landscape factors.

The area and quality of Ecological Infrastructure provided no clear basis for distinctions between the farms. Summer abundance inE.I . ofdominan t ground dwelling groups, predacious beetles andlinyphiids ,migh t bedetermine db y crop area conditions. Traditional management of E.I., that was expected tob e adverse, showed tohav e aqualitativel y different epigeic community andma ypositivel y support lycosid spiders, small staphylinids and epigeic herbivores as compared to othermanagement . Potential increase of arthropod abundance in latemow n E.I.ma y not involve ground dwelling arthropods but involve vegetation dwelling arthropods thatoccu r in ahighe r stratumo fth e vegetation.

Increased abundance ofpredaciou s species isno t equal toth e capacity ofth ecroppin g system to controlherbivorou s pests (e.g. Wood& Lenne , 1999).Lo w epigeicpredato r abundance (as observed inth e 'prototypes') could indicate both asufficientl y balanced system, inwhic h predators inhibit initial pest development, aswel l asa pes t outbreak susceptible system. Conversely highpredato r abundance maypositivel y correlate withhig h herbivorous pest densities because thesepredator s might havebee n supported by abundant prey. High numbers ofpredator s may indeed represent pest controlpotentia l ifthe ywoul d develop on detritivorous prey (Wiseet al, 1999) ornon-pes t herbivorous prey (e.g. Andow, 1988; 49 Altieri, 1994) and subsequently depresspes t populations. Tobrin g clarity to this subject further investigations areneeded .

4.5. Recommendations

Comparative studieso ngroun d dwelling arthropod functional groupswoul d benefit from a further restriction of geographical variation between farms. Definition of'intensity' should be based onmor e farm characteristics than only crop rotation intensity. Our explorative studies might berepeate d with emphasis onthre e extreme farm types:recentl y (1-2 years) converted farms, longorgani c duration (>8years ) and 'intensive'cropping , and long organic duration extensive cropping {e.g. >35%gramineou s crops, including leys).Becaus e the latter category isincreasingl y rare incommercia l organic farming, farm experiments may be needed, particularlywhe n investigating the effects of increased detrital input as suggested by Wiseet al.(1999) .

More specific research themesma y address; - the ecology ofth e aggregate functional group of saprophorous arthropods including staphylinids, Diptera, and Collumbola(e.g. Weber etal., 1997 ; Idingeret al., 1997) ; - the importance of intraguild predation in crops (e.g. Tscharntke, 1997); - the effect of different mowing regimes on arthropod communities (e.g. Haughton etal. , 1999),includin g vegetation dwelling functional groups aswel l as effects ofpredatio n pressure inth e E.I. duet o epigeicpredator s supported by rich resources inth e crop area.

Additional attention should bepai d to theofte n abundant but patchily distributed ants (Formica), because this species may compete with otherpredator s and mayprotec t aphids and related species infield margins .Thes e studies should extend to a larger part ofth e farm food web,whic h might benecessar y for soundlybase d practical recommendations with regard to enhancing farm biodiversity and itspotentia l ecosystem services. Although nofirm conclusions can be drawn yet from thisfield study , wethin k that results indicate future perspectives of afoo d web approach on farms.

Acknowledgements

The authorswis h tothan k the farmers who allowed us access to their land. Thanks aredu et o AndreMaasse n (Unifarm-WUR) for technical assistance,Bertu s Meijer, Bart Venhorst and JanNoorlande r (PRI) for arthropod measurements andArien a van Bruggen for her statistical help.Th ewor k at Wageningen University isfunde d by theMinistr y of Agriculture,Natur e Management and Fisheries,Departmen t of Science andKnowledg e Dissemination.

50 Chapter 4: Insectivorous breeding birds on organic farms in relation tocro p area, ecological infrastructure and arthropods

F.W. Smeding3& C.J.H . Booij" "Departmentof Plant Sciences, Biological Farming Systems group, Wageningen University, Wageningen, TheNetherlands Plant Research International, Wageningen, The Netherlands

Abstract Hypotheses on the relation of arthropod functional group abundance, crop area and ecological infrastructure (E.I.) variables to the territory density of functional groups of insectivorous birds were tested on organic arable farms. Bird territories were mapped on organic arable farms (eight in 1998,whe n epigeic arthropod were also studied, andnin e in 1999,whe n vegetation dwelling arthropod studies were included). The farms were all situated in an open landscape dominated by agricultural ; skylark, linnet and meadow pippit were the most numerous species. Most birds were more strongly related to vegetation dwelling arthropods in 1999, particularly those in the E.I., than to ground dwelling arthropods in 1998.Th e density of bird territories could usually be predicted by a combination of crop area and E.I. variables, including positive relations to improved E.I. characteristics. In 1999, according to the hypotheses, positive relations to organic farming duration were found; however in both years a general negative relation to extensive crop rotation was found. It will be discussed as to whether intensively managed crop areas may support birds due to the periodic and localised herbivore increases in a crop mosaic, but also whether extensively managed crops might offer less prey items due to the herbivore control exercised by epigeicpredator s as well as a less susceptible cropphysiology . Late mown E.I. might possibly offer an important food resource that may supplement the, often interrupted, food resources of the crop area.

1. Introduction

1.1. Biodiversityloss in agriculture

A serious decline ofth epopulatio n invariou splan t and animal species on more intensively farmed land in industrialised countries,ha sbee n shown by, for example Fuller etal. (1995 ) and Andreasen etal. (1996) . This decline concerns not only ecologists and agriculturists,bu t also thewide rpubli c (Paoletti, 1999;Woo d &Lenne , 1999).Th e development ofmulti - species or sustainable farming systems isprobabl y needed to reverse thistren d (Vandermeer etal, 1998;Altieri , 1994, 1999; Almekinders etal, 1995). Such farming systems utilise ecosystem servicesprovide d bybiodiversit y (e.g. pest control, soil fertility), and allowa greater chance for nature conservation centered biodiversity. Several examples ofmulti - species farming system practices are givenb y Altieri (1994),Woo d &Lenn e (1999) and other authors.

Development ofmulti-specie s farming systems mayrequir e specific research approaches. Firstly the research should take site-specificity ofbiodiversit y into account (Vandermeeret al.,1998 ) and should preferably start from 'nearly natural systems' (Brown, 1999a). Secondly the variables, should specifically address the farm-, orhighe r levels of aggregation (Almekinders etal., 1995).Thirdl y thepredicto r variables should clearly relate to decisions in the farm management (Vereijken, 1997;Kabourakis , 1998;Smedin g &Joenje , 1999;Par k& Cousins, 1995).

51 1.2.Organic farms andecological infrastructure

Organic farms that adheret o theproductio n guidelines 2092/91 ofth e EC (Anonymous, 1991) are suitable research objects asth e maintenance and utilisation ofbiodiversit y is inherent to thephilosoph y of organic farming. Research showstha t organic farming when compared to conventional farming, involves agreate r number and diversity of species and habitats for birds, vegetation dwelling arthropods, epigeic arthropods, earthworms and arableweed s(e.g. Kromp &Meindl , 1997;Paoletti , 1999;Va n Elsen,2000 ; Smeding, 1993;Hendrik s et al, 2000).

Because ofthei r increased similarity to 'nearly natural systems' farms that combine organic cropping with adeliberatel y enlarged, spatially diversified ecological infrastructure areo f particular interest (Brown, 1999a). Inthi s article ecological infrastructure (E.I.) isdefine d as the total of semi-natural (aquatic,woody ,tal l herb and grassy) habitats onth e farm including their spatial arrangement (Smeding &Booij , 1999).Enlarge d and diversified E.I. enhances thenumber s and diversity of indigenous plants and animals atth e farm level (e.g. Joenje et al, 1997;Boatma n etal, 1999).Thi s could contribute to soil fertility andpes t control on organic farms (Brown, 1999b;Krom p &Meindl , 1997;Theunisse n &Kohl , 1999).

1.3. Food web approach

Recent advances inresearc h of food webs (Polis& Winemiller , 1996;Pimm , 1991)offe r an appropriate methodology for biodiversity research, addressing variables athighe r levelso f aggregation. These advances, including some studies in agricultural habitats (e.g. DeRuite ret al., 1995;Tscharntke , 1997),ar eprovidin g inspiring examples ofho w (agro) ecosystems can be approached athighe r integration levels than the species level. Some ofthi s research indicates that farm food webs can mediate the effects of environmental factors on species (Wise etal., 1999;Tscharntke , 1997).

Our general hypothesis about the farm food web on organic arable farms with improved E.I. isbase d onthre e sub-hypotheses: Firstly, increased organic matter inputsfrom organi c manure, crop residues, cereals, ley pastures, and compost, relatet o increased numbers ofinvertebrat e detritivorous meso-an d macrofauna (e.g. Pfiffner &Mader , 1997;Webe r etal, 1997;Idinge r etal., 1996;Heimbac h & Garbe, 1996).Secondly , thenumber s ofvariou snon-pes t herbivorous functional groups are increased by improved E.I. (Holland &Fahrig , 2000;Boatma n etal, 1999)bu t alsob y organic crop characteristics (Hald &Redderson , 1990;Moreb y & Sotherton, 1997).Thirdly , combined increases of detrital andherbivor e invertebrates relate to acumulativ e positive effect on the higher , including both invertebrate andvertebrat e predators, which feed onpre yfrom bot h detrital and herbivore subsystems (e.g. Wise etal, 1999; Idingeret al, 1996).

Iti s likely that succession to ahypothetica l farm food web requires time (Idinger etal, 1997; Edwards etal, 1999),possibl y at least the length of oneo rtw o crop rotations. The succession speed mayb enegativel y influenced bypreviou s applications ofpesticide s and the depletion oforgani cmatte r caused by reliance on artificial fertilisers (e.g. Paoletti, 1999).

52 1.4. Insectivorousbirds and arthropod prey availability

The current field study was focussed onth e abundance of insect-feeding birds inrelatio n to crop area, E.I. characteristics and abundance of arthropod functional groups. Associated studies,t ob epublishe d elsewhere,wer edevote d to lowertrophi c levelso fth e farm food web, including both vegetation and ground dwelling arthropods (Smeding etai, 2001a,b). The ofinsect-feedin g birds represents ahighe r trophic level that may be enhanced by productivity atth ebas e ofth e food web (according toth econcep t ofOksane n etai, 1996), andtherefor e reflects the 'ecological ' of a farming system (Braae et ai, 1988).Birds ,an d their prey organisms, areinfluence d by farm characteristics. Birds are therefore affected both directly and indirectly by farm traits.Thes e indirect effects onbird s areprobabl y mediated through the food web onth e farm (e.g. Tscharntke, 1997).

Bird abundance on farms isno t only determined by food but also by other environmental factors (ascomprehensivel y outlined by Andrewartha &Birch , 1984).Importan t physical factors acting on farmland birds mayb e andnes t destruction bymachine s or trampling (e.g. Green, 1980),availabilit y of songpost s ornes t sites (Stoat, 1999),weathe r conditions orinaccessibl e vegetation (e.g. Schekkerman, 1997). Spatial features like field size, crop diversity and rotation mayno tnecessaril y relatet o food distribution. Howevermos t studies on farmland birds indicate that food availability is amajo r factor determining bird density (reviewed by Poulsen etai, 1998).Bir d chicks areparticularl y vulnerable to shortage ofprotein s supplied by their invertebrate prey (e.g. Hald &Reddersen , 1990;Aebischer , 1991;Poulse n etal, 1998).

Comparative studiesbetwee n organic and conventional farms (Braae etal., 1988;Hal d & Reddersen, 1991; Wilson etal., 1997;Chamberlai n etal., 1999;Moreb y & Sotherton, 1997) emphasized the effects ofpesticid e usageo n food availability, because pesticides depressed bothpre y andhos t plants ofpre y onconventiona l farms. Contributions ofothe r farm characteristics affecting invertebrates were difficult to assessbecaus e most studied factors were correlated with pesticide application. However, Braae etal. (1988 ) traced apositiv e effect ofmanure/fertilize r on 8o f 35numerousl y occurringbir d species;Hal d &Redderse n (1990) revealed thepossibl e importance ofhos tplant s ofFabaceae, Cruciferae, Polygonaceae, which arecommo n on organic farms ascro p orweed . Chamberlain etal. (1999)detecte d effects ofdifferen t landscape structures, which might be more relevant to organic farming practices (Hendrikset al, 2000) onth ebir d community.

Traditional farming systems may reveal theke y position of food resources for rich wildlife (e.g.o fdehesas ;Edward s etal, 1999); inTh eNetherland s atraditiona l (i.e.orthodo x biodynamic) mixed farm of 8ha ,possesse d territories of endangered bird species, red-backed shrike (Lanius collurio)an d ortolanbuntin g (Emberiza hortulana),probabl y duet o increase ofspecifi c preyitem s (Esselink etai, 1996).

Autecological research on farmland birds greatly contributed to deepening the understanding of therelatio n between crop management and chick food availability. Inparticular , work on greypartridg e (Perdrixperdrix) (e.g. Rands, 1985) and skylark (Alaudaarvensis) (e.g. Wilson etai, 1997;Poulse n etai, 1998;Wakeham-Dawso n etal, 1998) involved detailed investigations ofhabita t selection, nesting success,die t and chick condition in relation to farm characteristics and arthropod abundance. These studies were able to relate starvation ofchick s and low territory densities to apparant food shortage in thebreedin g habitat (e.g. Poulsenet al, 1998).

53 Arthropod functional groups, central toou rresearch , includepre yitem s for birds,bu t many taxama yb e insignificant because ofsize , digestibility, repellant taste orothe r characteristics as summarized by the concept 'availability'. Consequently farmland birdshav e distinctive preferences. In our explorative studyw e choset o relate insectivorous bird-density toth e abundance ofentir e arthropod functional groups. Contrary toe.g. Moreby etal. (1994 ) and Poulsen etal. (1998 )wh o selected 'chick-food' within arthropod samples, according todiet s ofconsidere d birds. Ourresearch , therefore, doesno t discern between direct feeding relations and shared indirect association with certain habitat factors. Distinguishing different bird functional groupswit hknow n specific food andhabita t requirements improves thebasi s for interpretation with regard topossibl e causal factors. However, thisresearc h was primarily concerned with relatingbir d functional group abundance (determined by its specific food with emphasis on chick diets) to arthropod functional groups (indicating the farm food web structure) in anattemp t to link autecological research (e.g. Poulsen etal, 1998)t o farming systemsresearc h (e.g.Swif t &Anderson , 1993; Vereijken, 1997;Wolfert , 1997;Va nKeule n etal, 1998).

7.5.Hypotheses and objectives

Five functional groups ofbird s feeding on invertebrates are examined. The distinction is based ondifferenc e in adult diet (insect, mixed insect-plant/seed or soil life) and habitat preference (field or field margin). For each group, specific hypotheses were derived from the above-explained general hypothesis withregar d toth e farm food web.

1.5.1. Insecti-granivorous birdswith field affinity Thesebird s feed on the ground onplants .The ywer e expected tob e supported by ground- and vegetation dwelling diurnal predators andherbivore s inth e crop area. Farm management features which were expected to enhance thepopulatio n ofthes e birdswer e areduce d number ofmachin e operations (which disturb their nests) and the occurrence of cereal crops and weeds which supply arthropods and seeds.Therefor e extensive crop rotations were expected to favour this group. This group isrepresente d by the skylark (Alaudaarvensis) whic h is knownt o feed itschick swit h large spiders,carabi d beetles and large sized herbivores (sawfly and lepidopteran larvae,beetles) . Young chicks receivebot h small and large soft-bodied insects,fo r example herbivorous larvae, and older chicksreceiv e more large hard-bodied insects, for examplebeetle s (e.g. Wilsonet al, 1997;Poulse n etal, 1988). Skylarks are particularly attracted by grass leys (set-aside) and spring cereals. Theyprefe r open cropst o forage, and avoid densecrops .

1.5.2. Insecti-granivorous birdswith E.I. affinity Thesebird swer e expected to look for their animal food inth evicinit y ofthei r seed resources atth e soil surface, onplant s inth eE.I . andi n crop edges.Her ethes ebird s find ground living andvegetatio n dwelling diurnal predators andherbivores . Farm management features which were expected to enhance this groupwer eth e amount of E.I.o nth e farm, the timing ofE.I . cutting andth evariatio n ofE.I . vegetation, including fruiting dicots.Thes e factors would mainly influence their food availability sincenest s areusuall y not located in herbaceous vegetation. This group ismainl y represented by the linnet (Carduelis cannabina), which feeds itschick swit h invertebrates to avaryin g extent, captured on the ground aswel l aso nplants ; chick diets include beetles, spiders,fl y adult and larvae,an d caterpillars, but also small herbivores like aphids (Cramp, 1988).

54 7.5.3. Specialistinsectivorous birdswithfield/E.I. affinity Thesebird s are able to capture flying insects. Therefore their density was expected to relatet o thenumbe r ofvegetatio n dwelling diptera, which dominate the airborne arthropod community; these diptera aremainl y adults ofdetritivorou s larvae and (root)herbivorous larvae {e.g. Tipulasp.) .Probabl y to alesse r extent, the group would be supported bynumber s ofvegetatio n dwelling predators {e.g. syrphids) and folivorous herbivores. Farm management factors that were expected to favour field-inhabiting insectivorous birds would be organic duration and extensive croprotatio n because these factors werepresume d to support the detrital food web.Als o the amount andqualit y ofE.I .wa s expected to influence field- inhabiting insectivorous bird numbers,becaus e adultso fman y flying insects from various trophic groups are attracted by flowers inth eE.I . This group isrepresente d byth e Motacillidaespecie sincludin gwagtail s {Motacilla sp.)an dmeado wpipi t {Anthuspratensis). Wagtails areknow n to feed on flies andmidges , including small individuals from 1-2 mm, but also onbeetle s (carabids aswel l asherbivorou s species) (Cramp, 1988).Meado w pipit adults seem to feed on smallpredominantl y dipteran insects (<5mm ) (Cramp, 1988) and are often seent o forage onth e ground beneath crop canopies.Howeve r chicks are often tob e fed with large (>1cm ) soft larvae (Cramp, 1988),tha t are generally herbivorous. Pipits on arable farms mostly occur infield margin s (Scharenburg etal, 1990).

1.5.4. Specialistinsectivorous birdswith E.I. affinity Thesebird s feed on vegetation dwelling arthropods ofvariou s trophic groups inth e E.I. and are expected toprefe r quantity (height and cover) above quality {e.g. speciesrichness ) of E.I. vegetation. The amount ofE.I . (boundary width, area and spatial density) might contribute to thedensit y ofthes ebird s(Stoat , 1999).Thi sgrou p ismainl yrepresente d by reed warbler {Acrocephalus scirpaceus) andree dbuntin g {Emebriza schoeniclus).I nnorther n clay districts in TheNetherlands , dominated by arable farming, abundance ofbot hbird s isclearl yrelate d to the length ofditche s with old (>1year ) reed stands (Scharenburg etal, 1990).Ditche s in Flevoland with amowin g regimepromotin g reed growth were rapidly colonized by reed warbler and also marshwarble r (Remmelzwaal &Voslamber , 1996).Diet s include awid e rangeo f arthropod taxatha t canb e found within thisvegetatio n (Cramp, 1988).

1.5.5. Soillife-feeding birds Prey items of soil-life feeding birds include mainly soilmacr ofauna , including detritivores (earthworms) androo t herbivorous insect larvae (particularly Tipulasp. ) and additionally epigeic arthropods (carabid beetles and spiders).Ope n fields without vegetation orwit h sparse orver y lowvegetatio nprovid e best accesst o soil life. However, fields that remain open during spring and early summer, receive intensive farming procedures that disturb nesting. The group ismainl y represented by Lapwing {Vanellus vanellus).Foo d availability for lapwing chicks isofte n critical because of agricultural practices. Dry soil isharde rt o penetrate and limits the accesst o soilmacr ofauna . (Schekkerman, 1997;Beintem a et al, 1991,1995).

1.5.6. Insecti-granivores versusspecialist With regard to food requirements ofcombine d groups of insecti-granivorous and specialist insectivorous birds,ou rhypothesi s wastha t the insecti-granivorous group will show acleare r relationship with ground dwellling arthropods (sampled by pitfall traps) than specialistbirds . Furthermore,w e expected that vegetation dwelling predators would be the most important common food resource ofth e four functional bird groupsbecaus e of their generally larger size. Total bird density isexpecte d tob erelate d both tothi s group ofvegetatio n dwelling predators aswel l ast o farm characteristics that favour detrital web numbers {i.e.duration , extensive crop rotation) and flowers/seeds inth e ecological infrastructure.

55 1.5.7. Landscape effects Birdsma yb e affected by differences inth e terrain. For example, inth enorther n clay district in TheNetherlands , skylark, blue-headed wagtail,meado w pipit and quail (Coturnix coturnix) prefer open terrain. Marsh bird abundance is favoured by continuous reed habitat of 1.5-3 km length (Scharenburg etal, 1990).Therefor e also landscape differences between farms were taken into account.

1.5.8. Objectives Thus, the objectives ofthi s studywere : 1. to determine theterritor y density of insectivorous birds of organic farms, and relatethi s density to the abundance of arthropod functional groups; 2. torelat eth eterritor y density ofinsectivorou s birds oforgani c farms to the extent ofon - farm ecological infrastructure, andvegetatio n and crop characteristics.

w

20 Kilometers

Fig. 1.Study area in 1998-1999:five farms with bird and arthropod studies in 1998as well as 1999are represented bysquares; onefarm with bird and arthropod studies only in 1998 isrepresented by a cross; two farms with bird and arthropod studies in 1998and only birdstudies in 1999are represented by circles; two farms with bird and arthropod studies in 1999are represented by triangles.

56 2. Materials and methods

2.1. Study area

The study was carried out on organic farms inth eprovinc e ofFlevoland , TheNetherlands . The farms are located inth ethre epolder sNoordoostpolder , Oostelijk Flevoland and Zuidelijk Flevoland (Fig. 1).Thes ethre epolder s were reclaimed around 1940, 1957 and 1968, respectively, and are dominated by agricultural land use.Th e landscape isver y flat and open with wood lots confined to farmyards, villages andmai n .Th e soil is acalcareou s clay- sandy clayo fmarin eorigin . Flevolandwa schose n asou r study areabecaus e itha sa concentration oforgani c farms that are similar with respect to their history and topography. There arecurrentl y 75 economically viable organic arable farms inFlevoland , which are generally large (c. 30-70 ha) and intensive relative to other organic arable farms inEurope .

TableI Characteristics of thefarms, ecological infrastructure and landscape (200 ha, including thestudied farm).

1998 Level Variable Abbrev. A N T F K Z R L E D Farm n year organic farm (n) DUR 12 12 8 7 1 10 4 8 Farm Cereals and ley pasture (%) INT 38 59 38 24 18 11 70 35 Farm % of crop area within 100 m range E.I. (%) EI10 0 47 40 74 74 51 80 90 72 Farm Semi-natural habitat (E.I.) (%) AREA 2 1 2 1 2 3 5 4

E.l. Mowing date (farm level) (month) MOW 7 5 9 5 5 6 6 7 E.I. Mowing frequency (n/year) MOWFRQ 2 3 1 2 3 2 2 2 E.I. Height vegetation (m) EIGHT 1.8 1.1 1.8 1.0 0.7 0.7 1.2 1.1 E.I. Cover vegetation (%) EICOV 98 82 97 88 85 92 97 94 E.I. Species diversity vegetation (n/4 m2) EISPN 12 8 11 9 6 10 10 15 E.I. Dominant species diversity (n/4 m2) EIDOM 3.4 2.8 3.4 3.1 2.3 3.0 3.4 3.8

Landscape Age polder (year) LSAGE 30 30 41 41 41 41 41 57 Landscape Semi-natural habitat landscape (%) LSNAT 4 8 5 5 5 4 7 4 Landscape Cereals/ley pasture landscape (%) LSINT 34 44 32 26 27 23 51 27 Landscape Vegetables landscape (%) LSVEG 38 23 17 16 29 22 21 28

1999 Farm n year organic farm (n) DUR 13 13 9 8 16 1 2 11 5 Farm Cereals and ley pasture (%) INT 2 24 32 42 33 18 12 20 72 Farm Weed abundance (class) WABUN 0.4 2.3 1.2 3.9 1.8 1.3 2.3 2.3 2.4 Farm % of crop area within 100 m range E.I. (%) Ell0 0 47 40 74 74 72 66 80 80 90 Farm Semi-natural habitat (E.I.) (%) AREA 2 1 2 1 2 1 1 3 5

E.I. Mowing date (farm level) (month) MOW 7 5 9 5 5 5 5 7 6 E.I. Mowing frequency (n/year) MOWFRQ 2 3 1 3 3 2 3 2 2 E.I. Height vegetation (m) EIGHT 1.8 1.1 1.8 1.0 0.7 0.4 0.7 0.7 1.2 E.I. Cover vegetation (%) EICOV 98 82 97 88 84 82 85 92 97 E.I. Species diversity vegetation (n/4 m2) EISPN 12 8 U 9 9 8 6 10 10 E.I. Dominant species diversity (n/4 m2) EIDOM 3.4 2.8 3.4 3.1 2.8 2.5 2.3 3.0 3.4

Landscape Age polder (year) LSAGE 31 31 42 42 42 42 42 42 42

Ten farms were selected based ondifference s in farm structure and ecological infrastructure (E.I.)trait s (Table 1).Th e selection included ninecommercia l organic farms and one experimental farm. The selected commercial farms are all arable although three (farms N,T , and K)posses s astabl ewit h livestock fed on fodder crops from thesam e farm but grazing 57 elsewhere during the summer. The organic experimental farm of Wageningen University isa mixed dairy and arable farm. Four commercial farms (A,T , Lan d D)wer e Ecological Arable Farming Systems (EAFS)prototyp e farms during 1992-1997(Vereijken , 1997, 1998).Amon g these farms, farm Ai sdifferen t because it is situated inth e youngest polder Zuidelijk Flevoland, andjoine d theEAFS-projec t a few years latertha n the others. Farms A andN started in 1985 on an experimental reclamation area wherepesticide s had neverbee n used (Remmelzwaal, 1992). Bird and arthropod studies in 1998 involved 8farm s (total 542 ha) inthre epolder s (Fig. 1). Bird and arthropod studies in 1999involve d 7 farms (total 392 ha) that were confined toth e polder Oostelijk Flevoland. Ontw o farms inpolde r Zuidelijk Flevoland (farms A andN ; 216 ha study area)onl ybir d studies weredon e in 1999.

2.1.1. Crop area characteristics Theorgani c duration ofth e selected farms ranged from one to sixteen years (Table 1). Ploughing isgenerall y at 25-30 cm depth. On all farms a6- 7 year crop rotation is maintained, with 6-14 different cropspe r farm. Theproportion s ofdifferen t types ofcro p per farm ranged inbot h years from 10-70% for gramineous crop (cereals and leypastures) , 10-28% for traditional lifted crops (potato,beet ) and 20-11% for vegetables (mainly onions,peas , sweet corn, various cabbages, red beet andrunne rbeans) .

The farms with themos t intensive crop rotations had upt o 60%vegetable s and therefore a lowpercentag e of gramineous crops (cereals, leypasture) . Ascompare d to other crops, gramineous cropsprovid e a large amount of carbon in crop residues and require fewer procedures that cause soildisturbance . Because input ofnutrient s (N,P,K) hast o be adjusted to crop requirements, andvegetable s need more nutrients than cereals, intensity ofcro p rotation isassume d torelat et o both the amount ofcro p residues and the amount ofN - application atth e farm level.

2.1.2.Ecological infrastructure characteristics Canals andbank s (together c. 3-4 mwidth ) that are laid out as linear herbaceous boundaries adjacent to the cropsmainl y determine the ecological infrastructure (E.I.) of arable farms in Flevoland. Onfiv e farms, boundaries arewide r because of adjacent (2-3 mwide ) grass strips, often including clovers (Trifolium spp.),o non e orbot h sides of thecanals . Farm Eals oha d grass stripsbetwee n cropplots .Percentage s of E.I.highe r than c. 2%(Tabl e 1)wer e mainly duet o thepresenc e ofthes e grass strips.Onl y farms K,L and Epossesse d aboundar y planted with ahedgerow . Spatial density ofE.I . ispurposivel y increased on farm E(Smedin g& Joenje, 1999) and farm L;th e spatial arrangement of E.I.o nth e other 6 farms mainly reflects landscape-scale features.

Extreme types ofvegetatio n management are: - 'traditional management' involving three or four more cutspe r year with aflai l mower leaving the shredded cuttings in situ; - 'latemowin g management' involving one ortw o cutspe ryea r after June 21st with a finger- barmowe r with hiab grab for removal ofcuttings .

The grass strips along the canals are required for transport on farms with a'lat e mowing management'. E.I. area istherefor e related to mowing date.I n this article the E.I.wit h enlarged area and latemowin g is defined as'improve d E.I.1.Th e improved E.I. on four farms (A,T , L, andD )wa s created during the EAFS-prototyping research program in 1992-1997 (Vereijken, 1997).Th e improved E.I. onth e experimental farm Estarte d in 1995 simultaneously with conversion of theconventiona l dairy farm to anorgani c mixed farm (Smeding &Joenje , 1999). 58 Dominant plant species in canalban k swards are:couc h (Elymus repens),perennia l meadow grasses (Festuca rubra, Poa trivialis, Loliumperenne, Agrostis stolonifera), commo n reed (Phragmitis australis)an d few perennial herbs:dandelio n {Taraxacum officinale) and stinging nettle (Urtica dioica).Canal stha t aretraditionall y managed have alow , relatively open, speciespoo r vegetation (<10 species/4 m2),dominate d by afe w grass species.Lat e mowing particularly relates to increased vegetation biomass inJun e and July asexpresse d by vegetationheigh t (upt o 1-3 m) andcove r (upt o90-100%) .However , latemowin g doesno t alwaysrelat et o increased higher plant species diversity for two reasons: a)tal l grass species may dominate the sward;b ) swards that havebee n traditionally managed mayb ecu t locally to ground level and therefore beinvade d by ruderals and annuals (increased species number up to 15species/ 4 m2 ),e.g. thistles (Cirsium arvense, Sonchusarvensis), annua l weeds(e.g. Sonchusasper, Polygonum aviculare) an d annuals that arerarel y perniciousweed s in Flevoland (e.g. Myosotisarvensis, Veronicapersica, Cardamine hirsuta).

Highperennia l plant diversities found in some locations mayb e caused by various interrelated factors (Kleijn, 1997); such asreduce d vegetation productivity, removal ofth e hay,nutrien t buffering by the grass strip,an d long organic duration. Artificial species introduction ishowever , also involved. Inth e EAFS-project, around 90 different native perennial dicotswer e artificially seeded intoth ecana lbank s to obtainhighe r plant diversity and flower abundance ofplant swit h limited dispersal capacity (Vereijken, 1998).Althoug h few species could settlewel l(e.g. Seneciojacobea, Crepis biennis, Heracleum sphondylium), the introduction clearly affected species diversity: vegetation including siteswit h>1 5 species/4m 2wa s confined to prototype farms.

2.2.Sampling and measurements

2.2.1.Birds Territory mapping of skylark wasdon e four times in 1998an d threetime s in 1999,i nth e beginning ofMa y (week 19), around the end ofMa y (week 22), around the middle ofJun e (week 25)and , only in 1998,aroun d thebeginnin g ofJul y (week 28),usin g themetho d described inVa n Dijk (1993).A representativ e part (80-100%) ofth e farm, excluding the farm yard, was selected. Farms werevisite d inth emornin gbetwee n 8:30-12:00h . Observations weredon ewhil e slowlywalkin g alongtranspor t tracks and canalbanks ,withi n a free observation range of 200m t obot h sides ofth etransect . Bird observations wereplotte d on a 1:5000fiel d map,includin g territorial behaviour (singing, fighting, mating) orothe r nesting indications (nest visits, feeding andpresenc e ofchicks) .

Field mapswer e interpreted according to VanDij k (1993):observation s were collected on individual bird species maps. Separate sightings within specific merger distances were attributed to the same territory, unlesstw o individuals were seen simultaneously.

Breedingbird swer e grouped according to food types andhabita t preferences: 1. Insecti-granivorous birdswit h crop areapreference : skylark (Alauda arvensis); 2. Insecti-granivorous birdswit h E.I.preference : linnet (Carduelis cannabina)an d goldfinch (Carduelis carduelis), both finch species (Fringillidae); 3. Specialist insectivorous birdswit h combined crop area and E.I.preference : blue-headed wagtail (Motacillaflava), whit ewagtai l (Motacilla alba)an d meadow pipit(Anthus pratensis) members ofth ebir d family Motacillidae; 4. Specialist insectivorousbird swit hE.I . affinity, defined as 'marshbirds '(accordin g to Remmelzwaal &Voslamber , 1996):ree d warbler (Acrocephalus scirpaceus),ree d bunting (Emberiza schoeniclus) andbluethroa t (Luscinia svecica). 59 5. Insecti-granivorous birds onth e farm, including groups 1.an d 2.mentione d abovean d also a few observations of quail (Coturnix coturnix) which isno t asongbir d but has requirements similar toth efirst group ; 6. Specialist insectivorous birds on the farm, including groups 3.an d4 .mentione d above; 7. Total songbirds including insecti-granivorous birds and (specialist) insectivorous birds; 8. Soil life-feeding birds: lapwing (Vanellus vanellus), oystercatcher (Haematopus ostralegus) and blacktailed godwit (Limosa limosa).

Observations ofswallo w (Hirundo rustica), housemarti n (Delichon urbica),starlin g (Sturnus vulgaris)an dhous e sparrow (Passerdomesticus) wereno t included because these species nest in farmyards and stables and mostly stay around the farmyard or are accidentally present inth efields o fadjacen t farmland. Linnet, goldfinch andwhit ewagtai l also do not nest inth e openfield bu t aremuc h moreresiden t inparticula r parts ofth e arablefields. Observation s of soil life-feeding birdswer e interpreted with caution because themos t common species, lapwing,require s observations inApri l (VanDijk , 1993).

2.2.2. Arthropods Samplingwa sdon e inwhea t crops and onadjacen t canalbank s (E.I.)- Wheat (Triticum aestivum) was chosen asa representativ e cropbecaus e thiscro p iscommo n onorgani c arable farms inFlevolan d and arthropod numbers in cereals are larger compared to numbers inothe r common crops likepotatoe s and onions (Booij &Noorlander , 1992).

In 1998groun d dwelling arthropods were sampled by sixpitfal l trapsplace d ina ro wwit h intervals of 10 m atdistance s of 0m (E.I.) and 30m (wheat).Weekl y sampleswer e taken during 8weeks ,startin g from 6th ofJun eunti l 28th ofJul y 1998.Individual s were sorted to taxaa torder - or family-level andplace d into six different arthropod functional groups (for further details see Smeding etal, 2001a): - Predacious beetles including carnivorous Carabidae and large (>6mm ) Staphylinidae; - Lycosid spiderswhic h are largean d represent diurnal wandering spiders; - Linyphiid spiderswhic h are small andrepresen t sheet-web spiders; - Small staphylinid beetles which represent animportan t detritivorous functional group; - Granivorous herbivores represented by granivorous carabidbeetles ; - Folivorous herbivores represented byweevil s (Curculionidae) andhopper s (Cicadellidae).

The observations ofth e six traps and eight sampling dateswer e accumulated giving onetota l numberpe r functional grouppe rhabita t (crop orE.I. ) per farm.

In 1999vegetatio n dwelling arthropods were collected bymean s of avacuu m sampler (ES 2100, Echo,Lak eZuric h U.S.A.) for 120s a t each 1 m2sample .Nin e samples of 1 m2a t distances of 0m (E.I.) and 30m (wheat)wer e selected. Sampling was doneonc e between June 7than d July 6th.Tax awer eplace d into trophic categories (for further details see Smeding etal, 2001b): - Detritivores; - K- or senescense-feeding herbivores (White, 1978); -r - or flush-feeding herbivores (White, 1978),represente d by aphids (Aphididae) - Predators; - Parasitoids,represente d by Parasitica (Hymenoptera). Theobservation s ofth enin e suction sampleswer e accumulated, giving onetota l number per functional grouppe rhabita t (crop orE.I. ) per farm.

60 2.2.3. Vegetation measurements inthe E.I. Vegetation studies included thebank s of smallcanal s (width <4m )tha t were adjacent toth e crops;investigatio n wasdon efro m May toAugust : latemow n vegetation wasinvestigate d in June andJul ybefor e thefirst cut . Early ando r frequently mown vegetation was investigated before thefirst cu t (May) or later (until August)whe n species composition could sufficiently be identified. Investigation was done once in 1997-1999. Speciesnumbe r were assumed tob e moreo r less constant overth etw oyears .Observation s onvegetatio n height and cover were influenced by sampling date and differences between years;howeve r by investigating many plots including all canalbank s (width <4m )o n the farm, therewa s sufficient representation ofgenera l performance, standing crop and occurrence of gaps inth e sward.

Theher b layer was sampled inplot s of4 m 2are awit h intervals of 50m betwee n plots. Plots included the ditchban k vegetation from the top toth ehelophyti c zone; and vegetation affected by submersion was excluded. Parameters were:cover , maximum height, plant species anddominan t plant species (totalling 80%o fth e cover).Average s were calculated for eachfarm . The database comprised 397plot swit h 30-50plot spe r farm.

2.2.4. Farmmanagement and lay out A 1:5000ma p was made of each farm, depicting the different crops and the ecological infrastructure (E.I.) ando nthes emap sth eare ao fth ecrop san do f semi-natural habitatswer e measured.

Measurements of farm characteristics wererestricte d to arepresentativ e farm areai nwhic h thebir d territories were mapped, which was 80-100%o fth etota l farm area.However , crop rotation intensity (% gramineous crops,includin g leypastur e and cereals) was calculated for thewhol e arable crop areabecaus e thisfigure indicate d both the input of carbon aswel l asa specific habitat within thebir d mapping area.

The 'E.I.area ' iscalculate d asth epercentag e of semi-natural habitat ofth e representative farm area. The spatial density of theE.I .wa s expressed by thepercentag e ofth e farm which lay within a 100m distanc e ofth e E.I. Incas e the canalban k vegetation was cutbot h after September 15th aswel l asbefor e May 15th, adistanc e of 70m wa s considered.

Weed abundance inth e crop areawa s measured inth efield usin g asimplifie d Tansley scale (with 0.5 = few individuals, 1 =rare ,2 = occasional , 4= frequent, 8= abundant; 10= co - dominant);wee d abundance was estimated during half ofJun e in each crop.A weighted averagevalu ewa scalculate d for each farm, based onth eproportion s ofth e cropswithi nth e areawher eth ebir d territories were mapped.

Datao n conversion to organic,operation s inth ewhea t crop in 1998,an d themowin g of canalsbank s and grass stripswa s obtained by using aquestionnaire . Duration is defined asth e number ofyear s since thewhol e crop areao fth e farm was certified according to the European regulations for organic farming (Anonymous, 1991).Th etimin g ofmowin g isrepresente d by themont h of thefirst cu t and the frequency per year.

2.2.5. Landscape Landscape parameters were sampled around 8farm s in 1998,i n anare a of 200h awhic h included the investigated farm. The areaha d arectangula r shapewit h alengt h and width ratio congruent to the farm andpositione d inth e same orientation asth e farm. The farm was situated inth ecentre . Insideth erectangl eth e areao fdifferen t crops, semi-naturalhabitats , roads,pavement s and urban occupation weremeasured . The semi-natural aquatic,woody ,tal l herb and grassy habitats were defined as'natural' . 61 2.3.Statistical analyses

All measurements wereteste d for normality. Arthropod numbers were In-transformed. All subsequent analyses were conducted using Statistical Analysis System (SAS Institute,Cary , North Carolina). Therelativ e importance of arthropods, crop area and E.I. characteristics on bird territory density were determined using multivariate REGMAX R regression. Eight explored response variables were:skylarks ,wagtails/pipit , finches, marshbirds , insectivores, insecti-granivores, total songbirds and soil life-feeding birds. Therewer efou r separate analyses concerning arthropods and farm environmental factors in two different years and one additional analysis concerning the landscape factors. These analyses includeth e following predictorvariables : 1. with regard to ground dwelling arthropods (1998), twelve variables in crop and E.I.: predacious beetles, lycosid spiders,linyphii d spiders,smal l staphylinids, granivorous carabids and folivorous insects; 2. withregar d to vegetation dwelling arthropods (1999), ten variables in crop area andE.I. : r-herbivores, K-herbivores, detritivores,predator s andparasitoids ; 3. withregar d to farm and E.I. characteristics in 1998,te n variables: organic duration, intensity, spatial density ofE.I. , areao fE.I. ,mowin gdate ,mowin g frequency, maximum height, cover,plan t species number anddominan t plant species number; 4. withregar dt o farm andE.I .characteristic s ofseve n farms inOostelij k Flevoland in 1999, the samete n variables as in 1998wer e used and also weed abundance: the same analysis wasperforme d includingtw o more farms inZuidelij k Flevoland because these farms were also involved in studies of 1998bu t not in arthropod studies of 1999tha t were confined to Oostelijk Flevoland; 5. with regard to landscape structure ofeigh t farms in 1998,five variables : age ofth e polder, proportion of semi-natural habitats,proportion s ofvegetabl e crops, lifted crops and gramineous crops.

Inth emultipl eregression s allpredicto r variableswer e included inbot h linear and quadratic form to find both linear and quadratic relations between response and predictor variables. Multiple regression models included maximally 4variable s (Maxr; STOP=4). For each response variable thebes t model was selected according toth e following criteria: - theR-squar e ofth emode l was largertha n50% ; - the significance level of the effects of all individual variables and themode l should be smaller than 0.05; - themode l with thehighes t probability was considered asth ebes t model. But ifth emode l with thehighes t probability, ascompare d toth ebes t model with onevariabl e less,ha d the sameprobabilit y classan d added lesstha n 5%t oth eR-square , than themode lwit h fewer variables was considered as better.

62 3. Results

3.1. Observedbird species and territory densities

In 1998withi n 422h a study area, 169territorie s of insectivorous and insecti-granivorous birdswer e counted and in 1999withi n 467h a study area, 133territorie s were counted. Observations inbot hyear sinclude d 10specie s(Tabl e 2). Additional observations ofsoi l life- feeding birds yielded 39territorie s inbot h years,includin g 3 species (Table2) . Most territories (c. 39%) concerned skylark (Alauda arvensisL.) ,tha t had territory densities ranging from 0.06-0.24pe rh a with average densities of0.1 5 and 0.12 territory perh a in 1998 and 1999respectively . Othernumerou s species were linnet (Cardueliscannabina L.) (c. 18% of observations) and meadow pipet (Anthuspratensis L.) (c. 13% ofobservations) . Soil life- feeding birdswer emainl y (c.90% ) lapwing (Vanellus vanellus L.).

3.2. Birdfunctional groupsin relation to arthropod functional groups

Skylark territory density was found tob epositivel y related to arthropod functional groups (Table 3)i nth e E.I., particularly toth ediurna l predacious groups: lycosid spiders in 1998 (Table4) and vegetation dwellingpredator s in 1999, but alsot o herbivores in 1999(Tabl e5) . With regard to arthropods in the crop area,r-herbivore s were an important predictor in 1999 (Table 5).Multipl e regression selected asbes t 1-variablemodel : ALAU = 0.0003 +0.002 9(rHERB) 2 (R2=0.66; PO.034).

Table2 Observations of bird territories: the territory numbers of different speciesperform, thesize of thestudy area, and thedensities of thefunctional groups. Theassemblages are explained in the method section.

1998 1999

Farm A N T F R L E D A N T F K Z R L E Birds Skylark 4 15 9 9 9 4 4 8 2 12 8 5 7 5 7 5 4 Linnet 3 4 3 4 2 3 5 5 2 5 2 2 6 1 2 5 4 Goldfinch 0 0 1 0 0 0 1 1 0 0 1 0 0 0 0 0 0 Quail 1 4 2 0 2 1 2 0 0 0 0 1 1 0 1 1 1 Blue-headed wagtail 1 2 2 1 1 2 1 2 1 1 1 2 0 0 1 1 1 White wagtail 2 2 1 2 1 1 3 1 2 3 2 1 1 1 2 1 2 Meadow pipit 0 1 2 5 4 7 1 9 0 0 1 2 0 0 0 6 0 Reed warbler 1 0 0 0 0 0 0 0 4 1 2 0 0 0 0 0 0 Bluethroat 1 0 0 0 0 0 0 0 2 0 0 0 1 0 0 0 0 Reed bunting 3 3 1 0 0 0 0 0 0 0 0 0 1 0 0 0 0 Lapwing 0 7 5 4 5 4 2 7 1 8 3 3 5 3 8 4 2 Oystercatcher 0 0 0 0 0 0 3 1 0 0 0 0 0 0 0 0 2 Black-tailed godwit 0 0 0 0 0 0 1 0 0 0 0 0 0 0 0 0 0

Bird mapping area (ha) 39 93 37 47 57 34 68 47 39 93 37 47 45 47 57 34 68

Skylarks n/10 ha 1.0 1.6 2.4 1.9 1.6 1.2 0.6 1.7 0.5 1.3 2.1 1.1 1.6 1.1 1.2 1.5 0.6 Finches n/10 ha 0.8 0.4 1.1 0.9 0.3 0.9 0.9 1.3 0.5 0.5 0.8 0.4 1.3 0.2 0.3 1.5 0.6 Motacillidae n/10 ha 0.8 0.5 1.3 1.7 1.0 2.9 0.7 2.6 0.8 0.4 1.1 1.1 0.2 0.2 0.5 2.3 0.4 Marsh birds n/10 ha 1.3 0.3 0.3 0.0 0.0 0.0 0.0 0.0 1.5 0.1 0.5 0.0 0.4 0.0 0.0 0.0 0.0 Insecti-granivores n/10 ha 2.0 2.5 4.0 2.8 2.3 2.3 1.8 3.0 1.0 1.8 2.9 1.7 3.1 1.3 1.7 3.2 1.3 Insectivores n/10 ha 2.0 0.9 1.6 1.7 1.0 2.9 0.7 2.6 2.3 0.5 1.6 1.1 0.7 0.2 0.5 2.3 0.4 Total songbirds n/10 ha 4.1 3.3 5.6 4.5 3.3 5.3 2.5 5.6 3.3 2.4 4.5 2.8 3.8 1.5 2.3 5.6 1.8

Soil life-feeders total n/10 ha 0.0 0.8 1.3 0.9 0.9 1.2 0.9 1.7 0.3 0.9 0.8 0.6 1.1 0.6 1.4 1.2 0.6

63 Table3 Observations of arthropods perform: datafrom 1998representing total numbers ofground dwelling arthropod functional groups captured in 6pitfall traps in weekly trappings over a eight weekperiod in the wheat crop (30 mfrom thefield margin) as well as in the ecological infrastructure; datafrom 1999representing thetotal number/9 m' perfarm of vegetation dwelling arthropods in the wheat crop (30mfrom the field margin) andin the ecological infrastructure, captured by vacuumsampling.

Crop area 1998 Arthropod functional groups (n/trap) Farms: Ground dwelling 1998 abbrev. A N T F K Z R L E D Predacious beetles in wheat 30 m COL 952 1064 1333 3765 * * 1011 1035 2889 1731 Lycosidae in wheat 30 m LYC 53 146 57 38 * * 18 83 35 8 Linyphiidae inwhea t 30 m LIN 3061 3311 2979 2622 * * 2515 2055 1687 1944 Staphylinidae <6 mm in wheat 30 m STSM 1308 329 893 425 * * 640 772 205 541 Carabidae granivorous in wheat 30 m CARH 5 83 7 4 * * 3 19 504 15 Folivorous insects in wheat 30 m FOLI 4 3 7 21 * * 13 6 8 5

Crop area 1999 Arthropod functional groups (n/9 m2) Farms: Vegetation dwelling 1999 A N T F K z R L E D r-Herbivores total rHERB * * 2379 219 836 743 413 377 189 * K-Herbivores total KHERB * * 20 12 28 15 36 42 58 * Detritivores total DETR * * 19 24 45 23 66 48 63 * Predator total PRED * * 33 31 85 33 93 39 221 * Parasitoids total PAR * * 94 60 68 66 156 152 79 *

EI 1998 Arthropod functional groups (n/trap) Farms: Ground dwelling 1998 A N T F K Z R L E D Predacious beetles in wheat 30 m COL 1365 1319 1205 2808 * * 832 1706 2603 926 Lycosidae in wheat 30 m LYC 87 535 208 302 * * 46 168 226 64 Linyphiidae in wheat 30 m LIN 2508 2577 880 4235 * * 1191 1607 1930 2554 Staphylinidae <6m m in wheat 30 m STSM 739 703 444 334 * * 365 217 411 709 Carabidae granivorous in wheat 30 m CARH 52 149 83 79 * * 253 291 279 66 Folivorous insects in wheat 30 m FOLI 104 275 150 33 * * 38 61 35 100

EI 1999 Arthropod functional groups (n/9 m2) Farms: Vegetation dwelling 1999 A N T F K Z R L E D r-Herbivores total rHERB * * 488 11 45 106 127 59 35 * K-Herbivores total KHERB * * 471 71 24 341 831 509 145 * Detritivores total DETR * * 395 67 43 79 45 440 98 * Predator total PRED * * 254 104 158 158 121 336 96 * Parasitoids total PAR * * 88 14 27 95 49 47 35 *

Finches (mainly linnet) territory density was found tob epositivel y related to vegetation dwelling predators inbot h crop area and E.I. andwa s found tob e negatively related to herbivorous groups in the E.I. (granivorous beetles and K-herbivores in 1998 and 1999 respectively) and also negatively related to linyphiid spiders in the crop area in 1998an d parasitoids in E.I. in 1999(Table s 4 and5) .

Motacillidae (wagtails/pipit) territory density was found tob epositivel y related to lycosid spiders inth e E.I. in 1998 (Table 4) and todetritivore s inth e E.I. in 1999(Tabl e 5).I n 1998 64 Motacillidae werenegativel y related to linyphiid spiders and curvilinearly to granivorous carabids inth ecro p area(Tabl e4) .

Marshbir d territory density was bestpredicte d by small staphylinid beetles inth ecro p areai n 1998 (Table4 ) andb y r-herbivores andparasitoid s inth e E.I. in 1999 (Table 5).

Insecti-granivorous birds territory density (including skylark, linnet, goldfinch and quail) was best predicted by vegetation dwelling predators inth e E.I. (Table 5).N o model with regard to ground dwelling arthropod observations, was obtained in 1998(Tabl e4) .

Insectivorous birdsterritor y density (including blue-headed andwhit ewagtail ,meado wpipit , reed warbler, reedbuntin g andbluethroat ) was found tob e curvilinearly related to detritivores inth eE.I . andt or-herbivore s inth eE.I . (Table 5).Th emos t important predictor was vegetation dwelling detritivores inth e E.I.:Multipl e regression selected asbes t 1-variable model: INSECT =0.0066(DETREI) 2 - 0.0054 (R2=0.71;PO.017) .N o model for insecti­ vorousbird s inrelatio n to ground dwelling arthropod observations was found (Table4) .

Table 4 Multiple regression models (P<0.05) with ground dwelling arthropods as predictor variables based on bird and arthropod observations on eight farms in 1998; Birds (territory/ha): Alau = skylark, Mota = Motacillidae, Fin = Finches, Marsh = Marsh birds, Insect = Specialist insectivorous birds; Insgran= Insecti-granivorous birds; Avis = total songbirds; Soilf = soil life-feeding birds; Ground dwelling arthropods; COL =predacious beetles; L YC = lycosid spiders; LIN = Linyphiid spiders; STSM = small staphylinid beetles; CARH = granivorous carabids; additional symbols: 'ca' = crop area; 'ei' = ecological infrastructure.

Birds Model R" Pr>F Alau -0.063'ln(LYCca)- 0.005«ln(CARHca)2+ 0.010*ln(LYCei)2 -0.004*ln(LINei) 2 +0.43 7 96 0.0162 Fin -0.11*ln(LINca) -0.04*ln(CARHei) 2 +1.061 2 72 0.0401 Mota -0.42*ln(LINca) +O.U'lnfCARHca ) - 0.025*ln(CARHca)2 +0.039*ln(LYCei ) + 3.071 99 0.0011 Marsh -0.96*ln(STSMca) + 0.08*ln(STSMca)2+ 2.839 74 0.0327 Insgran No model Insect No model Avis No model Soilf -0.36'ln(LINca) - 0.03*ln(CARHca) -0.15*ln(COLei ) + 0.1*ln(LYCei) + 3.53 94 0.0390

Table 5 Multiple regression models (P<0.05) with vegetation dwelling arthropods as predictor variables based on bird and arthropod observations on eight farms in 1999; Birds (territory/ha): Alau = skylark, Mota = Motacillidae, Fin = Finches, Marsh = Marsh birds, Insect = Specialist insectivorous birds; Insgran= Insecti-granivorous birds; Avis = total songbirds; Soilf = soil life-feeding birds; Vegetation dwelling arthropods: rHERB = r- herbivores; KHERB= K-herbivores; PRED =predators; DETR = detritivores; PAR =parasitoids; additional symbols: 'ca' = crop area; 'ei' = ecological infrastructure.

Birds Model R* Pr>F Alau 0.063•ln(rHERBca) +0.0014*ln(KHERBei) 2+ 0.038*ln(PREDei) -0.077*ln(PARei ) -0.24 4 100 0.0068 Fin 0.034*ln(PREDca) -0.014*ln(KHERBei )+ 0.139*ln(PREDei) - 0.004*ln(PARei)2- 0.632 100 0.0018 Mota 0.006*ln(DETRei)2 -0.05 9 62 0.0346 Marsh -0.044*ln(PARei) +0.003*ln(rHERBei) 2 +0.07 6 99 0.0002 Insgran 0.147*ln(PREDei)-0.527 62 0.0354 Insect -0.671*ln(DETRei)+ 0.075*ln(DETRei)2 -0.029*ln(rHERBei)2 + 1.703 97 0.0074 Avis 0.366*ln(PREDei) - 0.017*ln(PARei)2 - 1.284 94 0.0035 Soilf 0.0068*ln(PARca)2 -0.154*ln(KHERBei )+ 0.015*ln(KHERBei)2+ 0.33 2 99 0.0011

65 Total songbird territory density wasbes t predicted by vegetation dwelling arthropods inth e E.I. (Table 5) andwa spositivel y related to predators and curvilinearly to parasitoids. Total songbird territory densitywa s also predicted fairly wellb y aone-variabl emodel : AVIS= 0.027(PREDEI) - 0.38 (R2 = 0.71; PO.035). Nomode l for totalbir d density in relation to ground dwelling arthropod observations was found (Table4) .

Territory density of soil life-feeding birds had ground dwelling arthropods inbot h cropare a and E.I. aspredicto r variables and was found tob epositivel y related to lycosid spiders inth e E.I.an dnegativel y related to linyphiid spiders granivorous carabids inth e crop areaan d predaciousbeetle s inth eE.I . (Table 4). Withregar d tovegetatio n dwelling arthropods, soil life-feeding bird territory density could bestb epredicte d byparasitoi d abundance in thecro p area (which clearly isa n association because thesebird s definitly dono t feed onthi s group) andK-herbivore s inth e E.I. (Table5) .

3.3.Bird functional groupsand relations tofarm andE.I. characteristics

Skylark territory density was found tob epositivel y related to latemowin g and negatively related tohig hvegetatio n cover ofth eE.I . inbot h 1998 and 1999 (Tables 6an d 7),an d positively to organic duration ofth e farm in 1999.I fth etw o Zuid Flevoland farms were included inth e study area, then Skylark territory density was alsopredicte d byth e spatial density ofE.I . (Table8) .

Finches (mainly linnet) territory density was found tob epositivel y related to spatial density of E.I. andplan t species richness ofE.I .vegetatio n in 1998 (Table 6) and organic duration and areao fE.I .i n 1999(Tabl e 7);i n 1999a negativ erelatio n was found topercentag e of gramineous crops inth e crop rotation (Table7) .

Motacillidae (wagtails/pipet)territor y density in 1998 and 1999,wa s found to be positively related to spatial E.I. features (E.I. area in 1998an d E.I. density in 1999) and negatively related topercentag e of gramineous crops (Tables 6 and 7).I n 1998i t was negatively related tomowin g frequency and in 1999als o negatively to height of E.I. vegetation but positively to number of dominant plant species. Ifth etw o farms inZui d Flevoland were included inth e study area than wagtails/pippit were also found tob e related positively to weed abundance whereas spatial density ofE.I .wa s no longer included inth emode l (Tables 7an d8) .

Marshbir dterritor y density was found tob e exponentially related toheigh t ofth eE.I . vegetation in 1998an d 1999(Table s 6an d 7)an d curvilinearly to organic duration in 1999 (Table 7)an dwa s found tob enegativel y related to area and curvilinearly to latemowin g of the E.I.i n 1998 (Table6) .

Insecti-granivorous bird territory density (including skylark, finches, goldfinch andquail )wa s found tob epositivel y related to latemowin g of E.I. in 1998 and 1999(Table s 6an d 7)an d organic duration in 1999(Tabl e 7) andwa s found tob enegativel y related to frequent mowing and openvegetatio n in 1998(Tabl e 6) and also to E.I. dominant plant species number in 1999 (Table 7).I fth e two farms inZui d Flevoland were included inth e study area than insecti- granivorous bird territory density was also predicted byE.I . density andwa s found tob e negatively related to mowing frequency andE.I .plan t species number (Table8) .

Insectivorous bird territory density (including blue-headed and whitewagtail , meadow pipit, reed warbler, reed bunting andbluethroat ) was found tob epositivel y related to E.I. dominant plant species number andnegativel y topercentag e gramineous crops in 1998 and 1999 66 (Tables 6an d7 )an dnegativel yt oE.I .spatia l density in 1999.I fth etw oZui d Flevoland farms wereinclude dthe nth eselecte dmode linclude dwee dabundanc ea sa predicto r for insectivorous bird territory density,wherea s E.I.spatia ldensit ywa sno tinclude d (Table8) .

Total songbird territory densitywa sfoun d tob epositivel y relatedt oplan t speciesnumbe ro fE.I . vegetation in 1998an d 1999(Table s 6an d 7)an dlat emowin gan dmowin g frequency in 1999 (Table 7),an dnegativel yt opercentag e gramineouscrops ,mowin g frequency andcove r ofE.I . vegetationi n 1998(Tabl e 6).I fth etw oZui dFlevolan d farms wereinclude d thenth eselecte d modelinclude dorgani c duration asa predicto r fortota lbir dterritor ydensity ,wherea sE.I .plan t speciesnumbe rwa sno t selected (Tables 7an d8) .

Soillife-feedin g birdterritor y densitywa s found tob epositivel yrelate dt olat emowin gan d curvilinearly toheigh to fE.I .vegetatio n in 1998(Tabl e6 )an dpositivel yrelate dt oorgani c duration andnegativel yrelate d topercentag e gramineouscrop san d spatial density ofE.I .i n 1999 (Table7) .I fth etw oZui dFlevolan d farmswer einclude dthe nn osignifican t modelcoul db e found (Table8) .

Table 6 Multiple regression models (P<0.05) with farm and E.I. characteristics as predictor variables based on bird observations (response variables) on eight farms in 1998. The keys are in Table 5 (birds) and Table I (farm and E.I. characteristics).

Birds Model Pr>F Alau 0.0033*(MOW)2 -0.00006*(EICOV) 2+ 0.47 9 78 0.0231 Fin 0.0015«(EI100)- 0.0126*(AREA) + 0.0106*(EISPN) -0.09 4 99 0.0003 Mota -0.0002*(iNT)2- 0.6678*(AREA) +0.1478*(AREA) 2 -0.1219*(MOWFRQ ) + 1.214 99 0.0029 Marsh -0.0033*(AREA)2 + 0.185*(MOW) - 0.0146*(MOW)2+ 0.0446*(EIHGT) 2- 0.58 4 97 0.0163 Insgran 0.0036*(MOW)2- 0.0174*(MOWFRQ )- 0.00008*(EICOV) 2+ 0.841 95 0.0048 Insect -0.0037*(INT) +0.126*(EIDOM )- 0.1008 83 0.0117 Avis -0.0029*(INT) - 0.159*(MOWFRQ)2- 0.00007*(EICOV) 2+ 0.026*(EISPN ) + 1.23 98 0.0045 Soilf 0.0038*(MOW)2+ 0.309*(EIHGT ) - 0.187*(EIHGT)2 - 0.135 97 0.0011

Table 7 Multiple regression models (P<0.05) with farm and E.I. characteristics as predictor variables based on bird observations (response variables) on seven farms in Oostelijk Flevoland in 1999. The keys are in Table 5 (birds) and Table 1 (farm and E.I. characteristics).

Birds Model Pr>F Alau 0.0001*(DUR) 2+ 0.0035*(MOW) 2+ 0.0059*(MOWFRQ)2 - 0.0059*(EICOV) +0.48 3 100 0.0007 Fin 0.0088*(DUR) -0.0020*(INT ) + 0.0063*(AREA)2 + 0.0308 99 0.0005 Mota -0.0068*(INT) +0.00003*(E I100) 2 -0.044*(EIHGT) 2 + 0.057(EIDOM) 2- 0.37 99 0.0119 Marsh -0.0077*(DUR) +0.0006*(DUR) 2+ 0.021*(EIHGT)2 + 0.0014 98 0.0056 Insgran 0.014*(DUR) +0.036*(MOW ) -0.018*(EIDOM) 2+ 0.05 3 98 0.0049 Insect -0.00007*(INT)2 -0.00005*(EI100) 2 +0.198*(EIDOM ) +0.66 6 95 0.0186 Avis 0.0148*(MOW)2 +0.253*(MOWFRQ ) - 0.334*(EIHGT) +0.057*(EISPN) 2 - 1.015 100 0.0009 Soilf 0.0002*(DUR)2 -0.002*(INT ) - 0.00003*(EI100)2 +0.02 3 94 0.0251

Table 8 Multiple regression models (P<0.05) with farm and E.I. characteristics as predictor variables based on bird observations (response variables) on nine farms in Oostelijk and Zuidelijk Flevoland in 1999. The keys are in Table 5 (birds) and Table 1 (farm and E.I. characteristics).

Birds Model R2 Pr>F Alau 0.0033*(DUR)2+ 0.000012*(EI100) 2 + 0.0034*(MOW)2 -0.00006*(EICOV) 2+ 0.34 7 98 0.0015 Fin 0.0084*(DUR) -0.000009*(EI100) 2+ 0.0227*(MOW )- 0.075 1*(EIHGT ) - 0.109 99 0.0006 Mota -0.00002*(INT)2 + 0.0559*(WABUN) +0.3127*(MOW ) -0.0201»(MOW) 2 +1.122 93 0.0121 Marsh -0.216*(MOW) +0.0138*(MOW) 2 -0.144*(EISPN ) +0.0101*(EISPN) 2 + 1.243 98 0.0008 Insgran 0.0268*(DUR) + 0.0036*(EI100)- 0.129*(MOWFRQ)2- 0.0625*(EISPN ) +0.59 7 88 0.0418 Insect -0.0024*(INT)2 +0.0404*(WABUN )+ 0.0026*(EISPN)2+ 0.12 8 95 0.0012 Avis 0.0158*(DUR) + 0.112*(MOW) -0.0994»(EIHGT) 2 -0.37 1 89 0.0081 Soilf No model 67 3.4.Bird functional groups andrelations tolandscape characteristics in1998

Table 9i spresentin g the significant best multiple regression models for bird functional groups,usin g landscape characteristics aspredicto r variables; semi-natural habitats included canals,cana lbanks ,roa d verges,tal lherbs , shrubs andwoodlots .

Table9 Multiple regression models (P<0.05) withfarm and E.I. characteristics aspredictor variables based on bird observations (response variables) on eightfarms in 1998.Keys in Table5 (birds) and Table 1 (landscape characteristics).

Birds Model R" Pr>F Alau 0.048*(LSINT) - 0.00069(LSINT)Z- 0.0001 1*(LSVEG) 2- 0.557 94 0.006 Fin 0.000018»(LSAGE)2- 0.031«(LSNAT )+ 0.0037*(LSINT)- 0.003*(LSVEG) + 0.154 97 0.012 Mota -0.0067*(LSINT) 51 0.045 Marsh -0.024*(LSVEG) +0.0005*(LSVEG) 2 +0.26 8 83 0.011 Insgran 0.004*(LSAGE) +0.068*(LSINT ) -0.001*(LSINT) a -0.007»(LSVEG )- 0.866 96 0.021 Insect -0.28*(LSNAT) +0.02 1*(LSNAT) 2- 0.0001*(LSVEG) 2 + 1.065 97 0.001 Avis -0.057*(LSNAT) + 0.717 58 0.028 Soilf 0.00005*(LSAGE) +0.00005*(LSVEG) 2 +0.04 9 88 0.005

Skylark territories were curvilinearly related toa landscap ewit h many gramineous crops,an d negatively toa landscap e with ahig h proportion of vegetable crops.Finche s (mainly linnet) territory density was found tob epositivel y related to the age ofth epolde r and to the proportion ofgramineou s crops, andnegativel y related toth eproportion s of semi-natural habitat andvegetabl ecrops . Wagtails/pipet territory density was predicted by amarginall y significant model which included anegativ e relation to theproportio n ofgramineou s cropsi nth elandscape .Mars h birdsha d acurvilinea r relation to ahig h proportion ofvegetabl e production.

Insecti-granivorous birdterritor y density (including skylark, linnet, goldfinch and quail)wa s found tob epositivel y related to the age ofth epolder , negatively to theproportio n of vegetable crops and curvilinearly to gramineous crops.Insectivorou s bird territory density (including blue-headed andwhit ewagtail ,meado w pipit, reed warbler, reed bunting and bluethroat) was curvilinearly related to ahig h proportion ofsemi-natura l habitat inth e landscape andnegativel y toth eproportio n ofvegetabl e crops. Total songbird territory density wasbes tpredicte d by theproportio n oflandscap e with semi-natural habitat andwa s found to be negatively related to thisvariable . Soil life-feeding bird territory density was found tob e positively related toth e ageo fth epolde r andth eproportio n ofvegetabl e crops.

4. Discussion

4.1. Observations incomparison with hypotheses

Observations ofwagtails/pipe t showed most accordance with hypotheses because the territory density ofthi s group (insectivores with combined crop and E.I. affinity) was indeed related to both the crop area and E.I.characteristic s andmor e positively andmor eclearl y relatedt o vegetation dwelling arthropods (including abundant flying species) than to ground dwelling arthropods.Howeve r in general terms, similar observations were donewit hregar d to the other three groups of skylarks,finches an dmars h birds.Thi smean s thatfield observation s didno t provide indications for the expected E.I. preference of finches andmars h birds and crop area

68 preference of skylark. Also the expected association between insecti-granivorous birdsan d ground dwelling arthropod abundance was not observed.

Withregar d to crop area characteristics therewas , according to thehypotheses , agenera l positive relation to organic farming duration inth e 1999observations . Ourhypothesi s was that both organic duration aswel l as extensive crop rotation would enhance detrital subweb prey for birds. However anegativ e relation to extensive crop rotation was found withregar d to severalbir d functional groups,an dparticularl y toth e insectivorous groups.

With regard to ecological infrastructure variables, observations were largely in accordance with ourhypotheses ,becaus e selected models often included positive relations to characteristics of improved E.I.,particularl y latemowin g andplan t or dominant plant species number. However somemodel s had seemingly inconsistent combinations ofcharacteristic s as for example, earlymowin g with high vegetation.

These general tendencies,whic h havebee n discussed, canb e viewed in Table 10,whic h shows theoccurrenc e ofth e different (types of) variables in all selected models. The general tendencies are also well summarized inth ebes t models for total songbird territory density. Thesemodel s included positive relations tovegetatio n dwelling predators inth e E.I.,plan t speciesnumbe r in the E.I., and organic duration (in 1999)an d anegativ e relation to proportion of gramineous crops (in 1998).

4.2.Crop area effects

4.2.1.Food resources inthe crop area Preference ofbird s for arthropods from the E.I. may suggest that birds were attracted away fromfield s withpoo r food resources (O'Connor &Schrub , 1986).Preferenc e of skylarksan d blue-headed wagtails for field margins on conventional farms was found byD e Snoo (1995) inth e Haarlemmermeerpolder andb y Remmelzwaal &Voslambe r (1996) in 50h acro p plots inZuidelij k Flevoland. However, mean skylark territory densities on studied organic farms in 1998 (0.15/ha) and 1999(0.12/ha )wer e quitehig h when compared to means in Zuidelijk Flevoland andth e Groningen clay district: 0.09/ha and 0.11/ha,respectivel y (Remmelzwaal & Voslamber, 1996).Densitie s of around 0.2 territories/ha found on farms with high skylark abundance, correspond todensitie s of good skylark habitats involved in studies ofPoulse net al.(1998 ) and Wakeham-Dawson etal. (1998 ) in Great-Britain. Thismigh t be an indication that the studied organic arable crop areas contributed tobir d abundance.

4.2.2. Foodresources relatedto detritivore andherbivore subsystems Negative relations to extensive crop rotation ofinsectivorou s and insecti-granivorous birds were opposit to thehypothesis . Apparently distributions ofbird s and carabid beetles and linyphid spiders (involving >80%o f ground dwelling invertebrate numbers) did not match. Explanations may involve food aswel l asothe r environmental factors, however inthi s article weprimaril y explore explanations concerning food resources. Apossibl e explanation might betha t the increased detrital subweb,relate d to increased carbon supply, depresses the herbivore subweb (Wiseet al, 1999)whil eherbivore s arepreferre d prey items in 'chick-food' (Hald &Reddersen , 1990;Moreb y &Sotherton , 1997).Thi s explanation willb e explored below.

69 Table10 An overview of the variables that were selected by significant multiple regression models based on bird territory density (response variable) withfarm and E.I. characteristics aspredictor variables aswell as arthropod functional groups as predictor variables. Key: On the horizontal axis birds: alau =skylark, mota =Motacillidae, fin - finches, marsh = marsh birds, insect =specialist insectivorous birds; insgran- insecti-granivorous birds; avis =total songbirds; soilf=soil life- feeding birds; On thevertical axis, characteristics atthe farm and at the E.I. habitat level (key see Table I); down atthe left side: ground dwelling arthropods: col= predacious beetles; lye =lycosid spiders, lin =linyphiid spiders; sism = small staphylinid beetles; carh - granivorous carabids;foli=folivorous insects; down atthe right side: vegetation dwelling arthropods: Kherb—K-herbivores; detr =detritivores; pred= predators; par =parasitoids; rherb = r-herbivores; additional symbols: 'ca' =crop area; 'ei' =ecological infrastructure.

1998 1999

Factor Factor JS ed c** CA o 3 C*H CA CA 3 CA 3 s 4> 60 3 l-H (A CA o CA c '> 'o 00 'o o IS to .9 CA s a a CA 6 .s 6 CA B a dur dur X X X X X int X X X int X X X X wabun eilOO X eilOO X X X area X X X area X mow X X X X mow X X X mowfrq X X X mowfrq X X eihgt X X eihgt X X X eicov X X X eicov X eidom X eidom X X X X X X eispn eispn — colca Kherbca lycea X detrca stsmca X predca X linca X X X parca X carhca X X X rherbca X folica colei X Kherbei X X X lycei X X X detrei X X stsmei predei X X X X linei X parei X X X X carhei X rherbei X X foliei

The observation that Skylarkterritor y density related tor-herbivor e (i.e. Aphididae) abundance suggested that thisbir d isassociate d with an increased herbivore subweb in intensive crop rotation. However aphids areprobabl y not important food item for skylarks (e.g.Cramp , 1988;Poulse n etai, 1998) andma ytherefor e indicate other (food resource) factors.

Probably crops in intensive organic farms are generally morevulnerabl e topes t outbreaks than crops in extensive organic farms duet o crop physiology (e.g. Phelan, 1997) aswel l as a lowerpredatio n pressure from detrital subweb-based polyphagous predators. This type of crop vulnerability isdemonstrate d in the conventional plots of various comparative studies including various crop species (Kromp &Meindl , 1997).Tim e serieso farthropo d numbers in crop edgeso f conventional crops (Remmelzwaal &Voslamber , 1999)wit hregar d to various crop species, show similar exponential increaseso farthropo d numbers which are temporarily higher than arthropod densities in grassmargins .Hal d &Redderse n (1990) also found higher arthropod densities in conventional cerealfields tha n inorgani cfields i fth e exponentially 70 developing aphidswer etake n into account. They notetha t aphids are supplementary preybu t that theunpredictabilit y ofthei r densities limitth ecarryin g capacity ofconventiona l crops. On organic farms inFlevolan d spatial variation of individual crop species may compensate for the short time intervals of crop-specific herbivore increases. Accordingly, of skylark territories (Nan etal, inprep. ) showed an affinity of skylarks for within-field crop borders,whic h may facilitate foraging indifferen t crops and crop edges (Wilson etal, 1997).

The effect ofpredatio n bypolyphaguou s predators mightb e stronger onnumber s oflarge - sized invertebrate herbivores (K-strategists) than onnumber s of small-sized r-strategists, because K-strategist populations mayb evulnerabl e due the scarcity ofhos tplants ,bot h weedy andundersown . Poornutritiv e quality of adie t comprised ofmainl y r-herbivores {i.e. aphids) may also be involved (Toft, 1996).Herbivorou s preyresource s on organic arable farms inFlevolan d may therefore include only asmal lproportio n of favourite items (sawfly and lepidopteran larvae).However , inthi s caser-herbivore s mayb erelevan t as food resource for insectivorous birds.

In crops or crop rotations where epigeicpredator s exploit detritivores and subsequently inhibit herbivores,bird s have to feed onthes epredators . However, food web structures inthes e habitats represent apoo r resource for chick food. Observations ofPoulse n etal. (1998 )ma y be coherent with ourtheory : skylark chick diet and survival was compared inthre e distinct crop habitats (four year old set-aside, grass silage and springbarley) :nestling s in set-aside received many small loadings of small soft-bodied (i.e.herbivorou s but no aphids) organisms whichwer eabundan t intha t crop habitat;nestling s inbot h other cropsreceive d larger loads including hard-bodied organisms,particularl y carabid beetles. In spring barley arthropod abundance was lowest andnestlin g diet included ahig hproportio n of spiderstha t perhaps, compensated for the lack of larger items;th e observed chick mortality in spring barley was probably causedb y starvation.

Examination of arthropod communities provided preliminary evidence for theseideas : detailed analysis of ground dwelling arthropod communities in 1998 (Smeding etal, 2001a) indicated thatwhea t crops in intensively managed farms, ascompare d to extensive farms, had lownumber s of epigeicpredator s (and high numbers of fungivorous small staphylinids) whereaswhea t crops of farms with lessintensiv e (i.e.vegetable ) production couldb e associatedwit h an abundance ofcarabi dbeetles , granivorous carabidbeetle s and lycosid spiders.Detaile d analysis ofth evegetatio n dwelling arthropod community on arable farms in Flevoland in 1999(Smedin g etal, 2001b) indicated thatwhea t crops in extensive crop rotations of long duration farms had alo w abundance ofherbivore s and amoderatel y high abundance of detritivores (i.e.smal l diptera species) andparasitoids ,wherea s wheat cropso f intensive crop rotation farms were characterised by (potential) outbreaks ofr-herbivore s and absence ofhig h numbers of other arthropod functional groups.

The ideaspresente d correspond toth ehypothesi s ofPowe r etal. (1996) ,tha t partly contradicts the hypothesis of thepositiv e relation between increased primary productivity and toppredatio n (e.g. Oksanen etal, 1996).Powe r etal. (1996 ) found intempora l succession on river flood plains,tha t food chains shorten from threet o two steps,becaus e in alate r successional stage armoured secondary consumers inhibit food supply ofth eto p predators;i n the dynamic situation of early successional stageth eprimar y organisms arebette r accessible and giveris et o longer lengthsmor e than two steps.I nmor e or less stabilized farm food websdu et o detrital subsidy andherbivory-resistan t cropphysiology , dominant predacious invertebrates mayb e epigeic predators. This guild is 'armoured' from abird s perspective; their 'armour' may relatet obot h hard bodies (carabids) aswel l ashidin g in dense

71 vegetation ori n the soil, size(e.g. fo r wagtails) andnocturna l behaviour (Vickerman& Sunderland, 1975).

4.3. EcologicalInfrastructure effects

Territory densities ofmos t investigated bird functional groups showed thepositiv e effect of increased numbers ofdetritivore s andpredator s in late mown E.I. (Table 10).Accordingl y increased arthropod abundance in latemow n (uncut) vegetation, ascompare d to early mown, isreporte d in several studies (Boatman etal, 1999;Wakeham-Dawso n etal., 1998 ; Schekkerman, 1997;Curry , 1994).Wagtails/pipe t may have foraged on abundant small flies. Negative relations ofbird st o parasitoids andK-herbivore s indicated an avoidance of early mown E.I.becaus e these groups wererelativel y abundant in traditionally managed canal banks (Smeding etal., 2001b) .However , skylarks also showed some affinity for open E.I., possibly attracted by hoppers, lycosid spiders,granivorou s carabids and ants.

Apart from marsh birds,ther ei s atendenc y to avoid high E.I. vegetation. Thisma y havea behavioural background that thetypica l arable farmland birds prefer habitats that afford a good view. Marsh birds did not relate stronglyt o the average value of E.I. vegetation height because local stands ofree d or shrubsma y alreadyb e sufficient to establish anumbe r of territories.

Emergent may attract or catch flying arthropods (Hradetzky &Kromp , 1997). Many detritivorous diptera and aphids in late-mown vegetation areperhap s not autochthonous for thehabitat . r-Herbivores density in the E.I. and crop areawer ehighl y correlated (Smeding etal., 2001b) .Thi s may explain why marshbirds ,whic h mostly stay in and around E.I. vegetation, could bepredicte d by r-herbivores in the E.I. and, intensive production indicating (Smeding etal, 2001b), small staphylinids inth e crop area. Densities ofmars hbird si n emergent vegetation surrounded by arable crops might bepromote d by apre y 'precipitation' ofbot h detrivores and herbivores from the crop area.

Hald &Redderso n (1990) noted that additional sweep-net sampling, including morebir d prey items, exaggerated the difference between conventional and organic crops. Accordingly catches of four yellow water traps on farm E (Venhorst &Smeding , unpublished data) included taxaunde r represented by suction trapping, containing 75%dipter a ofwhic h c. 10% involved largertax a (>2m gdr y weight).Thi s under represented group,involvin g anthomyid flies, scatophagid flies, craneflies andpredaciou s flies might be important for various insectivorous birds (Poulsen etal, 1998;Cramp , 1988).

Soil life-feeding birds were expected torelat et o crop area characteristics. However, also association with certain E.I. characteristics were found. This is in accordance with impressions in thefield tha t chicks of lapwing often occurred in ditches,particularl y those with 'traditional mangement',wher e mud and shallow water innon-vegetate d open ditch bottoms provide soft ground and abundant aquatic prey, including molluscs.

4.4.Landscape effects

Differences inbir d territory densities between farms were also affected by landscape factors. TheZui d Flevoland polder had ashorte r life and largerfields whic h means a lower spatial density ofE.I .Territor y densities of skylarks, wagtails/pipet, linnet and lapwing relate positively toon e orbot h ofthes evariable s and therefore olderpolder s were preferred. 72 Negative relationst o linyphiid spidersca nb eexplaine db yth e fact that linyphiids preferred the larger scale found inth e younger polder (Smedinget al, 2001a). At the landscape level skylarks seemed tob e associated with ahig hproportio n of gramineous crops and alo wproportio n of vegetable crops.Thi s wasmor e or less opposite to the farm- level observations of skylarksbu t inaccordanc e with the expectation from literature. Landscape level crop areas included both conventional and organic crops.O n conventional farms cereals may offer the largest food resource for birds.However , on organic farms other crops than cereals could be relevant for prey provision.

According to expectations,mars h birds were enhanced by ahig h percentage of semi-natural habitats in the farm surroundings. Impressions inth e field were that vicinity of canalswit h reed stands,pond s or dense shrubs {e.g. in a2 h anatur e reserve near farm A) encouraged nesting ofthi s functional group inth e farm E.I. However, total songbird territory density is related negatively toth eproportio n of semi-natural habitat atth e landscape level. Accordingly spatial analysis of skylark observations with GIS (Nan etal., i nprep. ) revealed that skylark territories are clustered on farms, particularly at some distance from the farmyards, tree lines androads .Thi s mayreflec t thepreferenc e shown by typical arable farmland birds for open landscapewit hn o disturbance.

4.5. Conclusions

Results ofbir d studies in 1998an d 1999o nte n organic arable farms inFlevolan d suggest that bird territory density isparticularl y supported by organic arable farms with intensive crop production and improved (latemown , speciesrich ) ecological infrastructure. Bird food resources on intensive farms may bebase d onperiodi c and local herbivore increases in acro p mosaic. Extensive systems might bemor ebalance d due to thepest-contro l by beneficials (i.e. polyphagous predators) andt o ales s susceptible crop physiology. However, inthes e systems with greater detrital subsidized food webs (cf.Wis e etal, 1999),th ehighe r trophic levels(i.e. insectivorous birds) might be inhibited by secondary consumers which exploit the primary level andhav e limited availability for vertebrate predators (cf.Powe r etal, 1996).

Crop areas onorgani c arable farms inFlevolan d mayno t resemble the food rich habitats observed byHal d &Redderso n (1990)o r in set-aside (e.g. Poulsen etal., 1998)du et o perhaps intensity and/or abundance ofepigei cpredator stha t inhibitpopulation s of soft-bodied herbivorous larvae.

Late mown E.I.mos t probably offers animportan t food resource, supplementing both quantitatively and qualitatively, the crop area food resources andpossibl y filling temporary gaps in food availability.

Our observations provided preliminary evidence for these ideas; further research is required.

4.6. Recommendations

Further research on the topic requires restricted geographical variation between farms. Experimental manipulation ofbir d populations atth e farm level isextremel y expensive. The study areama y therefore include extreme types ofcommercia l farms, with regard to intensity and E.I. management and one ortw o experimental (pilot)farms which have, for example, increased detrital subsidy based on crop residues and organic manure.

73 The suggested importance ofr-herbivore s for birds inth e crop area ishar d to translate in practical recommendations because r-herbivores areusuall y pests. Itma y therefore bea challenge for farmers and agroecologists to design croprotation s with controlled herbivory in for example mixed crops, living mulches (Theunissen, 1994;Theunisse n etal, 1997),le y crops and grazing-resistant cultivars.

Closer examination of localbir d diets and foraging habitats isrequired , assessing the contribution of aphids, craneflies and carabid beetles inbir d diets,an d tracing important taxa. Other catching methods,preferabl y including adensit y assessment, should betrie d(e.g. Duelli etal., 1999).Als onon-foo d habitat factors (machine disturbance andpredation ) should be considered.

Special attention should begive n to the effect of local 'hot spots'o f food, represented by wagtails feeding ondun g deposits (e.g. Hald &reddersen , 1990)o r lapwings feeding atditc h bottoms.

Modification by farmers of E.I.managemen t isver yrewardin g interm s ofbir d abundance and therefore ofnatur e conservation value. Inparticula r because latemowin g contributes to bird food resources in early summer. However, managements with cautious, frequent mowing at>1 0c m above ground surface may also stimulate herbivores and predators important for birds.Optimall y managed E.I. mayinclud e apermanent , simpledivisio n into some frequently mown and late mownboundarie s (seee.g. Feber etal., 1999 ) and also local,broa d standso f tall vegetation or shrubswhic h facilitate nesting ofmars hbird s (seee.g. Stoate, 1999).Ree d or shrubs should preferably be situated close toth epresen t semi-natural habitat sotha t open space,attractiv e for typical farmland birdswil lno t become fragmented (Smeding & Joenje, 1999).

Acknowledgements

The authors wish tothan k the farmers who allowed usacces s to their land. Thanks are duet o AndreMaasse n (Unifarm) and Clasien Lock (PRI) for technical assistance,D r A.J. Beintema (Alterra) and Wouter Joenje (WU) for discussing our work during aon e day field visit, and Bart Venhorst, Bertus Meijer andJa nNoorlande r for arthropod measurements. Thewor k at the Wageningen University is funded by theMinistr y ofAgriculture ,Natur e Management and Fisheries, Department of Science andKnowledg e Dissemination.

74 Chapter 5: Effect offiel d margin management on insectivorous birds, aphids and their predators in different landscapes

F.W. Smeding"& C.J.H . Booijb "Department of Plant Sciences, Biological Farming Systems group, Wageningen University, Wageningen, The Netherlands 1'PlantResearch International, Wageningen, The Netherlands

Abstract An increased field margin area at farm level, is thought topromot e vertebrate and arthropod predators and to reduce pests. In a field trial in 1997 a comparison was made between animal communities on four organic farms thatpossesse d different amounts of ecological infrastructure (E.I.)- One P3^ was situated in young landscape and the other in old landscape. Only specialist insectivorous birds showed a clear positive relationship. With regard to important arthropod species in wheat (e.g. carabid, staphylinid and coccinellid beetles, spiders) abundance differences inrelatio n to amount of E.I. were not related tool d and young landscapes. High aphidophagous syrphid fly abundance and aphid (pest) densities occurred in young landscape farms.

1.Introductio n

1.1. Agricultureand biodiversity

Biodiversity inEuropea n agricultural landscapesi sdecreasing .On eimportan t solution for reversingthi sbiodiversit y lossi st odevelo p farming systemstha t integrate agriculture andnatur e conservationinterests .Organi cfarmin g triest oimplemen tthi side abecaus ebiodiversit yi s thoughtt oenhanc epes t control (Altieri, 1994). Farm-natureplan ssuppor tthes e farms todevelo p site-specific habitat diversification whichmean sth eestablishmen to fa necologica l infrastructure (E.I.)(Smedin g &Joenje , 1999).Th eE.I .i sdefine d asth etota lo fsemi-natura l (aquatic,woody , tallher b andgrassy )habitat s onth e farm includingit sspatia larrangement . Habitat diversification islikel yt oincreas enatura l enemy abundancean dt oreduc epes tpopulation s (Booij &De nNijs , 1996);thi smean stha tth eherbivorou s pestorganism srepresen t asmal lpar t ofth emainl ynon-pes t herbivore anddetritivor e communities inth efar m food web.Th eincreas e ofbeneficia l andharmles s(prey ) species enhancesth e food availability forth evertebrat e species onth efar m (e.g. Moreby &Sotherton , 1997; Hald& Reddersen , 1990), whichar eo f conservation interest.

1.2. Farmecological infrastructure

Practical recommendations tooptimiz eth eEcologica l Infrastructure often relyo nexperienc ean d common sense.Detaile d studiesabou tke yspecie sar eavailable ,e.g. skylark (Wilsonet ai, 1997)an dpartridg e (Potts, 1983), however, iti sdifficul t tocombin e andextrapolat ethi s knowledget o farm practice andtranslat ei tt oaction so fth e farmer.

Ourstud ywa sintende dt ocontribut et o asystem sapproac h that linksinformatio n onth eecolog y ofspecie s(groups )wit hfar m topography andmanagement . Therefore weundertoo ka simultaneous examination ofsevera lintegratio n levels(landscape , farm andhabitat/field ) and animalspecie sgroups .Far man d faunameasurement swhic har ebot heffectiv e inth efiel d (Den Nijs etai, 1998)a swel l asexplanator y factors in future theorywer e selected. 75 Ina field tria li n 1997a compariso nwa smad ebetwee n four contrasting organic farms that possesseddifferen t amountso festablishe d E.I.an dmigh tserv ea sgoo dexample sfo rth eabov e theory.I ti sprobabl etha tth eeffec t ofequa l efforts byth efarme r woulddiffe r between landscapes;a biodivers eneighbourhoo d mayoverrul eo ramplif y theeffect . Therefore onepai r oforgani cfarm s wassituate d ina youn g(culture ) landscapean danothe rpai ro forgani c farms in anol d(culture ) landscape;ol d landscapei sassume dt ob emor ebiodiverse . Ourhypothesi swa s that anincrease dE.I .woul db eassociate d withmor einsectivorou sbirds ,mor e arthropod predators anda reduce dpes t infestation.

2.Material s andMethod s

2.1.Study sites

Theresearc hwa scarrie dou to nfou r different organic farms (indicated inth etex t asY- ,Y+ ,O - and0+ ) oncla ysoil si nApril-Augus t 1997.Tw ofarm s weresituate d inth eyoun g landscape (=Y)o fth epolde rZuid-Flevolan dwhic hwa sreclaime d in 1968.Th eothe rtw ofarm s werepar t ofa nol d landscape(=0 ) nearth erivers i nth ecentr eo fTh eNetherlands .I neac h landscape farms witha smal lE.I.(=- )an da far m witha greate rE.I.(=+ )wer eselected . Withineac h farm winterwhea tfields wer echose n asa representativ e crop for arthropod andarabl ewee d measurements.Cerea laphid swer echose n asa nexampl eo fa pest .

ForDutc hstandard sal lstudie d farms canb econsidere d asol dorgani c farms. FarmsY -an dY + startedi n 1985o na nexperimenta l reclamation areawhic hneve rreceive d pesticides.Far mY + joined agrou po f 10 'innovation'pilo t farms in 1993 which impliestha tal lditche shav e adjacent threemete rwid egras s stripsan dtha tditc hvegetatio n includes severalartificiall y introduced indigenousplan t species (Vereijken, 1998). Farm O-converte d toorgani c farming practicei n 1980;it sintensiv eproductio n isfocusse d oncabbag ean dleek ;ol dhedgerow swer emaintained , mainlya tth eborder so fth efarm . Farm 0+ wasru nbetwee n 1971-1995 aspar to fa secondar y schoolfo rbiodynami c farming; extensiveboundarie s includinghedgerow swer ecreate ddurin g thisperiod .

2.2. Samplingand measurements

2.2.1. Landscape Landscapeparameter swer e sampledi na circl e aroundth emathematica l centreo fth efarm . The radiuso fthi scircl ediffere d betweenth efou r farms becausei tinclude d asectio no fth efar m and anextr a20 0m from th efar m border; the20 0m distanc e approximates toth ehom erang eo f severalsmal lmammal san dsongbirds .Insid eth ecircle ,th eare ao fdifferen t agricultural,semi - natural andothe rhabitat swa smeasured .Th esemi-natura l aquatic,woody , tallher b andgrass y habitatswer edefine d asE.I .Th escal eo fth e landscapewa sdetermine d bymeasurin gth etota l areas("blocks' )o finterconnecte d agricultural fields.

2.2.2. Farm Thelimit so fth e study areawer e equalt oth eadministrativ eborder so fth efar m becausethi si s theunitar ysyste munde rcontro lb yth e farmer (Swift &Anderson , 1993).Th earea so fsemi - naturalhabitat swer emeasured , asfo r landscape.

Thespatia larrangemen t ofth e ecological infrastructure wasexpresse db yth epercentag e ofth e farm whichla ywithi na 100m distanc eo fth eecologica l infrastructure. Theinfluenc e ofbound ­ arieso nth eabundanc eo flarg ecarabi dbeetle s andspider sprobabl y doesno t exceed 100m . 76 2.2.3. Vegetation Vegetation studiesinclude dboundarie s surroundingth ecrops .Th eher b layerwa ssample di n onemete r lengths atinterval s of25-10 0m betwee nsamples ,dependin go nth elengt ho fth e boundary.Parameter s were:cover ,maximu mheight ,plan t species anddominan tplan t species (which determine 80% cover).

Weed speciescompositio n inwinte rwhea twa srecorde db ylistin gth e species in4 0 sampleso f 0.25 m ineac h field.

2.2.4. Birds Territorymappin go fbird swa sdon ethre etime s atth ebeginnin g andth een do fMa yan dth e beginningo fJune ,usin gth emetho ddescribe d inVa nDij k (1993).Breedin gbird swer egroupe d accordingt othre e food types:insectivorous ,insectivorous/granivorou s andsoi l life-feeding.

2.2.5. Arthropods Weekly sampleswer etake no neac hfar m inwinte rwhea tbetwee n 29Apri l and2 4July .I neac h winterwhea t field four randomplot s of 10b y2 0m wer elai dout ,i nwhic hthre epitfal l trapsan d ayello wwate rtra pwer eplaced . Aphidswer ecounte do n8 0t o20 0tiller spe rplot ,dependin go n the expectednumbe r ofaphid spresent . Thecarabi d andstaphylini d beetlesan dspider scaugh ti n pitfall trapswer esorte dint osiz ecategories ;syrphi d flies andcoccinellid s inth ewate rtrap swer e determined tospecie slevel .

InSeptembe rthre e suction samples (3minute si n 1 m2)wer ecollecte d inadjacen t boundaries including atal lher b andgrass y anda shru b sample;herbivorou sbeetle s(curculionid san d chrysomelids)wer edetermine d tospecie slevel .

3. Results

3.1.Landscape characteristics

3.1.1. Farmsurroundings Theyoun glandscap ewa sdominate d byagricultura l land usean dha d largeblock so fcontinuou s agricultural land (Table 1).Th epercentag e semi-naturalhabita t appeared tob eunaffecte d by landscapetype .

3.1.2. Farmand its habitats Croprotatio ni nyoun g landscapetende dt oinclud e fewer extensivecrop s (cereals andle y pasture) (Table 1),howeve r farm Y-include d arelativel ylarg epercentag edu et oit slarg esize . Wheat yieldswer ehighe ri nth eyoun glandscap ewherea s arableweed si nthi scro pwer emuc h moreabundan t anddiver si nth eol dlandscap e(Tabl e 1).Farm si nyoun glandscap eha da large r cropare atha twa sremot efro m theboundarie s (Table 1).Boundarie s inth epolde r includedmor e aquatichabita t(ditche san dbanks )an di nth erive rregio ninclude dmor ewood y(hedgerow ) habitat. Thestructur e diversityan dspecie srichnes so fth eboundar yvegetatio n appearedt ob e unaffected byth elandscap etyp e(Tabl e1) .

77 Table 1 Characteristics offour farms inyoung (Y) andold (O) landscape with small (-) or increased(+) E.I. with regard to landscapeand habitats.

Y- Y+ O- 0+

Landscape % agricultural land use 71 94 54 52 block size (average)(ha) 23 26 3 2 % semi-natural habitat 27 4 40 18

Farm farm area(ha ) 10 55 22 28 % cereals/ley pasture incro p rotation 50 29 50 75 % farm areawithi n 100m range ofE.I . 61 72 81 100 % semi-natural habitat = E.I. 1.2 2.6 1.7 5.8

Crop no.o f weed species/0.25 m2 3.2 1.6 5.4 5.1

Boundary no.o fplan t species/m (average) 7.6 11.4 9.0 9.9

3.2. Effects ofEcological Infrastructure

3.2.1.Young landscape Characteristics offar m Y+(2.6 %E.I. ) incompariso nwit hfar m Y-(1.2 %E.I. )were :a mor e diverseboundar yvegetatio n andles sweed si nwhea t (Table 1);a highe r density andspecie s number ofbreedin gbird sdu et oth eoccurrenc e ofspecialis t insectivores inth eboundarie s(Fig . la); aslightl y greater abundanceo fmos tpolyphagou spredato rgroup si nwhea t(carabids , staphilinids andlycosi d spiders) (Figs lb and lc),wherea saphidophagou s syrphidsan d ladybirdswer eslightl ymor eabundan t inwhea t on farm Y-(Fig . lc);simila rhig haphi d abundancei nwhea t(Fig . Id); ahighe r speciesnumbe r andsimila r (low)abundanc eo f herbivorousbeetle si nboundaries .

3.2.2. Oldlandscape Characteristicso ffar m 0+ (5.8%E.I. ) incompariso nwit h farm O-(1.7 %E.I. )were :a mor e diverseboundar yvegetatio n (Table 1);a highe rdensit y ofspecialis t insectivorousbird s anda lowerdensit yo fmixe d insecti-granivorous birdsresultin g ina lowe rtota lbreedin gbir ddensit y (Fig. la);lowe rabundanc eo fpolyphagou s aphidpredator si nwhea t (carabid and staphilinid beetlesan dlinyphii d spiders) (Fig. lb);monophagou s aphidpredato r abundance werevariable : moresyrphids ,les sladybird s (Fig. lc) andles slacewings ;simila rlo w aphid abundancei nwhea t (Fig. Id) apart from anearl ymoderat epea ko fMetopolophium dirhodum; ahighe r species number andabundanc eo fherbivorou sbeetle s inboundaries .Th eabundanc eo fsyrphid scoul d havebee nunderestimate d becauseth eyello wwate rtrap scompete dwit hnumerou s flowers (DenNij sera/. , 1998).

3.2.3. Relationships with landscapetypes Increased ecologicalinfrastructure s ontw oo fth e four farms intw ocontrastin g landscapetype s involved adouble dpercentag e ofsemi-natura l habitats anda mor edivers evegetatio n (Table1) . Thispromote d inbot h landscapes speciestha t stronglydepen d onth eE.I .itsel f (insectivorous birdsdensit yan dherbivor ebeetle sdiversity) .I nth ecro ponl y abundance ofth eaphi dR. dirhodum seemedt oincreas e (Fig.Id).A decreas ewa s found withregar dt oth eabundanc eo f coccinellids (Fig.lc) andth edensit y ofth eaphi dSitobion avenae (Fig.ld).Howeve rmos t fauna parameterstha twer estudie ddi dno tsho wa relationship .Abundanc eo fpolyphagou saphi d predatorso nfarm s withincrease d E.I.decrease d inol dlandscap ewherea sthe yincrease di n younglandscap e(Fig.lb) ;aphidophagou s syrphids showedth erevers e(Fig.lc) .

78 a. Birds: number of territories per ha

0.8 0.6 0.4 0.2 0 ETP Y + o- o • insect feeding Hinsect/see d feeding

b. Polyphagous predators: total no. in pitfalls 12000

8000

4000 n±± -1 Y+ O 0 + ICarabidae HLycosidae •Linyphiidae D Staphylinidae

c.Aphidophagou spredators :tota lno .i nyello w watertrap s 600 400 200 HbzLJHbl3__JBS3_: Y- Y+ O- 0+ Iaphidophagou s Syrphidae DCoccinellidae

d. Aphids: sum of the average no.pe r tiller 75 50 25 SEfcE Y+ O- o+ iMetopolophium dirhodum OSitobion avenae

Fig. I. Occurrences of birds (a),polyphagous predators (b), aphidophagous predators (c),and aphids (d)on four farms: Y= young landscape; O =old landscape; - = small ecological infrastructure; + = large ecological infrastructure.

79 4. Discussion

Onlyobservation s ofspecialis t insectivorous birds andi nth eyoun glandscap e alsopolyphagou s predators supported ourhypothesis .Fe wparallel sbetwee n effects inth eol dan dyoun g landscapeswer e found. Howca nthes econtradictor y resultsb eunderstood ? Inou rattemp tt o give anexplanatio n wefocu s onsoi linvertebrate s andwee d abundancewhic h areonl ygroup s amongsevera lothers .However ,thes egroup sar elikel yt oinfluenc e numbers ofbird san d predatoryarthropods .

4.1. Younglandscape

Soilsi nth epolde r landscapear eknow nt ocontai n few earthworms andprobabl y havea lowe r soilmeso -an dmacr ofaun a biomasstha nol d landscapes.Thi si si naccordanc ewit hth eobserve d lowdensit yo fsoi llife-feedin g birds.Th elo w abundanceo farabl eweed s limitsdependen t foliar herbivores andgranivores .Thes e factors leadt oa cro phabita twher e aphidsbecom ea n importantcomponen t inth esumme rdie to fpolyphagou spredator s (Booij &De nNijs , 1996). Severalauthor srepor tthi spredominanc eo faphid si nconventionall ymanage dwhea t(Moreb y& Sotherton, 1997; Reddersen 1997).

Predator countso nfar m Y+suggeste d aninfluenc e ofincrease d E.I.bu tit sare ama yhav ebee n toosmal lan dit sdistanc efrom fiel d centresto o fart ohav ea stron geffect ; moreoverwee d abundance anda hig hpercentag e ofcereals/le ypastur eo nfar m Y-offere d better circumstances for grounddwellin g invertebrates ascompare d tofar mY+ .

4.2.Old landscape

Cropsi nth eol d landscapecontaine dmor eweed s andprobabl y ahighe rsoi l invertebrate biomass.Far mO -ha d ahig hwee d abundance;it sintensiv evegetabl eproductio n caused ahighe r general fertility levelo fth efield s whichma yhav ea quantitativ e effect onsoi linvertebrates . Thesefactor s could explainth ehig habundanc eo fth epolyphagou s carabidbeetle s (comparable withth eyoun glandscap enumbers ) andlinyphii dspider s inspit eo fth e absenceo faphids .Als o theapparentl yhig habundanc e ofinsectivorous/granivorou sbird s atfar m O-support sthi side ao f abundant food availability inth e fields.Thes erequirement spossibl y overruleth eE.I . influence oncro pcommunitie s inth egrowin gseason .

4.3. Improvingthe hypothesis

Inth e lighto fthes eresult sw econclud etha tth ehypothesi s should alsoconside r crop factors includingcro pdiversit y (e.g.cro protation ,weeds ) andsoi llif e characteristics.Ideall y comparativeresearc ho nE.I .influenc e shouldb erestricte d tofarm s wherethes echaracteristic s aresimilar .Th ehypothesi s could onlywor ko nfarm s withintermediat e levelso falternativ epre y onth efield s (associated withsoi l life andweed so rintercrops ) andwit hboundarie s thatar eno t toofa rfrom fiel d centres.I nou r studyi n 1997i tseeme dtha t animalcommunitie s inth ecrop s wereindependen tfrom E.I .I nth epolde rth eintensivel y cropped fields wereprobabl yto olarg e tob einfluence d byth eE.I .wherea si nth erive rregio nth efield s alreadyha d ahig h levelo f alternativepre yrelate dt ocro p factors. Withabundan t alternativepre ycause db ycro p factors (whichmigh tb eundesirabl e forth e farm, e.g. weeds)a smalle rE.I .ma yalread yb esufficien t to maintainbiodiversity .

80 Acknowledgements

Theauthor swis ht othan kth e farmers whoallowe d usacces st othei rland .Thank s aredu et o ClasienLoc k (PRI)an dAndr eMaasse n (Unifarm) fortechnica l assistance,Jan-Pie tZij pan d ElkeBoesewinke l for arthropod measurements andLoe sde nNij s for assistance indesignin gth e experiment. Thewor k atth eWageninge n University isfunde d byth eMinistr y ofAgriculture , NatureManagemen t andFisheries ,Departmen t ofScienc ean dKnowledg e Dissemination. Chapter 6:A concep t offoo d webstructur e in organic arable farming systems

F.W. Smeding*& G.R . deSnoo b "Departmentof Plant Sciences, Biological Farming Systems group, Wageningen University, Wageningen, The Netherlands bCentre of , UniversiteitLeiden, Leiden, The Netherlands

Abstract A proposal for a descriptive or topological farm food web is derived from field observations and from references in literature. Important themes in the food web theory are tentatively applied to this preliminary model, explaining differences between local farm food web structures andho w they are related to farm and/or ecological infrastructure (E.I.) management. Predictions aremad e for four different farm food web structures for extremes of farm and environmental gradients corresponding to the length of organic duration and amount/quality of E.I. The implications with regard to farming practices and nature conservation are that both organic duration and the amount/quality of ecological infrastructure may contribute to ecosystem services and nature conservation. However, an optimisation of the farm food web with regard to ecosystem services, may possibly run counter tonatur e conservation goals.

1. Introduction

1.1.The problem ofbiodiversity lossin agroecosystems

Species diversity of indigenousplant s and animals on farmland isdeclinin g dramatically due to agricultural intensification (Paoletti, 1999;Woo d &Lenne , 1999;Vandermee ret al., 1998 ; Swift &Anderson , 1993).Amongs t thosetha t are greatly concerned about this biodiversity loss are agriculturists, ecologists andmember s ofth epublic .Natur e conservation organisations usually addressth eintrinsi c value ofbiodiversit y and their viewpoint is translated intonationa lprogram s for species and habitat conservation (e.g.i nTh e Netherlands: Bal etal, 1995).Agroecologist s stress the economic valueo f biodiversity, related to itsecologica l services,e.g. pest control andbioti c regulation of soil fertility (Altieri, 1994, 1999;Paoletti , 1999;Pimente l etal, 1995).Biodiversit y lossha sbee n thought toresul t in artificial ecosystems,requirin g constant human intervention and costly external inputs.

Ecosystem servicesmigh t be sustained or improved by ageneral , indiscriminate, high species diversity (Altieri, 1994;Paoletti , 1999)o rb y afe w essential components ofth e local biodiversity, asdetermine d by site characteristics (Vandermeer etal, 1998;Almekinder set al, 1995).Functiona l redundancy of speciesmean stha tno t all specieshav e evident significance for ecosystem services (Duelli etal, 1999).However , itca nb e argued that agriculture has,t o acertai n extent, responsibility for all species and communities whichco - evolved with farming during around 10,000year s (Wood &Lenne , 1999),irrespectiv e their utility.

1.2. Multi-species agroecosystem research

One important solution for the reversal of the loss of biodiversity in agriculture of the industrialised countries is the development of farming systems that are economically based on

83 utilisation ofbiodiversit y and that harbour, partly duet o increased biodiversity, conservation- worthy species.Thi s idea iscompatibl e to the concept ofmulti-specie s agroecosystems (Vandermeer etal, 1998;Altieri , 1994, 1999;Almekinder set al, 1995;Swif t & Anderson, 1993).

Research aimed atthi s concept requires anappropriat e approach; three criteria, which take into account the studied objects and environmental variables, might be essential.Firstly , the research should applymethod s that are appropriate for higher levels of aggregation and avoid reduction ofth e system through emphasis on variables of lower levels (Almekinders et al, 1995). Secondly, investigations should address real farming systemsbecaus e biodiversity, relevant for apparent ecosystem services,migh t be site-specific (Vandermeer etal, 1998) and may notb e observable in small 'isolated' subsystems, or foreseeable by abottom-u p approach. Preferably study areas should include objects that express functioning of ecosystem services (Brown, 1999a).Thirdly , assessed variables should be clearly related tobot h the structural andth e daily decisions ofth e farmer sotha t the outcomes can be translated into farming practice andpolic y {e.g. Vereijken, 1997;Kabourakis , 1996;Par k &Cousins , 1995;Smedin g & Joenje, 1999).

Recent advances inresearc h on food webs (Polis& Winemiller , 1996)offe r a suitable approach for biodiversity studies athighe r levels of aggregation (Pimm, 1991; Gezondheidsraad, 1997).Thi s article therefore examinesth eperspective s of a food web approach for research that aims atmulti-specie s agroecosystem development.

1.3. Foodweb approach

1.3.1. Definitions Definitions and general information on food webs hasbee n examined extensively inth e volume edited by Polis &Winemille r (1996); some ofthei r main points are summarised here. A food web isdefine d asa networ k of -resource interactions among agrou po f organisms,populations , or aggregate trophic units. Species areusuall y aggregated in 'trophospecies' or functional groups which essentially shareresource s andpredator s but usually more criteria areuse d e.g. life history traits. Guilds are trophospecies of ahighe r order, andusuall y include several related functional groups.

In afoo d web diagram functional groups are arranged introphi c levels which haveth e same number of steps('chai n lengths') away from thebasa l resources plants or .A descriptive or 'topological food web'depict s species and feeding relations; in a'flo w web' components andrelation s arequantifie d ast onumbe r or . The most elaborate food web representation is a'functiona l web'tha t also reveals the relative strength ofrelations .

1.3.2. Researchon farm food webs Traditional studies on farmland communities often included considerations of thetrophi c structure which indirectly addressed local food webs.Example s ofthi s kind of studies areth e comprehensive empirical studyb y Potts &Vickerman n (1974), including awid erang e of invertebrate and vertebrate taxapresen t inwhea t crops in West Sussex, UK, orth e extensive surveyo fth e trophic structure of above-ground insect biomass in various agricultural landscapes by Ryszkowski &Kar g (1991) and Ryszkowski etal. (1993). Most studies directly addressing food webs wererestricte d to apar t ofth e farmland community, involving asectio n ofth eherbivor e food (sub) web or asectio n ofth e detrital (sub)web , for example studies ofpest-predato r complexes (examples ine.g. Wood &Lenne , 1999;Krom p &Meindl , 1997)o r the studies of food requirements of farmland birds 84 (reviewed byPoulse n etal, 1998). Someo fthes e studiesprovide d clear conceptual advances. For example Andow (1988) hypothesised 7'primitiv e relations' (i.e.foo d web structures) amongweeds ,crops ,herbivore s andpredators . Comprehensive topological food webso f arable cropping systems,includin g detrital andherbivor e subwebs,wer e first developed in 1977 for rice,i nth e context ofIntegrate d Pest Management programmes (e.g. Cohenet al, 1994; Schoenlyetfa/., 1996).

Amajo r breakthrough inagroecologica l application offoo d webtheory ,wer e studieso nth e detrital subwebsi n arable crops (e.g. DeRuite r etal, 1993,1995) and (e.g. Hunt etal, 1987),includin g also analyses ofnutrien t and energy flows. Progress in scientific understanding offoo d websi s expanding intoagroecosystem s asi sdemonstrate d by several recent review articles, for example:detrita l subweb control ofth eherbivor e subwebb y influencing vegetation performance (Brussaard, 1998)o rb y increased predatorpressur e on herbivores duet o externally subsidised detrital chains (Wise etal, 1999);an d interference between predacious functional groups,causin g 'trophic cascades' inth eherbivor e subweb (Tscharntke, 1997).

1.4. Objectives

However, despite the above-mentioned advances,ther eis ,t o ourknowledge ,n o comprehensive farm food web available for arable farming systems. Such a'topological ' food web, or amor e elaborate 'flow' web,ma y enable asynthesi s of fragmented data ontax a and habitats,an dcoul d subsequently form the foundation ofagroecologica l farming system research (e.g. Swift &Anderson , 1993;Va n Keulenet al, 1998)a swel l as applied agroecosystem planning (Smeding &Joenje , 1999;Visser ,2000) .

The objectives ofthi s articleare : - topropos e a simplified topological farm food web,encompassin g bothherbivor e and detrital subwebs andthei r interrelations, asoccurrin g atth e farm level of aggregation; - to demonstrate, usingbot h empirical dataan d theory,ho w thispreliminar y food webmode l couldb euse dt orelat eth eabundanc e offunctiona l groupst o farm management;

A case studywil lb euse d tomee t these objectives.

2. Methods and materials

2.1. Researcharea and delimitation ofthe system

2.1.1. Organic arablefarms in Flevoland The agroecosystem of arablefarm s inFlevolan d iswel l studied because of several large experiments (e.g. Booij &Noorlander , 1992; Booij, 1994;Brussaar d etal, 1988,1990;D e Ruiter etal, 1993,1995; Lantinga &Oomen , 1998;Smedin g &Joenje , 1999;Remmelzwaal , 1992;Remmelzwaa l &Voslamber , 1996;Vereijken , 1989,1997,1998). Flevoland consistso f threepolder s thatwer e reclaimed between 1940 and 1968 and aredominate d by agricultural land use.Th e landscape is flat andope nwit hwoodlot s confined to farmyards, villages and mainroads .Th epredominan t soiltyp ei scalcareou s clayo fmarin e origin. Organic farms may function as asteppin g stoneo nth eroa dt o developing multi-species agroecosystems (e.g.Stobbelaa r &Va nMansvelt , 2000;Vandermeer , 1995)an dwer e therefore chosen asobject s for thecas estudy .Aroun d 75organi c farms aresituate d in Flevoland including farms with improved ecological infrastructure, resulting from a 85 prototyping research project (Vereijken, 1997, 1998).Th e case study isbase d on empirical datao f 17organi c farms and one integrated farm, and literature concerning around 10 organic, integrated and conventional (experimental) farms.

Croprotation s oforgani c arable farms inFlevolan d include six ormor eyears .Commo n crops are spring and winter wheat (10-35%),le ypastur e (0-20%),traditiona l lifted crops (potatoes and sugarbeet ) (10-35%) and intensive vegetable crops (25-70%).Majo r vegetable crops are onions,peas , sweet corn,runne r beans,carro t and various cabbages.Ploughin g isgenerall y at 25-30 cm depth. The ecological infrastructure (E.I.) (Smeding &Booij , 1999)o n arable farms inFlevolan d mainly consists of linearherbaceou s boundaries of canalbank s and verges along tracks androad s (Smeding &Booij , 2001).Th e case study isbase d on empirical datao f around 20organi c farms.

2.1.2.Delimitation of the system The farming system ispositione d between commonly recognised aggregation levels ofhabita t and landscape andma ytherefor e be too small tob e considered as an (agro) ecosystem. However Swift &Anderso n (1993) consider the farm asth ebasi c unit in studies of system functioning in agricultural systems. The farm levelmigh t therefore be considered as aggregation level inth e ecological sense.Accordingly , Bockemuhl (1986) argues that taking local ecological and historical factors into account in farm management can consolidate the system borders of afar m unit.

Thesecon d criterion for spatial delimitation ofth econsidere d systemi stha t itmatche s the spacerequire d by awell-define d food web.Accordin g to these criteria Cousins (1988) defined theEcosyste m TrophicModule :th e ETM isbase d onth eterritor y size ofth eto p predator. Resident toppredator s on arable farms inFlevolan d areKestre l (Falco tinnunculus) and Stoat {Mustela ermined) (Remmelzwaal &Voslamber , 1996).Thei r territory sizes correspond to the scale and size of thestudie d farming system.

2.2. Generaldescription ofthe topological farm food web

The topological food web,includin gpreliminar y flow web characteristics, of aFlevolan d organic arable farm was developed using the following sources of information: - empirical data from farmland species inFlevolan d including data from our own research in Flevoland (references mentioned above and Smeding etal, 2001a, 2001b; Smeding& Booij, 1999,2001 ; Smeding,unpublishe d data); - literature on diets andpredator s of thetax a observed in Flevoland; - publications of research concerned with farmland species in arable crops andfield margins inWester n Europe.

The species (-groups) were listed in afoo d webmatrix . Functional groups and guilds were assembled according to commonly accepted criteria (Polis& Winemiller , 1996;Brussaard , 1998).Th eexample s ofassemblage s from theliteratur ewer e followed ifpossibl e(e.g. Brussaard etal, 1997;Brussaard , 1998; White, 1974).Som etax ama ybelon g to moretha n one functional groupbecaus e of shifts inthei r life history traits.

Thebiomas s ofth etrophospecie s reflects their ecological functions (Ryszkowski etal., 1993);biomas s estimations aretherefor e applied to illustrate the relative importance of groups and interactions in atwo-dimensiona l diagram of a farm food web.Roug h estimations, ina tenfold ordero fmagnitude ,wer ebase d on the available literature,preferabl y with data from organic farms inFlevoland , organic farms ingenera l (e.g. Kromp &Meindl , 1997)o r unsprayed crop edges (e.g. Boatman etal., 1999) .Th emea nbiomas s of arthropod taxa observed by Schekkerman(1997 ) was used torelat e density datat o biomass. Relative importancewa s assessed to derive a food web diagramfrom th emor e complex food web matrix. The farm food webdiagra m was drawni n a fashion similar to the food web diagrams ofOksanene/a/. (1996).

2.3.Relating the farm food webto farm management

Therelationshi p between food web structure and farm management was explored by comparing four management types oforgani c arable farms inFlevoland : short organic duration without improved ecological infrastructure (E.I.),shor torgani c duration with improved E.I., long organic duration without improved E.I., and long organic duration with improved E.I.Thes e four basic farm typeswer econsidere d torepresen t the extremes oftw o different gradients (cf.Hol t in Polis &Winemiller , 1996)wit h regard to duration of organic farming practice and quality ofth e ecological infrastructure. In addition, adistinctio n was madebetwee n long duration farms with improved E.I.wit h extensive and littleherbivor y in the crop area. Thus in total five food webswer e constructed.

Organic farming duration waspresume d torelat et oth e amount of input oforgani c matter in the crop area soil. This input includes organic manure, compost and crop residues ofi n particular, the gramineous crops (ley pasture and cereals).Additionally , organic duration also relates to an increase ofth e crop diversity aswel l asth ewee d diversity. The effects ofbot h an increased input oforgani cmatte r and crop/weed diversity may accumulate over years(Hal d & Reddersen, 1990;Idinge r &Kromp , 1997).

The quality ('improvement') of theecologica l infrastructure relatesparticularl y to vegetation structure and diversity within theboundary , and is largely determined byth emowin g regime. Mowing with a finger-bar mower after June 21st, seems tob eoptima l for arthropod abundance and diversity inFlevoland , when compared toth etraditiona l management ofmor e frequent mowingwit h aflai l mower (Smeding etal., 2001b) .I n addition toth e vegetation features, E.I. quality relates to the dimensions of field margins and consequently toth e proportion ofsemi-natura l habitat onth etota l farm area.

Performances of farm food web structures ofth e four theoretical farms were derived by roughly relating the available detritus andplan t resources to the abundance of detritivore and herbivore functional groups,respectively . An abundance of functional groups atth ehighe r trophic levelwa s assumed tob eprimaril y determinedb ypre y abundance ('bottom-up'). However 'top down' control andhorizonta l effects (i.e.competition ) were also considered, with reference to the evidence in literature (Wise etal., 1999 ;Tscharntke , 1997;Powe r et al., 1996;Booi j &De nNijs , 1996).Analogou s toth e approach ofFretwel l (1977),wh o coined the theory onpredato r andresourc e limitation, the four presented hypothetical food web structures werebase d on expertjudgemen t withreferenc e totheor y from literature and empirical data.

3. Results

Theresult s section includes threeparts : - anoutlin e ofth etopologica l food web; - consideration ofth e effect of farm management onth e food web structure (i.e.relativ e abundance of functional groups and strength ofthei r interactions); 87 anillustratio n ofth eprediction s with empirical datafrom arabl e farms in Flevoland.

3.1. Topologicalfarm food webon organic arablefarms in Flevoland

Because offrequentl y occurring omnivory andintraguil d predation, trophic levels are difficult to assess inth efar m food web.I t istherefor e convenient to make abroa d distinction among four aggregate subsystems (Table 1): - aprimar y guild ofherbivores ; - aprimar y guild ofdetritivores ; - anintermediat e levelo fmacr ofaun a predators of invertebrates; - ato p level ofpredator s ofvertebrates . Thecompositio n ofthes e four aggregate subsystems isdiscusse d below, and also some remarkswil l bemad ewit h regard to the relative size andth e extent ofth e energy flow through theherbivor e anddetritivor e based food chains.

Table1 Thefood web matrix of thefarm food web on an organic arablefarm inFlevoland (The Netherlands); in rows are the resources andprey functional groups respectively, in columns are the detritivores, grazers and predators, as represented by thesame functional groups; I = interaction and 2 = strong interaction (according to both observations and the literature references). Thelines represent theseparate resources and the four aggregate subsystems (detritivores, herbivores,predators of invertebrates and the top predators).

Subsystem Resource or functional groups 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 1 primary prod. roots 1 2 1 2 primary prod. leafs and stems 12 2 12 3 primary prod. flowers of dicots 1112 1 4 primary prod. seeds 1 1 1 1 2 2 2 5 detritus crop-residues, litter (C/N high) 2 1 2 1 6 detritus slurry, stable dung (C/N low) 2 2 1 7 detritivores earthworms X 1 2 1 1 8 detritivores microbivore predators X 2 1 1 1 9 detritivores detritivorous arthopods X 2 2 2 1 1 2 10 herbivores root herbivorous arthropods X 1 1 1 2 2 1 1 11 herbivores invertebr. flushfeeders (r) X 1 1 2 2 2 12 herbivores invertebr. senescence feeders (K) X 1 1 2 2 1 13 herbivores specialist anthophiles X 1 1 1 1 14 herbivores granivorous arthropods X 1 1 1 15 herbivores herbivorous vertebrates X 2 16 predators epigeic predacious beetles X 1 1 1 1 17 predators wandering and web spiders 1 X 1 1 18 predators vegetation dwelling arthropods 1 1 X 2 2 19 predators parasitoids 1 1 1 X 1 1 20 predators soil-life feeding vertebrates X 1 21 predators insecti-granivorous birds X 1 22 predators specialist insectivorous birds X 1 23 toppredators raptors, mustelids, opportunists X

3.1.1.Herbivore subweb primary guilds Guilds canb e assembled accordingt o the exploited part ofth eplant : roots,gree n material (leafs and stems),flower s and seeds(D e Snoo& Canters , 1988;Crawley , 1983)(Tabl e 1).

88 Theroo t herbivorous guild belongs to the soilbiot a (Brussaard, 1998)an d itmainl y includes herbivorous nematodes,mite s and insect larvae.Nematode s may determine the amount of root herbivory on farms (Wardle etal, 1999).Howeve r nematodes andmite s are somuc h interlinked with themicr opredato r food web (presented below) thatw e considered them asa component oftha t assemblage.Roo t herbivorous arthropods in arable agroecosytemsar e mostly polyphagous larvaeo fDiptera , Coleoptera and Lepidoptera. Their food also includes seeds and seedlings (Gange etal, 1991) (Table 1).Feedin g on detritus may also be significant insom etax a(Stron g etal, 1996).Th eroot-herbivor e guildbiomas s (Table2) inth e farm food web is generally larger than thebiomas s of allth e abovegroun d herbivores (Tscharntke & Greiler, 1995;Wardl e etal, 1999),particularl y when leatherjacket s (Tipulaspp. ) are enhanced by gramineous crops.

Table 2 Estimates of the biomass magnitudes (kgdry weightper ha) offunctional groups at organic arablefarms in Flevoland; the estimates are average values with regard to afarm of around 40 ha with 5% ecological infrastructure and 30%gramineous crops. Sources: (a) Van derBurgt, 1998; (b)De Ruiter etal, 1993, 1995; (c)Kromp &Meindl, 1997;(d) Schekkerman, 1997;(e) Smeding etal, 2001b; (f)Buys et al, 1996; (g) Hospers, 1991; (h)Booij, oral communication; (i)Remmelzwaal & Voslamber, 1996; (k)Lange et al., 1994; (m) Smeding & Booij, 2001.

Range of Functional group biomass Reference earthworms 10 1000 a microbivore predators 0.1 - 1 b detritivorous arthopods 0.1 - 10 c root herbivorous arthropods 1 100 d flushfeeding folivores 1 100 c senescense-feeding folivores 0.1 - 10 c, e specialist anthophiles 0.001 - 0.01 f,g granivorous arthropods 0.01 - 1 h herbivorous vertebrates 0.01 - 0.1 i,k epigeic predators 0.1 - 10 c,h oligophagous predators 0.01 - 1 c, e parasitoids 0.001 - 0.01 e soil life-feeding vertebrates 0.01 - 0.1 i ,m insectivorous birds 0.001 - 0.1 i, m top predators 0.001 - 0.01 i

Folivores canb esubdivide d intotw o invertebrate assemblages (Table 1):flus h and senescence feeding folivores. The flush feeding folivores (White, 1978) feed on young tissues with low C/Nrati o and lowfibre content . The characteristic modes of feeding areusuall y sucking, boring andminin g (Crawly, 1983).The yhav e asmal lbod y size, afas t development and good dispersal abilities. Senescence feeders feed onmatur e plant tissues. Originally White (1978) had defined this group as feeding ontissue stha t have increased nitrogen levels caused by stress or ageing. Inthi s articlew e also include folivores which feed on presumably healthy mature tissues which could have ahig h C/N ratio. Characteristic modes of feeding are clipping, holing orremova l (Crawly, 1983). Senescence feeders tend to have aslowe r development or alarge r body size (Crawley, 1983) andpossess , related to amor e sedentary behaviour, defensive traits topredatio n (Power etal, 1996).Th e flush feeders largely determine thebiomas s ofth e folivorous guilds onth e farm; and their exponential increase, dependant onho w susceptible the crop is,an d on weather conditions,ma y temporarily determine the aboveground herbivore biomass onth e farm.

89 Theanthophile s guild (Kevan, 1999)involve s insectswhic h aredependan t on flowers in their entire life cycle (specialists) oronl y inth e adult stage (non-specialists). The former are particularly bees,eithe r solitary (Tscharntke etal, 1998) or social (Svensson etal, 2000). Thelatte r are insectswhos e larvaeca nb eherbivorou s (e.g. butterflies andmoths) , predacious (e.g.syrphi d flies, cantharid beetles),parasitoids , ordetritivorou s (e.g. anthomyid flies). Specialist anthophile biomass,which , inFlevoland , mainly consists ofbumblebees , is small in comparisonwit hth ebiomas s ofothe rherbivorou s guilds(Tabl e2) .

The aboveground granivores guild involves insects,bird s andmammals .Becaus e ofth e difference inthei r life history, vertebrates areplace d inothe r functional groups.Th e granivorous invertebrate functional group include thecarabi d beetle generaAmara an d Harpaluswhic h arenotoriou s granivorous (aswel l asbein g omnivorous); and other granivorous insects likeweevil s and lygeidbug s which are mainly oligophagous (Westerman, inprep.) .

Herbivorous vertebrates onth e farm mainly involve rodents and a few bird species(e.g. pigeon andhous e sparrow).Thei r largebod y size and digestive capabilities allow themt o feed on awid erang e ofplan t tissues. Thedie t of for example thewoo d mouse, an abundant species inFlevolan d field margins, also includes animal food (invertebrates, bird eggs) (Lange etal, 1994). Seeds are animportan t component inmixe d dietsbecaus e ofthei r high nutritional value.

3.1.2. Detritalsubweb primary guilds Thedetrita l food web canb e divided into threemajo r divisions (Brussaard etal, 1997; Brussaard, 1998;Wardl e etal. 1999):th e micro predator food web, saprovorous arthropods andth e soil engineers (i.e.earthworms) .

Themicropredato r food web is in fact acomple x system in itself (DeRuite r etal, 1993, 1996;Brussaard , 1998)comprisin g bacteria-based and fungi-based compartments. Primary consumers aremicrofaun a likeprotozo a andnematodes ;to p species are micropredators comprised ofmesofaun a mites (Acari) and springtails (Collembola).Becaus e of respiration lossesb y several preceding transfers, thebiomas s ofmicropredator s isrelativel y small (Table2) .

The saprovorous arthropod guild involves meso- andmacrofaun a which feed directly onplan t debris ando nmicroflor a which develops ondecayin g tissues. Large springtails, Diptera larvae (e.g. Chironomidae, Sciaridae) and staphylinid beetle adults and larvae are important groups (Sinnige etal, 1992;Webe r etal, 1997;Frouz , 1999;Bohac , 1999).Taxonomi c composition of thisguil d isaffecte d by substrate quality, for example thenitroge n (Bohac, 1999) and sugar content (Weber etal, 1997).

The engineers guild (Brussaard, 1998),althoug h principally saprophagous, eats bulk-soil which may also include other kinds of food (microflora, seeds).Th emajo r characteristic of thisguil d isthei r capability of soil disturbance. Although earthworms play amino r direct role in soilmetabolism , theirbiomas s maypredominat e thetota lbiomas s of detritivorous meso- andmacrofaun a in arablefields (Tabl e2) .

In semi-natural habitats with apermanen t litter layer, itmigh t beconvenien t to distinguish within the saprovorous guild, an epigeic detritivore functional group that involves in particular woodlice (Paoletti &Hassal , 1999),diplopod s (Sinnige etal, 1992), slugs and snails. This group also exhibits transitionsfrom herbivorou s to detritivorous traits, for example abundant slugs feeding on sow thistles that had been weeded out ofth e crop area, 90 andplace d in the crop edge (Smeding,unpublishe d data).Fo rreason s of simplicity,thi s article doesno t distinguish thisgroup .

3.1.3.Predator guilds Sixdistinctiv epredato r functional groupso rguild srepresen t a shift inemphasi s from detritivore to herbivore subweb: - Epigeicpredaciou sbeetle s includingpolyphagou s carabid and staphylinid beetles; - Spiders,bot hwanderin g andweb-buildin g spiders,tha t are ground and facultatively vegetation dwelling polyphagous predators; - Vegetation dwelling arthropod predators including polyphagous species (e.g.canthari d beetles,myri dbugs ) andoligophagou s species.Oligophagou s species in crops are often aphidophagous (e.g. ladybirds, syrphid flies) butno t always so,fo r examplepredaciou s Diptera (e.g. Empididae, Dolichopodidae) which feed on small Diptera and Hymenoptera; - Parasitoids (mainly Hymenoptera) whichd ono tonl yexploit , asi softe n thought, aphids and lepidopterans but also, for example, detritivorous Diptera (Potts &Vickerman , 1974) or spiders (Topping &Sunderland , 1996); - Insectivorous birds including two functional groups: specialist insectivores that are capable of catching flying insects;insecti-granivorou s specieswhic h concentrate on feeding amongth evegetatio n on invertebrates andseeds ; - Soil life-feeding vertebrates, includingbird s (e.g. lapwing and starling) and mammals (e.g.mol e andshrew) .

Forreason s of simplicity it is convenient to combine thepredaciou s beetles and spiders ina guild of'epigeic polyphagous predators'(e.g. Poehling inBooi j &De nNijs , 1996).Th e epigeic predators have ake ypositio n inth e farm food web,i n that they exploit both the herbivore andth edetrita l subweb(Sunderlan d etal, 1996;Kromp , 1999;Wis eet al, 1999). Soalthoug h theybasicall y depend onth e abundance ofth e soil mesofauna, e.g. springtails (Potts &Vickermann , 1974;Hollan d &Thomas , 1997)the y also temporarily exploit the increased herbivorepopulatio n in the summer (Wise etal, 1999;Schoenl y etal, 1996),an d somigh t prevent outbreaks ofr-herbivore s if thesepopulation s build up slowly (Burn, 1988). Insectivorous birds also exploit both subwebsbu t are lessnumerous , compared to the other predator groups. Compared to otherpredaciou s groups,epigei cpredator s may achieveth e highestbiomas s in arable farm food webs (Table2) .

Intraguild predation, including hyperparasitism and cannibalism probably strongly affects the abundance of individual predator guilds (e.g. Tscharntke, 1997; Sunderland etal, 1996). Examples areth epredatio n of alreadyparasitise d aphidsb yvegetatio n dwellingpredators , carabid beetles feeding on spiders,o rinsectivorou s birds feeding on adult syrphid flies. Birds probably consume relatively more arthropod predators thanprimar y invertebrates when compared toth erelativ e abundance ofthes epre y species,becaus e theyma yprefe r larger individuals. Since epigeic predators canhid eunde r the soil surface orhav e nocturnal behaviour (carabid beetles),vegetatio n dwelling arthropods are animportan t prey for birdso n arable farms.

Haematophagous (blood feeding, parasitic) and scatophagous (dung feeding) insectsan d scavengers are small functional groups (e.g. Potts &Vickermann , 1974;Webe r etal, 1997) that exhibit transitionsfrom predaciou s to detritivorous traits.Thi s guild which feeds onth e produce ofvertebrate s may contribute significantly toth epre y of insectivorousbird s (e.g. the swallow).Thi s could occurwher e cattle are temporarily grazing on leypastur e or inth e stable,o rwhe n dung is stored on farm headlands (Hald& Reddersen , 1990; Smeding, unpublished data).Fo rreason s of simplicity we are not considering these functional groups.

91 3.1.4.The toppredator guild Toppredator s are thevertebrate s which aretru ecarnivore s (e.g. kestrel,bar n owl,weasel , stoat) oropportunisti c (e.g. carrion crow).Thei r common characteristic isth e abilityt opre y on other vertebrates andthei rpreferenc e for prey of alarge r size.Som eto p predators,whic h visit the farm by chance,belon g to the food web atth eregio n or landscape level (e.g. hen harrier, blueheron , and gulls).Th e abundance ofherbivorou s vertebrates strongly determines theresourc e availability for thetru ecarnivore s (Plesnik &Dusik , 1994). Thebreedin g density ofkestrel s in Oostelijk Flevoland, inth e years following reclamation, related positively to their food supply which was mainly common voles.However , nestling mortality wasno t affected by adecrease d vole abundance asth enestling s could alsob e fed with chicks,i nparticula r thoseo f starlings(Cave , 1968).

3.1.5.The relative sizes ofherbivore and detritivore subsystems Asi nmos t terrestrial ecosystems,th e farm foodweb consists of aherbivor e subweb and a detrital subweb (Odum, 1973;Poli s &Winemiller , 1996).Foo d chains inth etota l farm food web may,b y chance, achieve alengt h of 9steps .Howeve r thenumbe r of functional trophic levels (Power etal, 1996)i smuc h less,presumabl y approximating the common chain length of4 (Winemiller &Polis , 1996).

In agroecosystems the detrital subweb is expected tob emuc h larger than the herbivore subweb interm s ofbiomas s and species diversity (Wardle etal, 1999; Swift &Anderson , 1993).Thi s could beexplaine d from theviewpoin t of resource availability. In most natural systems, onaverage , about 10%o fne tprimar y production (NPP) is generally takenb y herbivores and about 90%b y ; insect herbivores consume usually less than 1% ofNPP ,wherea s ruminants can easily assimilate moretha n 10%(Crawly , 1983;Edward s& Wratten, 1980). Onarabl e farms the average herbivore consumption might alsob e lesstha n 1%becaus e harvesting removes alarg eproportio n ofNP P and herbivory isinhibite d bybot h thedirec t andindirec t measures ofth e farmer. However even asmal l amount of consumption may causegrea t damage to theplan t (Tscharntke &Greiler , 1995;Crawly , 1983)an d therefore involve large economic loss.

Incontras t to theherbivor e subweb,th e detrital webreceive s extra food interm s of organic manure,organi c wastes and crop residues ofbot h allochthonous and autochthonous origin. However 90-95%o f energy indetritu s isuse d by therespiratio n of soil micro organisms (Swift &Anderson , 1993).Consequentl y thebiomas s ofth emesofaun a (i.e.th eto p organisms ofth emicropredato r food web) isamazingl y small (DeRuite r etal, 1993,1995) (Table2) compare d with,fo r example,earthworms .Th erelativ e difference between the biomass ofmacr ofaun a (i.e.larg einvertebrates ) ofth e detritital subweb compared toth e macrofaun a ofth eherbivor e subweb maytherefor e not bever ymarked . In otherwords : although the size atth e food webbas ema y differ extremely between both subwebs,th e quantitative difference seemst ob emuc h smaller at ahighe r trophic level. This supposition is supportedb y studies ofRyszkowsk i etal. (1993) onth e aboveground insect biomassi n several agricultural landscapes: herbivore subwebbiomas s (including also the adults ofroo t herbivorous larvae) was generally largertha n thebiomas s of insectsoriginatin g from the detritivore subweb. However, comprehensive datao n this subject are scarce and could be biased by seasonal and annual effects.

92 3.2.Farm food webstructure related to farm management

Inth etex tbelow , thefive foo d web structures arepresente d and illustrated with food web diagrams.Firstly , attention isdraw nt o thecro p area contributions to the farm food web; contrasting the food web ofa short organic duration farm and low E.I. quality (typeA ;Fig . la), with the food web of afar m with long organic duration and lowE.I . quality (typeB ;Fig . lb). Next the effect ofimprove d E.I. (i.e. with an increased area andvegetatio n diversity)o n the farm food web isconsidere d by contrasting the above twomodel s without improved E.I. withshor t anda lon gduratio n farms withimprove d E.I.(type sC andD ;Figs , lc and Id).

3.2.1.Short duration farm withoutimproved E.I. The farm food web oftyp e A(Fig . la) represents arelativel y small and low food web structure. The food web is largely determined byth e detrital food web,howeve r soilmeso - andmacrofaun a densities are lowwhe n compared to the long duration farms (types B andD) . After the conversion to organic farming herbivore densities incrop sma y initially remain low.

However, cropsma ybecom e susceptible toherbivor y because the farm hast o adapt toa changed fertility management andth e epigeicpolyphagou spredato rpressur e is relatively weak. In cases ofunbalance d luxurious development ofth ecrop ,herbivores , (including senescence feeding folivores, flush feeding folivores androo therbivores )ma yrapidl y establish inth e crop and inducetemporaril y high vegetation dwellingpredato r densities.

Characteristics of traditionally managed E.I.are : - ahig h abundance of foliar herbivores related toth eprolonge d vegetative development of the gramineous sward; - epigeicpredator s that are attracted byth ewar m microclimate of sparse vegetation; - herbivores (e.g. granivores) relatedt o annual orrudera lplant s in disturbed swards; - specific detritivorous arthropod taxa related torelativel y easily decomposable cuttings.

Arthropod communities oftraditionall y managed E.I.probabl y have less influence onth e crop areatha n arthropod communities ofimprove d E.I.,becaus eo f amor erestricte d food supply (both quantitative and qualitative), andconsequentl y lower densities aswel l asa reduce d variety of functional groups.

Vertebratepredator s mayb emainl ybird s feeding on soil organisms (e.g. lapwing) and foraging insectivorous birds, exploiting accidentalherbivor e outbreaks (e.g.swallow s feeding above infested crops).Fo r insectivorous birdstraditionall y managed E.I.ma y offer a useful but limited quantity of food.

3.2.2. Longorganic duration farms withoutimproved E.I. The farm food webo ftyp e B (Fig. lb) compared totyp eA i scharacterise d by an increased detrital web inth e crop area. Particularly densities of saprovorous arthropods and earthworms might be increased when this iscompare d to short duration farms (e.g.Heimbac h &Garbe , 1996;Idinge r etal, 1997).Th e increase ofthes e groupsma y support, inparticular , epigeic predator functional groupsbu t also somevegetatio n dwelling predators like Empididae that feed on small Diptera. Longduratio n organic farms might possess arelativel y diverse population ofpotentia l croppest s andwee d herbivores, duet o theaccumulate d effect of increased crop andwee d diversity (Hald &Reddersen , 1990;Moreb y & Sotherton, 1997).

93 Top- predators

Epigeic predators

1 Oligo- Wv"" phagous predators \ Parasitoids-

., -Vy \

Earth­ Microbi- Sapro- Root Flush- Senescence Antho- Granivores Vertebrate worms vore vorous herbivores feeding -feeding philes herbivores predators arthropods herbivores herbivores

Detritus Crop plants Weeds Perennial plants E.I.

Fig. la. Predictedfood webstructure for an organic arablefarm withshort organic duration, without an improved ecological infrastructure. Thicklines indicateprobable cases of strong interactions. Continuousthin lines refer to interactions of intermediate strength. Dotted lines refer to interactions of minor importance. Boxes with thick lines indicate a relatively high quantitative importance of thefunctional group within thefarm food web. Top- predators

Soil life- feeding vertebrates

Fig. lb. Predictedfood webstructure for an organic arablefarm with long organic duration, butwithout an improved ecological infrastructure. Symbols as inFig. la. 94 However, therelationship s between andwithi npredato r and herbivore guilds may alternate between two extreme trophic structures.Figs . Id and le illustrate this alternation withregar d to the crop area of farm typeD .Th e extremes depend onth erelativ e strength ofpredatio n pressure by thepolyphagou s epigeicpredator s andth e degree ofpes t susceptibility ofth e crop (Van Emden, 1988),whic h isrelate d toweather , soil nutrients, soil organic matter levels (Phelan, 1997) and interaction between crop and intercrop/weeds (e.g. Theunissen, 1994): - If epigeicpredato r abundance isrelativel y low ori fcrop s are susceptible toherbivory , the flush feeding herbivores maybecom e apes t (Fig. lb).Herbivore s support adivers e predator system because neither epigeicpredator s nor oligophagous predators are dominant. This system may control increase of senescence-feeding herbivores butno t outbreaks ofth emor e vigorous flush feeding herbivores (e.g. aphids).Thes e situations may occurmor e frequently inintensiv e organic croprotations ; - If epigeicpredato r abundance ishig h orth e crop not susceptible,the n allherbivore s and connected epigeicpredator s might depress oligophagous predator numbers.Epigei c predatornumber s areprobabl y limited by food shortages (DenNij s etal, 1997;Booi j et al, 1997)an dma y therefore exert apotentiall y highpredatio n pressure on the herbivore subweb, aprincipl e referred to as'allochthonou s energy subsidy viamult i channel omnivory'b y Wiseet al. (1999) .Herbivore s higher inth e crop canopyma y experience lesspredatio n pressure (Kromp, 1999)an dmigh t bepreye d uponmainl y byparasitoids . Parasitoids mayb erelease d from interference by oligophagous predators (cf.Tscharntke , 1997),o rhav ebette r abilities to cope with lowdensitie s ofpre y (cf.Powel l &Nickless , 1996).Thi s food web structure mayoccu r more frequently in extensive organic crop rotations duet o large detritusresource s and sparser crops.

Withregar d to vertebrate predators, increased densities may occur in soil life-feeding and insectivorous birds when compared to farm typeA .However , inparticular , the situation with increased herbivory inth e crop area may support insectivorous birds,becaus e herbivores are an essential component ofnestlin g diets (Poulsen etal, 1998;Hal d& Reddersen , 1990), whereasth enumerou s epigeicpre y items arehidde n by nocturnal behaviour (Vickerman& Sunderland, 1974),o r shelteredb y physical structures orprotecte db y lowpalatabilit y(e.g. toohar d for the chickst o eat). Therefore the functional food chain length of the farm food webma y get shorter when epigeicpredator s predominate anddepres sth eherbivor e subweb, andreduc e food availability for insectivorous birds.Accordingly , Power etal. (1996 ) hypothesised that food chainsma y shorten in asuccessio n from thepionee r staget oa n established community duet o the increasingly armoured features of sedentary taxa atth e2 nd or 3rd trophic level.

Herbivorous bird andmamma lnumber sma y alsob e supportedb y an increased seed availability ofweed s and cereals onlon gduratio n farms. Asmal l top-predation layerma y subsequently bebase d on anincreas e ofherbivorou s and insectivorousvertebrat eabundance .

3.2.3.Short duration farms withimproved E.I. The farm food web of ashor t duration farm with animprove d E.I. (Type C;Fig . lc) will combineth erelativel y small food webi nth ecro p area(occurrin g on farm type A)wit h an increased and more functionally diverse food web inth e field margins. Theresultin g food web structure maybecom e extended andmor e satiated.

Theeffec t ofimprove dE.I . on the farm food web structure ismos t significant withregar d to resources and connected functional groupstha t are lacking or scarce in the crop area: flowers, seeds, anthophiles, granivores and senescence feeding herbivores. Specific taxao f detritivorous arthropods are supported by thevariet y of litter types.Bot h increased arthropod numbers, aswel l asvarie d vegetation structure, support vegetation dwellingpredators . 95 Primary functional groups and invertebratepredator s maydispers efrom th e E.I. into thecro p (Sotherton, 1994, 1995; Lys& Nentwig , 1992).

Vertebrate functional groups include species thatb ypreferenc e reside inth e E.I., feeding on the locallypresen t seeds and arthropods butprobabl y also on imported flying insects that adheret otal l vegetation.

3.2.4. Longduration farm withimproved E.I. Improved E.I.probabl y adds similar resourcest o the food web of alon g duration farm (Type D;Fig . Id) ast oth efoo d webo fa shor tduratio n farm (TypeC ;Fig . lc). However, ona long duration farm the addition ofherbivor e subweb functional groupsma y further satiate the food web structure and cause cumulative effects. The greater the amount and variety of resources could, for example,giv e abette r absorption byth e food web of fluctuations orinterruption s in the food supply, duet o operations inth ecro p area.I nparticular , populations oflarg emobil e arthropod predators, insectivorous birds andto ppredator s might be supported and stabilised byimprove d E.I. Foodweb s of farms including some degreeo fherbivor y on crops andweed s (Fig. le), are likely to support more insectivorous birds and toppredator s than farms with depressed herbivore subwebsdu et o abundant polyphagous predators orherbivor e resistant crops (Fig. Id) (as explained inparagrap h 3.2.2.). In crop areaswit h little herbivory, insectivorous birds may havet o depend more on improved E.I. for highqualit y sources of prey.

Soil life- feeding vertebrates

Earth­ Microbi- Sapro- worms vore vorous predators anhropods

Detrims Crop plants Weeds Perennial plants E.I.

Fig. !c. Predictedfood webstructure for anorganic arablefarm withshort organic duration including an improved ecological infrastructure. Symbols as inFig. la.

96 Soil life- feeding vertebrates

Crop plants Weeds Perennial plants E.I.

Fig. Id.Predicted food web structure affected by increased detritalsubsidy, for an organic arablefarm with long organic duration including an improved ecological infrastructure. Thefood web structure expresses the indirectlyinhibiting effect of detrital subsidy on the herbivore subsystem. Symbols as in Fig. la.

Top- predators

Soil life- feeding vertebrates

Cropplant s Weeds Perennial plants E.I.

Fig. le. Predictedfood web structure affected byan r-herbivore outbreak,for an organic arablefarm with long organic duration including an improved ecological infrastructure. Thefood web structure expresses the 'bottom up' effect of increased herbivorousprey availability. Symbols as in Fig. la.

97 3.3.Support from empirical research onorganic arablefarms in Flevoland

Resultsfrom empirica l research on arable farms inFlevolan d give some supporting evidence for themodel spresente d above. Observations ofth e four farm types showed: - short organic duration farms (type Aan d C)ha d inth ecro p area low densities of allth e functional groups ofvegetatio n dwelling arthropods or, in more dense luxurious crops,hig h densities of senescence feeding folivores associated with increased predator abundance (Smeding etal., 2001b) .Bir d densities onthes e farms were low compared to other farm types (Smeding &Booij , 2001); - long organic duration farms (typeB andD )ha d in the crop area two distinctive compositions ofvegetatio n dwelling arthropods:on ewit h abundant flush feeding folivores related to intensive crop rotation, and another withmoderatel y abundant detritivores and parasitoids related to extensive crop rotation (Smeding etal., 2001b) . Accordingly epigeic arthropods had alo w abundance inlon g duration farms with intensive crop rotation (Smeding etal., 2001a ) ascompare d to farms with ales s intensive vegetableproduction . However, evidence for increased epigeicpredato r abundance on long duration organic farms with ahig h proportion of gramineous crops,wa sno t found, because these were not well represented in the study area; - longduratio n and improved E.I. farms (typeD) ,includin g mainly intensive vegetable producing farms, had thehighes t territory densities of insectivorous birds,probabl y related to both E.I. and crop characteristics (Smeding &Booij , 2001).Insectivorou s bird abundance seemed tob enegativel y correlated to more extensive croprotations ; - improved E.I.bordere d by large conventionally managed crop plots (resembling type C)ha d high densities of insectivorous bird territories and supported also herbivorous mammals (Remmelzwaal &Voslamber , 1996).

4. Discussion

4.1. Onthe validation ofthe model

Inthi s articlew epropose d atopologica l farm food web (based on the farm food web of organic arable farms in Flevoland, TheNetherlands ) andrelate d this food web to two important farm traits:th eduratio n oforgani c farming andth e amount of ecological infrastructure. This exercise led toth edistinctio n of four basic food web structures.A n additional couldb emad ebase d onth e level of cropherbivor y in long-duration organic farms. Itmigh t be clear that theoutline d responses of the farm food web to farm management werebase d on ami x of field evidence, theory and interpretation. The above presented empirical data illustrate the food webbu t areprincipall y not applicable as validation,becaus e the theory waspartl y inspired by the same data. Further research is required tovalidat e ourmode l food webs.However , wethin k that the ideas that havebee n presented, despite theweaknesse s inthei rvalidation , provide valuable hypotheses and incentives for further research.

4.2. Integration ofecosystem servicesand nature conservation

Theduratio n oforgani c farming aswel l asth e amount of ecological infrastructure affect the farm food web.Bot hvariable s areusefu l for implementing the integration of ecosystem services (Altieri, 1999)an dnatur e conservation-centred aims.However , the effect of duration does not always coincide with the effect of amor e extensive crop rotation (i.e.th e proportion of gramineous crop inth e crop rotation), since current organic practice could include various 98 degrees of intensity.Foo d web structures may differ significantly between extensive and intensive long duration organic farms; difference in 'intensity' among organic farms involves input ofC (particularly crop residues and straw inmanure ) aswel l aswee d control andcro p performance duet o nutrient availability.

Theresult s ofthi s article suggest that food webs that support beneficials (i.e.th e polyphagous epigeicpredators ) mayno t necessarily support conservation-worthy species like insectivorous birds. Longduratio n organic farms with anintensiv e cropping system aswel l asa n improved E.I. seem to combine together, to areasonabl e degree,bot hth e aims of ecosystem services ando fnatur e conservation. However, ifthes e systems weret ob e optimised, the increased employment ofbeneficial s might run counter tonatur e conservation aims.A n increased efficacy ofwee d control inth e future, might also further diminish the food resources inth e organic crop area available for thesebirds .

Thepresente d models ofth e farm food web showed thepositiv e effect of E.I. on nature conservation aims.Thi s effect may occur relatively independentlyfrom organi c duration. It must, however, benote d that short duration farms areno tth e same as conventional farms, wherepesticid e and fertiliser drift might happen (De Snoo, 1995).Indirec t effects of E.I.o n 'ecosystem services'i nth e crop area couldb epresume dbu tma yb e less evident thanth e effects ofE.I . on conservation-worthy species atth e farm level.

4.3Recommendations for research

Hypotheses of further research might concentrate onth e effects ofincrease d detritus supply in the farm food web(e.g. Wise etal, 1999),th e importance of detritivorous mesofauna for epigeicpredator s (e.g. Holland &Thomas , 1997;Idinge r &Kromp , 1996)an dth e relation between herbivory andhighe r trophic levels ofth e farm food web (e.g. Poulsen etal., 1998 ; Tscharntke &Greiler , 1995).Applie d research based on food webmanagement , may address themanagemen t of cropresidue s and othernon-harvestabl ecomponent s (NHC) (Vandermeer etal., 1998) aswel l asth epossibilitie s for controlled herbivory in living mulches, leysetc . (e.g.Theunissen , 1994, 1997).I twoul db eworthwhil et o linkth e farm food webmodel , as proposed inthi s article,t o a farming system model (e.g. Oomen &Habets , 1998;Wolfer t et al, 1997)whic h canbroadl y estimate theresourc e availability of variousdetritu s andplan t tissue qualities,base d on farm management traits.

Finally, also the spatial distribution ofresource s for functional groups should be addressed (e.g. Topping, 1997;Kreus s &Tscharntke , 1994)a swel l asth e effect ofphysica l disturbance and shelter on mortality (e.g. DenNij s etal, 1996) orbir dnestin g success (e.g. Wilson et al, 1997).However , according to authorities in food webresearc h (Poliset al, 1996),whe n referring toth e current state ofknowledge , care isneede d sotha t the complex field study scenario isno t entangled with themethodolog y usedb y the researcher.

Acknowledgements

The authors wish to express their thanks to Wouter Joenje and Kees Booij for discussing the article and suggesting appropriate references andArien a vanBrugge n for commenting onth e manuscript. Thewor k atth e Wageningen University is funded by theMinistr y of Agriculture, Nature Management andFisheries , Department of Science and Knowledge Dissemination.

99 Chapter 7: Farm-Nature Plan: landscape ecology based farm planning

F.W. Smeding3& W . Joenjeb "Department of Plant Sciences, Biological Farming Systems group, Wageningen University, Wageningen, TheNetherlands bDepartment ofPlant Sciences, Cropand WeedEcology group, Wageningen University, Wageningen, TheNetherlands

Abstract A procedure ispresente d for restyling the lay-out and management of farms in order to increase the biodiversity inth e agricultural landscape as well as the farming. The protocol for the development of an on- farm nature management plan uses landscape ecology characteristics, local biotic and abiotic data and potential, as well as the farming system data and farmer's personal interests. Itma y also enhance the farmer's interest and understanding of agroecological pattern and processes and a more conscious approach to elements of nature on the farm, which may be the starting point in ecologising agricultural practice.

1. Introduction

The integration of nature and farming isa n important issue in scientific aswel l aspolitica l circles of industrialized European countries, inth e first place,population s ofman yplan t and animal species associatedwit hagricultura l land are declining inman y Western European countries. Changes inagricultura l land-use areofte n suggested asth emai n cause ofthi s decline (e.g. Wilson etal, 1997;Fulle r etal, 1995;Andrease n etal, 1996).Thes e changes have concomitantly had an effect atth e landscape-level, resulting in alos s of landscape differentiation in Europe (Baldock etal, 1993).

At theIUC N conference 1992o nbiodiversit y inRi o deJaneiro , most European governments committed themselves to implementing biodiversity conservation measures. InTh e Netherlands,thes e commitments areno wbein g translated into deeds.Agriculture , the land- use accounting for about 50%o fth e country's area, isno t excluded from these efforts to conserve biodiversity, sinceth e occurrence ofman y species and communities isrelate d to farmland and farming practices.

Apart from its intrinsic value,biodiversit y has a function in anumbe r of ecological processes highly relevant for food production. These 'life support functions' ofbiodiversit y include for example the soilnutrien t cycles (Brussaard, 1997)an dregulatio n ofpest sb ymean s of biological control (Altieri, 1994).Agro-technolog y hasbroadl y increased the independence of these functions atth e cost ofhig h inputs of fossil fuel, artificial fertilizers andpesticides .Th e concept of'sustainable' agriculture aimst o re-establish an equilibrium: life support functions andtechnolog y should be farm attributes of equal importance.Here , the integration of farming andnatur e comes into play.

From bothperspective s -th e intrinsic and functional valueo fbiodiversit y in agriculture - there isa stron gnee d todevelo p farming systemstha t includebiodiversit y (Almekinderset al, 1995;Vandermee r etal, 1998).Thes e ideas areparticularl y manifest in the current practice oforgani c farming (EEC, 1994).Th e standards for organic farming guarantee safer conditions for biodiversity development bybannin g the application ofpesticide s and artificial 101 fertilizer application andb y encouraging long crop rotation andmoderat e stocking rates. Nevertheless, organic farms also lack aclea r perspective when it comes to integrating nature onth e farm.

Inthi s article,w e arguetha t the integration of farming andnatur e inth e context ofa (including themanagemen t andconservatio n ofbiodiversity) , should comprise two levels:th e landscape level (100-1000 ha) andth e farm level (10-100 ha) (Table 1).A sregard s the landscape level,ther e shouldb e coherence,betwee n the structure and distribution ofth enon-productiv e (green) habitats onth e farm and in the surrounding landscape. Thereaso n istha t conditions supporting biodiversity onth e farm are strongly related toprocesse s and structures atth e landscape level.Th e first question therefore, is:wha t areth e implications ofthi s landscape-level focus for farm-nature? At the farm level, these green habitats should be clearly andpurposefull y integrated with farm management. Only if habitat structures and distribution are compatible with (orbeneficia l to)th e agricultural production, willthe y be ablet o survive and, as aresult , biotic complexity be able to increase.

The second question tob e addressed is: what does this farm-level focus imply for nature on thefarm ? Thethir d question is anextensio n ofth eprecedin g ones:ho wca n these implications beworke d out for real farms, e.g. in a'Farm-Natur e Plan'? How can farms actually be 're-styled' inorde r to achieve abette r coherence with landscape patterns and processes and abette r adjustment of farming tonature ? The outlines of aprocedur e for such anon-far m implementation will bepresented . Thisprotoco l for aFarm-Natur e Plan is currently being tested inDutc h agricultural practice, mainly on organic farms, without subsidies on avoluntar ybasis .

2. Farm in the landscape

Thequestio n ofwh y this 'coherence'betwee n the greenhabitat' s structure and distribution on the farm and atth e landscape level iss o important for thebiodiversit y can only be approached at landscape ecology level andha s to include agroecosystem functioning.

Atth ebeginnin g ofthi s century, traditional farming wasbroadl y in accordance with landscape ecology characteristics andprocesses . The technical and economic constraints to farming were directly related to limiting natural resources and conditions. Crops, livestock, tillage andmanurin g had aspatia l distribution reflecting theprevailin g abiotic conditions. This land-use pattern resulted ina gradien t inus e intensity andproductivit y from the farm settlements onwards. In addition tothi s spatially differentiated land-use, the use of each area was consistently the same over theyears ,i.e. constancy ofprocesses . Thistraditiona l land-use resulted in ahig hbiodiversit y (Westhoff etal, 1970;Va nLeeuwen , 1966). However, increasingly and especially since 1950 improved technology and welfare allowed agricultural land-use to gain independence from natural resources andth e increasing inputs of artificial fertilizer andmechanisatio n have levelled out thehabita t variation. Nevertheless, modern agricultural landscapes stillharbou r abiotic andbioti c diversity. This variation often refers to vital ecological processes,particularl y related toth e eco- (Everts &D eVries , 1991).

The data from plant and animal species inventories are extremely valuable. Because vegetation succession and species immigration mostly take decades to be accomplished (Van Dorp, 1996;Hermy , 1994),i t isdifficul t to conclude thepotentia l suitability of alocatio n for certain species.Therefore , structures andprocesse s atth e surrounding landscape level havet o beuse d asreference s and resources for biodiversity atth e farm level. 102 TableI Important structures andprocesses (italics) on the landscape and thefarm levels as related to ecosystem components.

Component: Landscape level (reference): Farm level (field of operation):

Climate exposure to sun and wind, mesoclimate shade, shelter, erosion, deposition

Soil soil type, inclination, soil/height transitions managing nutrient and organic matter {soil formation) status, tillage, disturbance

Water catchment area (river), water transport: ditches, drains, ponds, wells, water inundation, seepage, infiltration, leaching management or accumulation minerals, land-use planning

Physical structures roads andbanks , dikes,urba n zones, buildings, tracks, fences, dumps/piles, (man-made) construction transports of matter, construction

Vegetation present communities (including , pastures, arable crops, weeds, vegetation of nature reserve),succession, population linear habitats, dispersal, management and dynamics, dispersal control

Fauna vertebrates and migratory invertebrates, present species, domestic animals, migration, dispersal,population dynamics

Howca npatter n andproces s atth e landscape level (Table 1)b e assessed inorde r to achieve such coherence between thehabita t infrastructure of farm and landscape? Ifw e look at the landscape level there arethre e ecological keynotes tob e addressed: - Similarity: thepromotio n on the farm ofplan t and animal populations already occurring in theregion , most effectively enhances the farm biodiversity. Semi-natural (green) habitat types found in the surrounding landscape couldb epresen t or constructed orreinstate d on the farm; a farm in anope n landscape could have asolitar y tree or some shrubs,bu t is better not planted with woodlots orhedges ; - Connectivity: ecologically related habitats (wet, wooded, herbaceous) should be connected orb e located within acertai n distance inorde rt o guarantee aminimu m territory size within thereac h of dependent animal species (Opdam etal., 1993);th eon-far m green habitats should link upwit h the habitat network present inth e surrounding landscape; - Variationpotential: the variation in soil andhydrologica l conditions at the landscape level aswel l asth eprocesse s atwor k on alandscap e scale acting as agents for variation {e.g. causing groundwater seepage zones, flooding) shouldb eutilize d inplannin g habitat arrangement and habitat management at the farm level.

Ifthes ethre ekeynote s are addressed on anumbe r ofcontiguou s farms, their common landscape context will lead to compatible green that willpositivel y influence eachothe r andma y alsorestor e characteristics ofth elandscape . This synergy or positive feedback mayhav eth e character of arestoration , but includes economically feasible farming.

When the three landscape level keynotes are applied toth e ecosystem components atth e farm level, anumbe r ofrecommendation s canb e identified (Table2) : - Climate:genera l exposure to sun andwind , rainfall areth e same for farm and landscape;

103 Soil:differen t soil types andtransition s may extend into the farm area,whic h could be expressed by botanical composition; this canb e enhanced by applying special management on selected (promising) habitats (e.g. establishing grassy strips,removin g biomass after mowing); Water: influx ofnutrient-ric h orpollute d water from the surrounding area may strongly influence on-farm habitat conditions;thi s should berestricte d ifpossible ; differences in groundwater quality extending into the farm area could be expressed in the local species composition (Everts &D eVries , 1991); deepening of aditc h or construction ofa pon d should be linked with the surface-water network; distances up to c.30 0m enable amphibians todisperse ; Vegetation: woody elements (hedgerows, woodlots, individual shrubs and trees),ca nb e planted to expand thewood y network orreinforc e existing patterns or isolated woody elements inth e landscape;managemen t ofwoody ,dr y orwe t herbaceous vegetation should refer to examples in the surrounding areatha t demonstrate subsequent developmental stages and species-rich communities; Fauna: measures for thepromotio n ofcertai n species could focus onresiden t (vulnerable) species;t o avoid disturbance of large species,particularl y meadow birds,thes e efforts couldbes t be at least 200m away from settlements andmai nroads .

3. On-farm nature

Thepreviou s section considered the coherence inpatter n andproces s between the farm and landscape andrecommendation s werederive d for the farm. Now,w e will emphasize pattern andproces s at the farm levelproper , including the related activities andmanagement . The challenge here ist o establish an explicit link between the farming measures and the desired developments in farm nature.

Anumbe r of structures andprocesse s atth e farm level can be important. Features atth e farm level aresummarize d inTabl e 1.Ho wca n these structures and processes be assessed and addressed in aFarm-Natur e Plan? Herew e will use four farm-ecology keynotes comparable toth ethre ekeynote s atth e landscape level: - Surfacearea: acertai n part ofth e farm area should comprise green (i.e.semi-natura l or extensively managed)habitat s and serve as arefug e and asourc e ofvariation . In anumbe r of cases an area of 5% ofth e farm hasbee n recommended (Vereijken etal, 1994); - Connectivity: similar habitats onth efar m (wet,woody , herbaceous) shouldb e connected or located within acertai n distance inorde r to ensuretha t dependent animal species have at least theminimu m territory size and can disperse across the farm (Opdam etal, 1993); - Variation: avariet y ofhabitat s should occur onth e farm; variation canb e added within thepredominan t habitat types by applying different management forms and thus accommodating different plant species and animals that e.g. require variation inpre y and resources;thi shold s true for generalist predators ofpest swh onee d 'alternativeprey ' when pest populations are low (Altieri, 1994).Variatio n also relates directly to biodiversity; - Habitatquality: extensively managed habitats should beprotecte d from manuring and soil disturbance in order to allow communities ofolde r successional stages to develop;t o facilitate vegetation succession and small scale differentiation in ecological gradients there should be asmuc h continuity inth e year-to-year management aspossible , including the useo fth efields (Westhof f etal, 1970); some vertebrate species require additional measures to compensate for orprotec t against threats inth e production area,e.g. protection of meadow bird nests (Beintema, 1991).

104 Table2 Developing a Farm-Nature Plan by translating ecological information (about structures andprocesses) to recommendations.

Component: Landscape level recommendations: Farm level recommendations:

Climate utilize differences in microclimate

Soil utilize relief, soil-type, transitions nutrient application in consistent patterns in space and time; no input outside fields

Water correspond to flow patterns, seepage water manage water quality, quantity and links: quality (i.e. the general ecohydrology) retention, rising water table, creating new ponds and marshes

Physical structures consider vicinity of roads and buildings enable nesting and hiding of animals (disquiet)

Vegetation reference communities under consistent creating and/or linking of similar habitats; (nature) management; improve spatial consistent management method ina linkages or maintain separation consistent distribution pattern; mowing mainly with biomass removal

Animals reference communities, including species facilities and protection measures with a large home range

When the four farm-ecological keynotes are applied toth e ecosystem components atth e farm level, anumbe r ofrecommendation s canb e identified (Table2) : - Microclimate and climate: establishment of sheltered and sunnyplaces ; in open landscapes shelter canb e provided by shrubs,ree d or even highperennia l herbs;ditc h banks orpond swit h asunn y aspect should have priority when extra efforts aremad ei n vegetation management; - Soil and soil fertility: anymanurin g or emission ofnutrient s outside theproductio n area should be avoided. The margins ofpasture s and, ifpossible ,th e edges of arable crops could, therefore, receive lessmanure ; difference in fertility levels of fields should be emphasized; onmargina lfields o n arable farms ales s intensive crop rotation canb e considered; - Farm buildings: access and nesting facilities for birds (swallow, barn owl) and mammals (hedgehog, stoat) could be offered; - Water:th epotentia l for conserving relatively nutrient-poor groundwater and rainwater on the farm should be exploited; sometimes thedirectio n offlo w can bechange d andth e effects ofgoo d water quality extended; the water level in some ditches couldb e raised by damming. If the landscape type permits,we t elements or deeper water canb e created; - Vegetation: ifth e landscape permits new linear woody elements of at least 3m widt ht o be established; therul e of thumbregardin g themovemen t ofbirds ,bat s and butterflies the 100mete r between trees and shrubs;th epatter n ofwet , woody and herbaceous elements around fields should be asinterconnecte d aspossible ; aga p should never exceed 200m (radius of action for stoat and songbirds). Plots should preferably ben o larger than 8ha . There should be arealisti c and not toocomple x annualplannin g of the vegetation management ofverge s and ditch banks,tha tprovide s for differences in growing stage between or even within linear elements; there should be aconsisten t differentiation in land-use patterns, if stocking ratepermits .

105 4. Protocol for aFarm-Natur e Plan

How can organic orothe r farms bemodifie d in order to achieve coherence atth e landscape level and integration ofnatur e atth e farm level? The aboverecommendation s were derived from both levels.Th eFarm-Natur e Planprotoco l (Smeding, 1995)wa s developed for this purpose. Iti s astepwis e procedure that leads to ama p ofth e farm including site-specific recommendations.

TheFarm-Natur e Planprotoco l hasfive steps : - Theattunement: atal k with the farmer about hisperceptio n ofnatur e onth e farm, his wishes,technica l possibilities andconstraints ; - Thesurvey: afield visi t of two dayst o identify and describe all relevant features resulting in (1)a 1:5,000ma p ofth e farm, (2) a 1:25,000ma p ofth e landscape, (3) alis t ofth e different habitat types and their areas and (4) ashor trepor t onobservation s ofhabita t structure and composition; only conspicuous indicator speciesnee d to be identified; information is also collected about land-use planning policy andnatur e management subsidies relevant for the farm; - Theappraisal: an evaluation of thefar m with regard to: (1)th e area of non-farmed habitats, (2)th econnectednes s of similar (wet,wood y or tallherbaceous ) habitats onan d around the farm area, (3)th evariatio n and (4)qualit y of farmed and non-farmed habitats; - The choiceof objectives: adecisio n on site-specific andrealisti c objectives for the farm; the decision isbase d on ecological criteria (appraisal), the farmer's preferences and management possibilities (attunement) and local land-use policy and subsidies (survey); - The design: starting from theobjective s and the farm map several points of attention are checked; atthi s stage,th e list ofrecommendation s (Table 2) isconsulte d to establish landscape coherence andcompatibilit y with farming practices;th eresul t is ama p ofth e farm area including the existing andplanne d distribution ofgree n infrastructure, with practical recommendations; particularly inthi slas t stepth e scientific/ecological judgement ist ob e combined with creativity whenfinding alternative s for acceptable solutions.

Bywa y of anexample , let usconside r theFarm-Natur e Planmad e for the A.P. Minderhoud- hoeveexperimenta l farm (Lantinga &Oomen , 1998),a 8 9h amixe d organic farm in apolde r dominated by arableproduction . Thepla n (Fig. 1)involve d increasing thetota l areao fnon - farmed biotopes from 2.8t o 5.8%.Fou r sown grassy banks are cross-linking the herbaceous network anddividin g extended areas of arablelan d into plots of 9h a orless . Themanagemen t inth e ditches isvaried , with the aim of achieving atal l herb structure (common reed) and apoorl y productive vegetation with increased flowering performance. Thevegetatio n quality ofth e ditch slopes isguarantee d by means of agrass yfield margi n and anoveral l mowing regime with biomass removal. The ditch with tallherb s together with shrubby corners and anewl yplante d hedgerow have toprovid e a shelter network for insectivorous songbirds and small mammals like stoat andweasel . Additional measures are takent oprotec t and facilitate nesting of swallow andblack-taile d Godwit.

Theprotoco l was developed incollaboratio n with the Dutch National Extension Service.Th e procedure was tested in 1995a tthre e organic farms (Smeding, 1995).I n 1996-1997 extensionists made 27Farm-Natur e Plans for Dutch organic farms, ranging from dairy and arablet o fruit production, and distributed overvariou sregion s in theNetherland s (Van Almenkerk &Va n Koesveld, 1997). In 1998 70%o fth e involved farmers responded toa questionnaire: 40%o fth e farmers had implemented the recommendations and 35% intended to do soi nth e future. About 70%o fth einvolve d farmers areusin g the Farm-Nature Plan in discussions withth e local government, water boards andnatur e conservation authorities. 106 Fig. I. Farm-Nature Planfor theA.P. Minderhoudhoeve-WU experimental organicfarm (89 ha), a, arablefield; p, permanent pasture; s.farm buildings; c, canal; r, with verge(s). Measures (width): I,grassbanks (3m); 2, ditch/common reed (5 m); ditch rich inflowering herbs (5 m);4, grassy field margins (3 m); 5,shrub corners (25 m~);6, hedgerow (3 m).

5. Discussion

The development of fanning systems that include biodiversity management is acomple x topic.Th e approach presented inthi s article emphasizes coherence between farm and landscape features and integration ofnon-productio n habitat and farm management. As explained above,thi s approach issubsequentl y elaborated into an advisory instrument (protocol for Farm-Nature Plan)whic h assumes thatth e farmer will collaborate and participate actively.

The approach does not pretend tob efinal an d comprehensive; the Farm-Nature Plan protocol follows only oneparticula r lineo freasoning , applying field experience and scientific evidences mixed with professional judgement. However, the complex system of afar m can also be approached from another perspective, for example,historica l development, spatial relations or from different integration levels (crop,community , region). Certainly, more scientific research could alwaysb edon ebefor e givingpractica l recommendations atth e farm level. However, despite such criticism every alternative line ofplannin g of farm-nature, including the implementation, tends to pose comparable problems.

107 Avaluabl epoin t ofth eapproac hpresente d isth ewell-structure d framework that enablesope n discussion and improvements. Theresult s ofpractice-oriente d scientific research can be fitted into this framework, improving the criteria and guidelines. Important research topics include: the selection of indicator-parameters at different levels, food webs on farms, dispersal requirements for plants and animals, necessity ofplan t species introduction, etc.

A further advantage istha t theprotoco l istransparen t and canb e applied by non-academics, because the emphasis iso nobservatio n of structures and general characteristics of species and vegetation. Last,bu t not least: the farmer's involvement indesignin g aFarm-Natur e Plan is essential. Theresult s are strongly dependent onth e farmer's good intentions, understanding and accuracy in timing andpositionin g of actions. Theprotoco l and Farm-Nature Plan seem to be effective tools inrestorin g landscape and farming sustainability. Aswel l ashavin g a high-tech context,precisio n farming (Rabbinge, 1997)ma y also refer to akee n understanding andus eo f ecological patterns andprocesse s and atoleran t attitude towards nature elements onth e farm: managingbiodiversit y asth e startingpoin t in ecologising agriculture.

108 Chapter 8:Genera l discussion

1.Observation s andthei r implications from afoo d web perspective

1.1. Empiricaldata on food web structures

1.1.1. Initial hypotheses Itwa s expected that the duration of organicmanagement , anextensiv e crop rotation and an improved ecological infrastructure (E.I.) {i.e. enlarged area, aregim e oflat emowing) ,woul d relatepositivel y to the abundance ofnon-pes t arthropod functional groups, including primary levelso fth edetrita l aswel l asth eherbivor e subweb.Th e crop areawa s expected to support inparticula r thedetritivore s inth e farm food web, andth eE.I . was expected to support particularly the (K-)herbivore functional groupstha t mightb e scarce in the crop area.A n increase ofth eprimar y functional groups would consequently enhance invertebrate predator andparasitoi d numbers aswel l asth enumber s ofinsectivorou s vertebrates (i.e.birds )tha t feed onbot h primary andpredaciou s groups.Th e increased abundance ofth e predator complex was expected to inhibit outbreaks ofpes t herbivores.

1.1.2. Observations inthe crop area Field studies on organic arable farms inFlevolan d indicated that all the factors included inth e abovehypothesis , affected abundance and functional group composition ofbot h arthropods andbirds .Howeve r the effects differed from what was expected.

Inth e crop area (i.e.whea t crops),K-herbivor e abundance was associated with farms that had been recently converted, while r-herbivore abundance was associated with farms of alonge r organic duration with an intensive croprotation . Both observed herbivore increases might be explained by acombinatio n ofcro p susceptibility toth eherbivor e group and limited predation pressure (Smeding etal, 2001b).Ther ewas , according to thehypothesis , some evidence found for detritivore increase related toth e extensive crop rotations on long duration farms (Smeding etal, 2001b), and for predator increases associated with detrital subweb onth e more extensive farms (Smeding etal, 2001a). Inthi s lattercase ,a broade r definition of 'extensive'wa s applied, involving both crop rotation and farming practices. Predator abundance in the crop canopy was particularly related to high K-herbivore numbers on recently converted farms; however the highest species diversity was found on long duration farms which had an intensive crop rotation (Smeding etal, 2001b).Th e farms which had an extensive crop rotation compared with the other farms had alo wpredato r number inth e crop canopy, whereas theparasitoid s had intermediate numbers.Th edensit y ofbirds ' territories seemed tob e enhanced by the crop areas of longduratio n organic farms with an intensive crop rotation. Thiswa spossibl y related to increased herbivory (Smeding &Booij , 2001).

1.1.3.Observations inthe ecological infrastructure Increased arthropod abundance was found inth evegetatio n ofth e improved E.I. (Smedinget al, 2001b).Th e increased functional groups includedbot hpredator s and detritivores. However K-herbivore numbers did not increase inth e improved E.I. when they were compared toth e traditionally managed E.I.;perhap sbecaus e theprolonge d vegetative development of frequently mown vegetation might extend the availability ofnutritiv e grass tissue. Also,th e abundance of epigeicpredators , inparticula r ofth e quantitatively important group ofpredaciou s beetles,wa s not related to an improved E.I. (Smeding &Booij , 2001). Observations suggested that their numbers wererelate d toth ecro p areaconditions . Open, frequently mown E.I. vegetation enhanced the more specific ground dwelling functional

109 groups compared to improved E.I. Theterritoria l density of insectivorous birds was positively related toth ehig h arthropod abundance inth e E.I. (Smeding &Booij , 2001).

1.1.4. Observations onthe interactions between theE.I. andthe crop area Field studiesprovide d no convincing evidence for thedispersa l of functional groups, abundant inth e E.I., into the crop area.Th e effects of cropcondition s onth e arthropod functional group abundance inth e crop area seemed tooffse t the influence ofth eE.I . (Smeding etai, 2001b).Contrar y to expectations, observations suggested the effects were reversed regarding epigeic predacious beetles (Smeding etai, 2001a), and flying insectstha t werepossibl y transported intoth eE.I . (Smeding etal, 2001b; Smeding &Booij , 2001).

Bird studies revealed positive correlation between bird functional groups and a combination of crop area and E.I. characteristics (Smeding &Booij , 2001).Thi s may suggest thatbot h the crop area andth e E.I. habitats are important to thesebirds .Thi sma y also hold true for birds that typically live inth e E.I.,becaus e acertai nproportio n ofthei r prey could originate inth e crop area.

1.2.Observations relatedto the food web theory

1.2.1.Top downand bottom up determination ofabundance The initial set ofhypothese s wasbase d onth e generalisation that anincrease d food webbase , duet o anenlarge d resource supply, relates toth e subsequently increased trophic layers ('bottom upcontrol' , cf Oksanenet ai, 1996).Ou r findings largely confirm the importance of "bottomup 'determinatio n ondetritivores , K-herbivores, r-herbivores, invertebrate predators and insectivorous birds.However , theplac e ofth eresource s was different than expected: long duration organic farms did not provide alarg e food resource to K-herbivores and incas e ofa n intensive crop rotation also not to detritivores. Inth e E.I., contrary to expectation detritivorous arthropod numbers were enhanced by latemowin g and not particularly K- herbivore numbers.

Additional to indications for 'bottom up'determination , indications were found for 'topdown ' effects {cf Wise etai, 1999;Tscharntke , 1997;Powe r etal, 1996). Observations ofbot h the K- and r-herbivore abundance suggested that the explanation ofherbivor e abundance might, next toth eherbivor e susceptibility ofth ecrop ,als o involveherbivor e control bypredators . An increase ofpredatio n pressure on farms, along an intensity-gradient beginning with recently converted farms, may first depress K-herbivore numbers (on intensive farms of long duration) andthe n also themor e vigorousr-herbivore s (on extensive farms of longduration) . Thispresume d predation pressure couldb e due to epigeic predators that are supported by subsidised detrital food chains (Wise etal, 1999;Poli s& Hurd , 1996).Th epotentia l strength of suchpressur e is suggested by several reports on food shortages for epigeic predators (Sunderland etai, 1996;Booi j etal., 1996;Wis e etal, 1999).Thes e 'top down' effects may modify to acertai n extent the generally 'bottom up'determine d farm food web structure. However, muchmor eresearc h isrequire d to find conclusive answers.

Thepredominanc e of aparticula r predator functional groupma y affect other predator groups by means ofpredatio n aswel l as competition (e.g. Tscharntke, 1997).Fo r example,a n abundance ofcarabid s or linyphiid spiders might affect syrphid fly larvae (oligophagous predator) andparasitoids ,whic h also interact with each other. Consequently parasitoid distribution might to acertai n extent, bedetermine d byth e occurrence ofpredator s competing with orpreyin g onparasitoids . Theseinteraction s (i.e.'to p down' and 'horizontal effects' inth e food web) defined as'predato r interference' or'intraguil d predation' may influence arthropod 110 functional group compositions. Theabundanc eo f insectivorousbird smigh tb enegativel y affected by the 'multi-channel' omnivoryo f epigeicpredators , sinceherbivores , especially K-herbivores are the preferred food items for insectivorous birds (Poulsen etal, 1998).Th ebird s feed, to acertai n extent, on ahighe r trophic level than arthropod predators; areductio n ofbir d density therefore implies a shortening ofth e functional chain length ofth e farm food web.Accordingly , Power etal. (1996)argue d that food chain lengthwil lge t shorter in asuccessio n from thepionee r staget o the secondary stagebecaus e the grazers or secondary consumers possess armoured traits protecting againstpredation . Thisviewpoin t would beth erevers e ofth ecommonl y held view by ecologists that therei s asuccessio n towards an increased complexity.

Thehighes t bird densities were observed on farms that combined intensive crop production with improved E.I. This observation suggested, according to thehypothesis , that the broadening ofth e food webbas e may facilitate ('bottom up1)highe r trophic levels.Th e negative indirect ('top down') effect of detrital subsidy onth e food availability tobird s atth e farm level,migh t be compensated by food availability tobird s inothe r habitats onth e farm (e.g.i n E.I.) and around the farm.

1.3. Extrapolationof Flevoland observations toother landscapes

The comparative study of arthropod communities andbird s on four organic arable farms in Flevoland and therive r region (Smeding &Booij , 1999;De n Nijs etal., 1998) demonstrated mainly contrasts between the landscapes.Th erelatio n ofmos t invertebrate functional groups to E.I. seemed tob e ambiguous.However , the farm typewit h increased E.I. in each landscape was associated with high densities of specialist insectivorous birds.

Observations suggested that wheat crops inth e river region compared to wheat crops in Flevoland included amuc hhighe r speciesdiversity , involving for example an apparent abundance and diversity ofweeds . Abundant flowering weeds like camomile, attracted large numbers of anthophiles and also for example large-sizedpredaciou s Diptera (asilid flies) that did not occur inFlevoland ; the territory of a stonechat (Saxicola torquata) on farm 'O-',a n insectivorous bird, and the weed Venus' looking-glass (Legousia speculum-veneris) on farm 'O+'represente d highnatur e conservation values (Smeding,unpublishe d data).Th e observed increase in speciesrichnes s and associated complexity of farmland communities on organic farms inol d culture landscapema yrequir e amor edetaile d assessment of functional group assemblages inth e farm food web.

The increased food availability tobird s inth eriver regio nwa s suggested by the total songbird territory density: 10-12territories/1 0ha ,whic hwa s twot o sixtime s more than onte n selected Flevoland farms in the 1997-1999observations . In one ofbot h river region farms, skylarks had adensit y of2. 7territories/1 0ha ,whic h was higher than thehighes t observed density (2.4 territories/10ha ) in Flevoland. In farm food webs inol d culture landscape comparedt o polders, resources supplied byweeds ,bot h directly (seeds) and indirectly (herbivores), might haveparticula r importance. Also contributions ofth eE.I . toth e farm food web inol d culture landscape,ma y not only include organisms from herbaceous vegetation but also from habitats (e.g.woody , aquatic)tha t aremor econtrastin g to farmland communities.

It canb e argued that itwoul d be more difficult to assess (i.e.isolate ) the effects of farm management (e.g.involvin g detritis input in the crop area) in old culture landscape than in polders. More complex systems may involve more extensive ecosystem services and nature conservation contributions that dono t (yet)occu r in simpler agroecosystems (Brown, 1999a; 111 Altieri, 1994).Studie s should, therefore, sooner or later, also addressth emor e complex farmland communities inmor ebiodivers e landscapes (Stobbelaar &Va n Mansvelt, 2000; Edwards etal., 1999;Vandermee r etal., 1998) .

1.4. Implications

1.4.1. Implicationsfor improvement ofecosystem serviceand nature conservation aims Withregar d to improving theutilisatio n ofecosyste m services (Altieri, 1999),ou r results implytha t enhancing thepre y availability for epigeic predators,b y subsidising the detrital subweb,migh t contribute topes t control on organic arable farms. Thismechanis m may alreadyb ea ninfluenc e on longduratio n organic farms with extensive crop rotation (including a leypasture) . However, further research onthi s topic isnecessar y and promising. Increasing the abundance ofvegetatio n dwelling (oligophagous) predators inth e crop canopy, might not be anappropriat e aim,becaus e the great abundance of cropherbivore s would be needed to increase thesepredators . This isi n agreement with reports on the limited predation capacity ofoligophagou s predators (e.g. Coccinellidaei nHemptinn e& Dixon , 1997).

Withreferenc e toth e nature conservation goals,result s suggest that improvement ofE.I . enhancesth edensit y ofinsectivorou sbirds ,whic h areth emai nconservation-worth yspecie s on arable farms. Thecro p areas of intensive organic arable farms supported relatively high densities of insectivorous birds (when compared with recently converted farms), possibly related toth e occurrence ofr-herbivores .

However, ifthes e systemswer et ob e optimised, in accordance withth eprinciple s outlined above,th eincrease d employment ofepigei cbeneficial s might run counter to highbir d densities i.e. nature conservation values.A n increased efficacy ofwee d control inth e future, might further diminish the food resources inth e crop area available for these birds.I tmigh t appear curioustha t current organic farming systems inFlevolan d with an intensive cropping system (and particularly when there is also animprove d E.I.) combine together, toa reasonable degree,bot h aims ofprovidin g ecosystem services and nature conservation, but that further improvement oforgani c farming systems might reduce this integration.

1.4.2. Furtheragroecological implications offood web management Theimportanc e ofK-herbivore s aspre y for insectivorous birds aswel l asth e importance of herbivorous vertebrates for toppredator s suggeststh erelevance , from an energypoin t of view,o frelativel y small food chains('flows' ) inth e farm food web.Probabl y most consumption by herbivores iscarrie d outb y small organisms liker-herbivore s (i.e.aphids ) (Smeding &D e Snoo,2001) .Becaus e 90%o fenerg y is lost in each transfer inth e food chain (Polis& Winemiller , 1996),foo d chainstha t start from smallprimar y organisms, dissipate more energy,befor e tissues are assimilated into the larger organism-'meat';primar y consumption by largeherbivore s (e.g. leatherjackets , chafers, grasshoppers, sawfly larvae, caterpillars,mice ) might bemor e effective for thehighe r trophic levels ofth e farm food web. Inth e detrital food web asimila r arbitrary division between the small primary organism chains and largeprimar y organism chains could bemade :micro-organis m based chains (De Ruiter etal., 1995)versu s saprovorous arthropods (e.g. flies, large springtails) and earthworms based chains,whic h connect tovertebrates , for example, lapwing and insectivorous mammals.

Foodwe bmanagement , which is aimed atnatur econservatio n goals,ma y address,i n particular, the large organism chains. This could be doneb y increasing the amount ofles s nutritive andmor e complex plant tissues, aswel l asth e detritus that does not decompose 112 easily inbot h thecro p area and the E.I.;thes eresource s require the employment ofth e larger organisms mentioned above.Accordingly , Ryszkowski etal. (1993 ) found more large species in arable crops inmosai c landscapes {i.e. including ahig h amount of'E.I.') than in arable crops in uniform landscapes.

Cousins (1996), whowa s asourc eo f inspiration for the chapter on food web structure, stressed the importance ofweightin g the food 'value' ina food web.Fro m this viewpoint, for example seeds and flower products, although principally belonging toprimar y production, mayb epu t onth e sametrophi c level asaphids .Accordingly , detritus with ahighe r fibre content, mayhav e ahighe r 'value'tha n easily decomposable material which is immediately decomposed by themicropredato r system.However , theseprinciple s should not be applied toorigorously ; for example, Steffan-Dewenter &Tscharntk e (1997) found more large sized catterpillars inpionee r successional fields than in late successional fields. Another aspect is that larger organisms often have longer life cycles;therefor e alsophysica l disturbance and shelterpossiblities , onbot hfield an d farm level, should be considered {e.g. Ryszkowski et al., 1993;Brown , 1999b).

Themanagemen t ofcomponent s that areno tharveste d (NHC)(Vandermee ret al, 1998), included inth e concept ofmulti-specie s farms,provide s aconceptua l basis for a food web management that specifically addresses the 'largeprimar y organism chains'. Central toth e management ofNH C isth e recognition ofth e importance of cropresidues ,ha y from thecana l banks, andothe rnon-harveste d materials (including thestandin g cropo fwil d vegetation and trees) for the farming system and its food web.Primar y production capacity, which doesno t feature inth e economic returns ofth e farm {e.g. greenmanure ,E.I . vegetation) mightb euse d asa lif e support for thenon-pes t herbivore subweb and subsequently support the detritivore subwebmacrofauna . Betterhabita t conditions for theseherbivore s maycompensat et o some extent for thepredatio n pressure by epigeicpredators , due to detrital subsidy. In sucha n extended food web structure, ecosystem services andnatur e conservation might againb e integrated. However, the question whether complex food webs arebette r i.e. more stabletha n simple food webs isa perennia l topic for discussion among ecologists (Polis& Winemiller, 1996),an dmigh t bemor e difficult to approach than otherquestion s raised by this paragraph.

Swift &Anderso n (1993) aswel l asWardl e (1999) arguedtha t thedetrita l subweb on arable farms ismor e developed and less depleted than theherbivor e subweb. Our studies,however , suggest, inreferenc e to the above ground community, that therei s an equal significance for bothth eherbivor e subweb and the detrital subweb (Smeding &D e Snoo,2001) : agroecosystemsi nwhic h all theNH C enter the detritivorous food chainsma yno t support highdensitie s ofvertebrates . From afoo d webperspectiv e itmigh t therefore be achalleng e to incorporate controlled herbivory inmulti-specie s farming systems (e.g. Theunissen, 1994, 1997;Andow , 1988;Altieri , 1994;Hollan d &Fahrig ,2000) .

2. Recommendations

2.1Improvement of topological web

The assemblage of functional groups andth e identification of interactionsbetwee n these functional groups are the empirical foundation of food web studies (Winemiller &Polis , 1996).Peculiaritie s of species may, in some cases,determin e food web structure. Therefore it is important to elaborate onthi s foundation by extracting information ondiet , habitat preferences and life history shifts from thevas t amount of available literature {e.g. Paoletti, 1999).Thi s information iscurrentl y being collected in adatabas e AURYN (Booij,Loc k and 113 Smeding, unpublished data),an d will beuse d toupdat e the topological food webs presented inthi sthesis .

2.2. Linkingfood Miebs tofarm models

The changes of farm food web structure related to farm and E.I. characteristics (Smeding& De Snoo,2001 )coul db edescribe d by amathematica l model. Inth e exploratory phase of food webmanagement , simplemodel sma y helpt o explore thehypothese s and to detect theoretical inconsistency. Apreliminar y model isbein g developed using Fuzzy Cognitive Mapping (Wolfert, in prep),whic h can calculate,base d on certain farm and E.I.traits ,relativ e sizes of the functional groups in the farm food web.

Food web analysis may inth e future profit from extensive farm models that calculate the availability ofresource s for theprimar y functional groupsi nth e farm food web(e.g. Oomen & Habets, 1998; Wolfert, 1998; Wolfert etal., 1997). Thesemodel s could estimate,base d on datao fth ecro protatio n andmanur e strategy, theamount s ofvariou stype s ofprimar y production and crop residues (NHC) that areneede d to support theprimar y consumption in the food web. Suchmodel scoul d beuse dt odesig nrealisti c experimental and commercial farms, with for example optimised detrital subsidy.

The subject ofthi s thesis was explorative research describing arthropod and bird communities that gave an indication of food web structure.However , inth e future morequantitativ e data sets andbette r understanding ofprocesse s in farm food webs are expected to deliver useful simulation models.

2.3Spatial approaches

Thespatia l approach (e.g. Opdam, 1986;Wratte n &Thomas , 1990;De nBelde r etal., in prep.) and the food web approach arebot h useful as aconceptua l framework for agroecological studies atth ecommunit y level (Van Wingerden &Booij , 1999).Authoritie s of the food web approach (Poliset al, 1996;Winemille r &Polis , 1996) arguetha tbot h approaches should becombined ,bu t alsonot etha tth eresultin g complexity couldb ea seriou s constraint. Accordingly, connecting spatial models to important functional groups inth e food web model willpos e greatproblems ,bot h intechnica l andmethodologica l respect. However, itmigh t be achalleng e for both food web and spatial approaches to experiment with rather simple comprehensive models (e.g. Kreuss &Tscharntke , 1994).Th e effect on the farm food web structure ofth eE.I . arthropod community, and also of cropmosaics , exemplify spatial heterogeneity. Accounting for the mobility of functional groups isnecessar y for the successful application of food web management.

The spatial approach,base d onth edispersa l ofindividual s andpopulation s (e.g. Topping, 1997)i sessentiall y a'botto m up'approach . However, characteristics on ahighe r integration level should alsob e approached withth eappropriat emethod s for thishighe r level (Almekinders etal., 1995;Cousins , 1996).Example s ofa spatial approach that addresses higherintegratio n levels areprovide d byth ewor ko fOldema n (1990),analysin g spatial patterns based onth e distinction of'eco-units'.

114 2.4.Field experiments

Field research and experimentation arerequire d to investigate thepreliminar y conclusions provided by this thesis.A stud y areamigh t include both commercial farms aswel l as experimental farms. The experimental farms could represent extreme regimes of management which areno tye t acceptable for farming practice.Car eshoul d betake nwit h experiments that involveplot swher ether e areman y smallfields, whic h may represent amosai c from a farm levelperspective . The context and scale of a(farming ) system may in some casesb ea precondition for theoccurrenc e ofparticula r effects (e.g. Pimm, 1991;Brown , 1999a).

Comprehensive field studieswit h indiscriminate assessment ofman y taxa, asdon eb y Potts& Vickerman(1974) ,ma ymode lfield researc htha t couldoptimall y support food webstudies .

2.5.Practical recommendations

Procedures for farm-nature planswer erecentl y elaborated by Van Almenkerk &Va n Koesveld (1997), Visser (2000) andDaeme n(2000) .Protocol s orprocedure s for such combined agricultural and ecological plansmigh t be supported by arestricte d food web assessment based on simple assumptions.Fo r example, asimpl emode l that applies datafrom the farm map (areas of cropsan dhabitats) , coarsemanurin g strategy, and E.I. management, mayprovid e estimates ofresourc e availability for several functional groupso n the farm. These outcomes may help the farmer or advisert o setpriorities ,wit hregar d to habitats or species groupstha t could be and aret ob e enhanced.

115 References

Aebischer, N.J., 1991.Twent y yearso fmonitorin g invertebrates andweed s incerea l fields in Sussex. In: L.G. Firbank, N. Carter, J.F.Darbyshire , J.F.Pott s (eds),Th e 32nd Symposium of theBritis h Ecological Society andth e Association ofApplie d Biologists,Universit y of Cambridge,Blackwel l Scientific Publications, Oxford, 305-331. Afifi, A.A.,Clark , V., 1984.Computer-aide d Multivariate Analysis.Va nNostran d Reinhold, New York, 458pp . Almekinders, C.J.M.,Fresco , L.O., Struik, P.C., 1995.Th enee d to studyan d manage variation in agro-ecosystems.Netherland s Journal ofAgricultura l Science 43, 127-142. Almenkerk,J.J .van , Koesveld,F .van , 1997.He t agrarisch natuurplan, een innovatieve voorlichtingsmethode (projectdocument). Biologica, Organisation for Organic farming and Trade,Utrecht . Altieri, M.A., 1994.Biodiversit y andPes tmanagemen t inAgroecosystems . Food Products Press,Ne w York, 185 pp. Altieri,M.A. , 1999.Th e ecological roleo fbiodiversit y in agroecosystems. Agriculture Ecosystems andEnvironmen t 74,19-32. Andow,D.A. , 1988.Managemen t ofweed s for insect manipulation in agroecosystems. In: M.A. Altieri &M . Liebman, WeedManagemen t inAgroecosystems : Ecological Approaches, CRC Press,Boc aRaton ,Florid aUSA , 265-301. Andreasen, C, Stryhn, H., Streibig,J.C. , 1996.Declin e ofth e flora inDanis h arable fields. Journal of 33, 619-626. Andrewartha, H.G., Birch, L.C., 1984.Th e ecological web -Mor e onth e distribution and abundance of animals. The University ofChicag oPress ,Chicago/London , 506pp . Anonymous, 1991.Counci l Regulation (EEC) No 2092/91 of2 4 June 1991o n organic production of agricultural products and indications referring thereto on agricultural foodstuffs, Official J.L.I98, 1-15. Bal, D.,Beye ,H.M. ,Hoogeveen , Y.R., 1995.Handboe kNatuurdoeltypen . Rapportnr . 11, Informatie enKennisCentru m Natuurbeheer, Wageningen, 408pp . Baldock, D.,Beaufoy , G., Bennett, G., Clark, J., 1993. Natureconservatio n andne w directions inth e EC Common Agricultural Policy. Report for the Ministry ofAgriculture , NatureManagemen t andFisheries , TheNetherlands . Institute for European Environmental Policy, London and Arnhem, 224pp . Barker, A.M. &Reynolds , C.J.M., 1999.Th evalue so fplante d grass fields margins asa habitat for sawlies and other chick-food insects.Aspect s of Applied Biology 54, 109-116. Beintema, A.J., 1991. Breeding ecology ofmeado w birds (Charadriiformes) ; implications for conservation and management. Ph.D.Thesis,Universit y of Groningen, Groningen, 127pp . Beintema, A.J.,Moedt , O.,Ellinger , D., 1995.Ecologisch e atlas van deNederlands e weidevogels. Schuyt &Co ,Haarlem , 352pp . Beintema, A.J., Thissen, J.B.,Tensen , D.,Visser , G.H., 1991. Feeding ecology of charadriiform chicks in agricultural . Ardea 79,31-44 . Belder, E.den , Elderson, J., Brink, W.J. van den, Schelling, G., (inprep. ) Effect of woodlots ononio n thrips density in leek fields: alandscap e analysis.

117 Boatman, N.D.,Davies ,D.H.K. , Chaney, K., Feber, R., Snoo, G.R. de, Sparks,T.H . (eds), 1999.Fiel d margins andbuffe r zones:ecology , management and policy. Aspects of Applied Biology 54,37 4pp . Bockemuhl, J., 1986.Dyin gForest s- a crisesi nconsciousness . Social Ecology Series. Hawthorn Press, Gloucestershire UK, 96pp . Bohac,J. , 1999. Staphylinid beetles asbioindicators . Agriculture Ecosystems and Environment 74,357-372 . Booij, K., 1994.Diversit y patterns in carabid assemblages inrelatio n to crops and farming systems. In:Desende r etal. (eds) ,Carabi d beetles:Ecolog y and Evolution, Kluwer Academic Publishers, TheNetherlands , 425-431. Booij, K., Nijs, L.den , 1996.Arthropo d natural enemies in arable land, II Survival, Reproduction and Enhancement. ActaJutlandic aLXXI :2 ,natura l science series 10. AarhusUniversit y Press,Denmark , 278pp . Booij, C.J.H.,Noorlander , J., 1992.Farmin g systems and insect predators. Agriculture, Ecosystems and Environment 40,125-135 . Booij, C.J.H., Topping, C, Szysko,J. , Dijk, Th. van,Paoletti ,M. , Helenius, J., 1996. Assessment of survival andmortalit y infiel d populations ofbeneficia l arthropods.Act a Jutlandica LXXI:2 ,natura l science series 10,Aarhu sUniversit y Press,Denmark ,69-78 . Braae, L.Nehr , H.,Petersen , B.S., 1988.Fuglefaunae n pakonventionell e ogokologisk e landbrug. [acomparativ e study ofth ebir d fauna in conventionally and organically farmed areas with special reference toth e effects ofpesticides] . Miljoprojekt, 102.Danis h Evironmental Protection Agency, Copenhagen. Brown, M.W., 1999a. Applying principles of community ecology topes t management in orchards. Agriculture, Ecosystems and Environment 73, 103-106. Brown, R.W., 1999b.Gras smargin s andearthwor m activity in organic and integrated systems. Aspects of Applied Biology 54,207-210 . Brussaard, L., 1998.Soi l fauna, guilds, functional groups and ecosystem processes. Applied Soil Ecology 9, 123-135. Brussaard, L., Behan-Pelletier, V.M., Bignell,D.E. , Brown, V.K., Didden, W.A.M., Folgarait, P.J., Fragoso, C, Freckman,D.W. , Gupta, V.V.S.R., Hattori,T. ,Hawksworth , D.L., Klopatek, C, Lavelle,P. ,Walloch , D.,Rusek , J., Soderstrom, B.,Tiedje , J.M., Virginia, R.A., 1997.Biodiversit y and ecosystem functioning in soil.Ambi o26(8) ,563-570 . Brussaard, L.,Veen ,J.A .van ,Kooistra , M.J., Lebbink, G., 1988.Th edutc hprogramm e on soil ecology of arablefarmin g systems I. objectives, approach andpreliminar y results. Eological Bulletins 39,35-40 . Btichs, W.Harenberg , A.,Zimmerman , J., 1997.Th e invertebrate ecology of farmland asa mirror ofth e intensity ofth e impact ofman ?-A n approach to interpreting results of field experiments carried out in different crop management intensities of asuga rbee t and anoi l seed raperotatio n including set-aside. Biological Agriculture &Horticultur e 15,83-108 . Burgt, G.J. van der, 1997.Inventarisati e regenwormen AKC-bedrijven. Intern rapport, Louis BolkInstituut , Driebergen NL,2 5pp . Burn, A.J., 1988.Effect s of scale on themeasuremen t ofpesticid e effects on invertebrate predators and predation. In:M.P . Greaves,P.W . Greig-Smith &B.D . Smith, Field methods for thestud yo f environmental effects ofpesticides ,Monograp h 40,Britis h Crop Protection Council, 109-177

118 Buys, J.C., Oosterveld,E.B. ,Ellenbroek , F.M., 1996.Kanse n voornatuu rbi j braaklegging II -Verslag van een tweejarig praktijkonderzoek, CLM 253-1996, Centrum voor landbouw en Milieu,Utrecht , 163pp . Canters,K.J. , Dormer,J.H. ,Poll ,R.J . van der, Tamis, W.L.M., 1997.Wa t komt erin enwa t gaat emit. CMLrappor t 132,Centru m voor Milieukunde, Rijksuniversiteit Leiden, 27pp . Cave,A.J. , 1968.Th ebreedin g ofth e kestrelFalco tinnunculus L., inth ereclaime d area Oostelijk Flevoland. Ph.D.Thesis ,Rijksuniversitei t Leiden, 404pp . Chamberlain, D.E., Wilson, J.D.,Fuller , R.J., 1999.A compariso n ofbir d populations on organic and conventional farm systems in Southern Britain. Biological Consrvation 88, 307-320. Cohen,J.E. , Schoenly, K., Heong,K.L. ,Justo ,H. ,Arida , Barrion, A.T., Litsinger, J.A., 1994. A food-web approach to evaluate the impact of insecticide spraying on insect pest population dynamics in aPhilippin e irrigated rice ecosystem. Journal ofApplie d Ecology 31, 747-763. Cousins, S.H., 1990. Countable ecosystems deriving from ane w food web entity. Oikos57 , 270-275. Cousins, S.H., 1996.Foo dwebs : from the Lindeman paradigm to ataxonomi c general theory of ecology. In:G.A . Polis & K.O. Winemiller,Foo d webs: integration ofpattern s and dynamics. Chapman &Hall ,Ne w York, 243-254. Cramp, S.(ed) , 1988.Handboo k of theBird s ofEurope ,th eMiddl e East andNort h Africa. TheBird s ofth e Western Palearctic. Oxford University Press. Crawly, M.J., 1983. Herbivory -The dynamics of animal-plant interactions. Studies in Ecology 10,Blackwel l Scientific Publications, Oxford/London, 437pp . Curry, J.P., 1994.Grasslan d invertebrates: ecology, influence on soil fertility, and effects on plant growth. Chapman &Hall , London, 437pp . Daemen,N. ,2000 .Bedrijfsnatuurplanne n -een onderzoek envoorste l voor een standaard. Project analyse en standaardisering bedrijfsnatuurplannen van Landschapsbeheer Nederland Intern rapport, Landschapsbeheer Nederland, Utrecht, 58pp . Dijk, A.J. van, 1993.Handleidin g SOVON Broedvogel Onderzoek. SOVON, Beek-Ubbergen NL, 54 pp. Dorp, D.van , 1996.See d dispersal in agricultural habitas and therestoratio n of species-rich meadows.Ph.D. Thesis ,Wageninge nAgricultura l University, 177 pp. Duelli, P.,Obrist , M.K., Schmatz,D.R. , 1999.Biodiversit y evaluation in agricultural landscapes. Agriculture Ecosystems and Environment 74,33-64 . Edwards,P.J. , Kollmann,J. , Wood,D. , 1999.Determinant s of agrobiodiversity inth e agricultural landscape. In:D.Woo d & J.M. Lenne(eds) , Agrobiodiversity: characterization, utilization andmanagement . CABIpublishing , Oxon/New York, 183- 210. Edwards,P .J. , Wratten, S.D., 1980.Ecolog y of insect-plant interactions. Studies inBiolog y no.121 , Edward Arnold, Londen, 60pp . EEC (Europese Commissie), 1994.D ebiologisch e landbouw. Bureau voor officiele publikaties der Europese Gemeenschappen, Luxemburg. Elsen,T . van, 2000. Species diversity asa tas k for organic agriculture in europe.Agriculture , Ecosystems &Environmen t 77, 101-110.

119 Emden,H.F .van , 1988.Th epotentia l for managing indigenous natural enemies of aphidso n fieldcrops .Philosphica l Transactions ofth eRoya l Society London, B318, 183-201. Esselink, H., Smeding, F.W., Schaminee, J.H.J, 1996.Biologisch e landbouw ind e Achterhoek: nostalgie ofnieuw ewege n voor natuurherstel.-Beschrijvin g van het bedrijf van de familie van Ipenburg. In:A . Stortelder (ed),Natuu r en landschap in de Achterhoek en Liemers,Jaarboe k 1995/96,Uitgav e Stichting Staring Instituut/Mr. H.J. Steenbergenstichting, Doetinchem NL, 8-39. Everts,F.H . &Vries ,N.P.J , de, 1991. Devegetatieontwikkelin g vanbeekdalsystemen ; een landschapsecologische studie van enkeleDrents e beekdalen. Historische Uitgeverij, Groningen, 224pp . Feber, E.E., Smith,H. ,Macdonald , D.W., 1999.Th e importance of spatially variablefield margin management for twobutterfl y species.Aspect s ofApplie d Biology 54, 155-162. Fretwell, S.D., 1977.Th eregulatio n ofplan t communities by the food chains exploiting them. Perspectives inBiolog y and Medicine 20, 169-185. Frouz, J., 1999.Us e of soildwellin g diptera (Insecta, Diptera) asbioindicators : arevie w of ecological requirements andrespons e todisturbance . Agriculture Ecosystems and Environment 74, 167-186. Fuller, R.J., Gregory, R.D., Gibbons,D.W. , Marchant, J.H., Wilson, J.D.,Bailie , S.R., Carter, N., 1995.Populatio n declines and range contractions among lowland farmland birdsi n Britain. 9,1425-1441 . Gange,A.C. ,Brown , V.K., Farmer, L.M., 1991. Mechanisms of seedling mortality by subterranean insect herbivores. Oecologia 88,228-232 . Gezondheidsraad, 1997.D e voedselweb-benadering ind e ecotoxicologische risicobeoordeling, 1994/14,Rijswijk , 82pp . Green, R.E., 1980.Foo d selection by skylarks and grazing damage to sugarbee t seedlings. Journal ofAppie d Ecology 17,613-630 . Hald,A.B. ,Redderse n J., 1990.Fuglefod e ikornmarker-insekte r og vildeplante r [Bird food items incerea lfield-insects an d weeds. Studies on conventional andorgani c farms 1987- 1988].Miljeprojek t 125.Danis h Environmental Protection Agency, Copenhagen, 107pp . Halley, J.M., Thomas,C.F.G. ,Jepson , P.C., 1996.A mode l for spatial dynamics of linyphiid spiders in farmland. Journal of Applied Ecology 63,471-492 . Haughton, A.J., Bell, J.R., Gates, S., Johnson, P.J., Macdonald, D.W., Tattersall, F.H., Hart, B.H., 1999.Method s ofincreasin g invertebrate abundance within field margins.Aspect so f Applied Biology 54, 163-170. Haysom,K.A. , McCracken, Foster, G.N., Sotherton, N.W., 1999.Gras s conservation headlands - adapting an arabletechniqu e for the grassland farmer. Aspects ofApplie d Biology 54, 171-178.

Heimbach,U. , Garbe,V. , 1996.Effect s ofreduce d tillage systems in sugarbee t on predatory andpes t arthropods. ActaJutlandic a LXXI: 2,natura l science series 10,Aarhu s University Press,Denmark , 195-208.

Hemptinne,J.L. , Dixon, A.F.G., 1997.Ar e aphidophagous ladybirds (Coccinellidae) prudent predators? Biological Agriculture &Horticultur e 15, 151-159.

120 Hermy,M. , 1994.Effect s of former landus eo nplan t speciesdiversit y andpatter n in European deciduous woodlands. In:Biodiversity , Temperate ecosystems and Global Change, Springer-Verlag, Berlin, 123-144. Hendriks,K. , Stobbelaar, D.J., Mansvelt, J.D.van , 2000.Th e appearance of agriculture -An assessment ofth e quality of landscape ofbot h organic and conventional horticultural farms inWes t Friesland. Agriculture, Ecosystems &Environmen t 77,157-176. Holland, J., Fahrig, L., 2000.Effec t ofwood yborder s on insect density and diversity in crop fields: alandscape-scal e analysis. Agriculture, Ecosystems and Environment 78,115-122. Holland, J.M., Drysdale, A., Hewitt, M.V., Turley, D., 1996. The LINK IFSProjec t - The effect ofcro p rotations and cropping systems on Carabidae. Aspects of Applied Biology 47,119-126. Holland, J.M., Thomas S.R., 1996.Phacelia tanacetifolia flower strips:thei r effect on beneficial invertebrates and gamebird chick food in anintegrate d farming system. Acta Jutlandica LXXI:2 ,natura l science series 10,Aarhu sUniversit y Press,Denmark , 171-183. Holland, J.M., Thomas, S.R., 1997.Assessin g therol e ofbeneficia l invertebrates in conventional and integrated farming systemsdurin g anoutbrea k ofSitobion avenae. Biological Agriculture &Horticultur e 15, 73-82. Holland, J.M., Winder, L., Perry, J.N., 1999.Arthropo dpre y of farmland birds:thei r spatial distribution within aspraye dfield wit h andwithou t buffer zones.Aspect s of Applied Biology 54,53-60 . Hospers, M., 1991.Bloeme n tegen luizen -deinzaa i van bloemstroken langs akkerranden. Publikatie Louis Bolk Instituut, DriebergenNL . Hradetzky,R. , Kromp,B. , 1997.Spatia l distribution of flying insects in anorgani c ryefield and an adjacent hedge and forest edge.Biologica l Agriculture &Horticultur e 15, 353-360. Hunt, H.W., Coleman, D.C.,Ingham , E.R, Ingham, R.E., Elliot, E.T., Moore, J.C.,Rose , S.L., Reid, C.P.P., Morley, C.R., 1987.Th edetrita l food web in a shortgrass prairie. Biology and Fertility of 3,57-68 . Idinger, J., Kromp,B. , 1997.Groun d photoeclector evaluation ofdifferen t arthropod groups inunfertilized , inorganic and compost-fertilized cerealfields i n Eastern Austria, Biological Agriculture &Horticultur e 15, 171-176. Idinger, J., Kromp,B. , Steinberger, K.H., 1996.Groun d photo-eclector evaluation ofth e numbers of carabid beetles and spiders found in and around cereal fields treated with either inorganic orcompos t fertilizers. ActaJutlandic a LXXI:2 ,natura l science series 10,Aarhu s University Press,Denmark , 255-269. Isart,J. , Llerena, J.J. (eds), 1996.Biodiversit y and Land Use:Th erol e of Organic Farming. Proceedings ofth efirst ENO Fworkshop , Barcelona, 155pp . Joenje, W., Burel,F. ,Gerowitt , B.,Paoletti , M.G., Thomas,C.F.G. ,Moonen , C, Baudry, J., LeCoeur , D.,Kleijn , D.,Schippers ,P. ,Kopp ,A. ,Thenail ,C , Marshall, E.J.P., 1997.Fiel d boundary habitat for wildlife, crop and environmental protection, Long Ashton Research Station, Bristol, 313pp . Kabourakis, E., 1996.Prototypin g and dissemination ofecologica l oliveproductio n systems: amethodolog y for designing and dissemination ofprototyp e ecological olive production systems (EOPS) in Crete.Ph.D .Thesis , Landbouwuniversiteit Wageningen, 121pp . Kevan, P.G., 1999.Pollinator s asbioindicator s ofth estat e ofth e environment: species, activity and diversity. Agriculture Ecosystems and Environment 74,373-394 . 121 Keulen, H.van , Lantinga, E.A., Laar, H.H. van, 1998.Mixe d farming systems in Europe. APMinderhoudhoeve-reeks nr.2,Landbouwuniversitei t Wageningen, 231pp . Kleijn, D., 1997. Species richness andwee d abundance in thevegetatio n of arable field boundaries.Ph.D .Thesis ,Landbouwuniversitei t Wageningen, 176pp . Kreuss,A. , Tscharntke, T., 1994.Habita t fragmentation, species loss,an d biological control. Science264 , 1581-1584. Kromp,B .& Meindl , P.(eds), 1997.Entomologica l Research in Organic Agriculture. Biological Agriculture &Horticultur e 15,No s 1-4. 382pp . Kromp,B. , 1999.Carabi d beetles in sustainable agriculture: arevie w onpes t control efficay, cultivation impacts and enhancement. Agriculture Ecosystems and Environment 74, 187- 228. Lange,R. , Twisk, P.,Winden , A.van , Diepenbeek, A.van , 1994.Zoogdiere n van West- Europa.KNNV-veldgid sno.8 , KNNV-uitgeverij, Utrecht, 400pp . Lantinga, E.A., Oomen,G.J.M. , 1998.Th eMinderhoudhoev e project: development ofa n integrated and ecological mixed farming system. In:H .va n Keulen, E.A. Lantinga &H.H . van Laar, Mixed farming systems in Europe. APMinderhoudhoeve-reeks nr.2, Landbouwuniversiteit Wageningen, 115-118. LaSalle,J. , 1999.Insec tbiodiversit y in agroecosystems: function, value and optimization. In: D.Woo d & J.M.Lenn e (eds),Agrobiodiversity : characterization, utilization and management. CABIpublishing , Oxon/NewYork , 155-182. Leeuwen, C.G. van, 1966.A relatio n theoretical approach topatter n andproces s in vegetation. Wentia 15, 25-46. Lys,J.A. , 1994.th epositiv e influence of stripmanagemen t on ground beetles in a cereal field: increase,migratio n and overwintering. In:Desende r etal. (eds) ,Carabi d beetles: Ecology and Evolution, Kluwer Academic Publishers, TheNetherlands ,451-455 . Lys,J .A. ,Nentwig , W., 1992.Augmentatio n ofbeneficia l arthropods by strip management 4: surface activity, movements and activity density of abundant carabid beetles in acerea l field.Oecologi a 92,373-382 . Macleod, A., Sotherton,N.W. ,R.W.J . Harwood, Wratten, S.D., 1995.A n improved suction sampling devicet o collect aphids andthei rpredator s in agroecosystems. Acta Jutlandica LXX: 2,natura l science series 10,Aarhu sUniversit y Press,Denmark , 125-131. Magurran, A.E., 1988.Ecologica l diversity and itsmeasurement . CroomHelm , London, 178 pp. Marc, P.,Canard , A.,Ysnel , F., 1999. Spiders (Aranae) useful for pest limitation and bioindication. Agriculture Ecosystems andEnvironmen t 74,229-274 . Moreby, S.J., Aebischer, N.J., Southway, S.E., Sotherton,N.W. , 1994.A compariso n ofth e flora and arthropod fauna of organically and conventionally grown winter wheat in southern England. Annal of Applied Biology, 125,13-27 . Moreby, S.J., Sotherton, N.W., 1997.A compariso n of someimportan t chick-food insect groups found inorgani c and conventionally grown winter wheat fields in Southern Engeland. Biological Agriculture &Horticultur e 15, 51-60. Nan,H. , Smeding, F.W., Stuiver, J., Bruggen, A.H.C., (inprep. ) Spatial distribution of skylarkAlauda arvensis territorie s onorgani c farmland in The Netherlands - spatial analysis using GIS.

122 Nijs, L.J.M.F. den, Boesewinkel, E., Zijp, J.P.H., Smeding,F.W. , Booij, C.J.H., 1998. Methods to study the effect of ecological infrastructure on aphids and their enemies on winter wheat. Proceedings Experimental and Applied Entomology 9,93-98 . Nijs, L.J.M.F. den,Booij , C.J.H.,Daamen ,R. , Lock,C.A.M. ,Noorlander , J., 1996.Ca n pitfall trap catches inform us about survival andmortalit y factors? An analysis of pitfall trap data for Pterostichus cupreus(Coleoptera : Carabidae) inrelatio n to crop rotation and crop specific husbandry practices.Act a Jutlandica LXXI: 2,natura l science series 10, AarhusUniversit y Press,Denmark ,41-56 . Nijs, L.den ,Booij , K., Daamen, R., Lock, C, Noorlander, J., 1997.Th e effects ofcro p rotation onbeneficia l arthropods in agroecosystems. Biological Agriculture &Horticultur e 15,364. O'Connor , R.J., Shrubb, M., 1986.Farmin g andBirds .Cambridg e University Press. Odum,E.P. , 1974.Fundamental s of ecology, 3rd Ed., W.B. Saunders, Philadelphia, 574pp . Oksanen, L., Oksanen, T., Ekerholm, P.,Moen , J., Lundberg, P.,Schneider , M., Aunapuu, M., 1996.Structur e and dynamics of arctic-subarctic grazer webs in relation toprimar y productivity. In:G.A . Polis & K.O.Winemiller , Foodwebs : integration ofpattern s and dynamics. Chapman &Hall ,Ne w York, 231-242. Oldeman, R.A.A., 1990. :Element s of Silvology. Springer-verlag, 624pp . Oomen,G.J.M. ,Habets ,A.S.J. , 1998.Usin g thewhol e farm model FARM and the dynamic modelNDICE A to integrate arable and animal production. In:H . van Keulen, E.A. Lantinga &H.H .va n Laar, Mixed farming systems inEurope . APMinderhoudhoeve-reeks nr.2, Landbouwuniversiteit Wageningen, 199-206. Opdam,P. ,va n Apeldoorn, R., Schotman,A. ,Kalkhoven , J., 1993. Population responses to landscape fragmentation. In: C.C.Vo s & P. Opdam (eds),Landscap e Ecology ofa Stressed Environment. Chapman and Hall, London, 147-171. Paoletti, M.G.,Hassal ,M. , 1999.Woodlic e (Isopoda: Oniscidea):thei r potential for assessing sustainability andus e asbioindicators . Agriculture Ecosystems and Environment 74, 157- 166. Paoletti,M.G.(ed) , 1999.Invertebrat e biodiversity asbioindicator s of sustainable landscapes: practical use of invertebrates to assess sustainable land use.Agricultur e Ecosystems and Environment 74,no s 1-3, 446pp . Paoletti, M.G.,Boscolo , P., Sommaggio, D., 1997.Beneficia l insects in fields surrounded by hedgerows inNort h Eastern Italy.Biologica l Agriculture &Horticultur e 15,311-324 . Paoletti, M.G., Sommagio, D.,Bressan , M,. Celano,V. , 1996.Ca n sustainable agricultural practices affect biodiversity in agricultural landscapes? Acas e study concerning orchards in Italy. Acta Jutlandica LXXI: 2,natura l science series 10,Aarhu s University Press, Denmark, 241-254. Park, J., Cousins, S.H., 1995.Soi lbiologica lhealt h and agro-ecological change. Agriculture, Ecosystems and Environment 56,137-148 . Pfiffner, L., Mader,P. , 1997.Effect s ofbiodynamic , organic and conventional production systems on earthworm populations.Biologica l Agriculture &Horticultur e 15,3-10 . Phelan, P.L., 1997. Soil-management history and the role of plant mineral balance asa determinant ofmaiz e susceptibility toth eeuropea n cornborer . Biological Agriculture& Horticulture 15,25-34.

123 Pimentel, D.,Harvey , C, Resosudarmo, P., Sinclair, K., Kurz, K.,McNair , M., Christ, S., Shpritz,L. ,Fitton , L., Saffouri, R., Blair, R., 1995.Environmenta l and economic costso f soil erosion and conservation benefits, Science 267,1117-1123. Pimm, S.L., 1991. Thebalanc e ofnature ? -Ecological issues in the conservation of species and communities. TheUniversit y ofChicag o Press,Chicago/London , 434pp . Plesnik, J., Dusik, M, 1994.Reproductiv e output ofth eKestre l Falcotinnuculus i nrelatio n to smallmamma l dynamics inintensivel y farmland. In:Rapto r conservation today. The PicaPress ,Meyburg . Polis, G.A., Holt, R.D.,Menge ,B.A. , Winemiller,K.O. , 1996.Time , space and life history: influences on food webs.In :G.A .Poli s& K.O.Winemiller , Foodwebs : integration of patterns and dynamics. Chapman &Hall ,Ne w York,435-460 . Polis, G.A., Hurd, S.D., 1996.Allochthonou s input acrosshabitats , subsidized consumers and apparent trophic cascades: examples from theocean-lan d interface. In: G.A.Poli s& K.O. Winemiller, Food webs: integration ofpattern s anddynamics . Chapman &Hall ,Ne w York,275-285 . Polis,G.A. , Winemiller, K.O., 1996.Foo d webs:integratio n ofpattern s and dynamics. Chapman &Hall ,Ne w York, 472pp . Potts, G.R., 1986.Th epartridge :pesticides ,predatio n and conservation. Collins, London. Potts, G.R., Vickermann, G.P., 1974.Studie s ofth ecerea l ecosystem. Advances in Ecological Research, 8, 107-197. Poulsen,J.G. ,Aebischer , N.J., 1995.Quantitativ e comparison oftw omethod s of assessing diet ofnestlin g skylarks (Alaudaarvensis) . Th e auk 112, 1070-1073. Poulsen, J.G., Sotherton,N.W. , Aebischer, N.J., 1998.Comparativ e nesting and feeding ecology ofskylark sAlauda arvensis o n arable farmland in Southern England with special reference to set-aside.Journa l ofApplie d Ecology 35, 131-147. Power, W.,Nickless ,A. , 1997.Th e foraging efficiency ofth eparasitoi dDiaeretiella rapae (Hymenoptera: Braconidae) in relation to the spatial distribution of aphids onplants .Act a Jutlandica LXXI:2 ,natura l science series 10,Aarhu s University Press,Denmark , 97-106. Power, M.E.,Parker , M.S.,Wootton , J.T., 1996.Disturbanc e and food chain length inrivers . In: G.A. Polis & K.O.Winemiller , Food webs: integration ofpattern s anddynamics . Chapman &Hall ,Ne wYork ,286-297 .

Rabbinge,R . 1997.Chairman' s introduction. In:Precisio n Agriculture: Spatial and Temporal Variability ofEnvironmenta l Quality. CibaFound . Symp. 210. Wiley, New York, 1-5. Rands,M.R.W. , 1985.pesticid e useo n cereals and the survival of greypartridg e chicks:a fieldexperiment . Journal ofApplie d Ecology,49-54 . Reddersen, J., 1997.Th e arthropod fauna oforgani c versus conventional cerealfields i n Denmark. Biological Agriculture &Horticultur e 15,61-71 .

Remmelzwaal,A.J. , 1992.Rijkswaterstaa t alsbiologisch e boer. Landinrichting 32(5),22-27 . Remmelzwaal,A.J. , Voslamber, B., 1996.I n demarg e-ee n onderzoek naar ruimte voor de natuuro plandbouwbedrijven . Flevobericht 390.Ministeri e vanVerkee r en Waterstaat, RWS-RDIJ ,Lelysta d NL, 136pp . Ruiter, P.C.de ,Moore , J.C.,Zwart , K.B.,Bouwman , L.A., Hassink, J., Bloem,J. , Vos,J.A . de,Marinissen , J.C.Y., Didden, W.A.M., Lebbink, G., Brussaard, L., 1993.Simulatio n of

124 nitrogen mineralization in thebelow-groun d food webs of twowinte r wheat fields. Journal ofApplie d Ecology 30,95-106 . RuiterP.C . de,Neutel , A.M., Moore, J.C., 1995.Energetics ,patter n of interaction strengths and stability inrea l ecosystems. Science 269,1257-1260. Ruiter, P.C. de,Neutel , A.M.,Moore , J.C., 1996.Energetic s and stability in below-ground food webs.In : G.A. Polis & K.O.Winemiller ,Foo d webs:integratio n ofpattern s and dynamics. Chapman &Hall ,Ne w York, 201-210. Ryszkowski, L,Karg ,J. , 1991.Th e effect ofth e structure of agricultural landscape on biomass of insectso fth e above-ground fauna. Ekologia Polska 39(2), 171-179. Ryszkowski, L.,Karg , J., Margarit, G., Paoletti, M.G., Zlotin, R., 1993. Above-ground insect biomass in agricultural landscapes ofEurope .In :R.G.H . Bunce, L. Ryszkowski & M.G. Paoletti, Landscape Ecology and Agroecosystems, Lewispublishers , BocaRaton , 71-82. Scharenburg, Kvan ,Hoff , J. van 't, Koks,B. ,Klinken , A.van , 1990.Akkervogel s in Groningen. SOVON-district Groningen en Avifauna Groningen, Groningen NL, 112 pp. Schekkerman,H. , 1997.Graslandbehee r en groeimogelijkheden voor weidevogelkuikens. DLG-publicatie 102,IBN-rappor t 292,Instituu t voor Bos-e n Natuuronderzoek, Wageningen, 92pp . Schoenly, K.G., Cohen, J.E., Heong, K.L.,Arida , G.S.,Barrion , A.T., Litsinger, J.A., 1996. Quantifying the impact of insecticides on food web structure ofric e arthropod populations ina Philippin e Farmer's irrigated field: acas e study.In : G.A. Polis & K.O. Winemiller, Food webs:integratio n ofpattern s anddynamics . Chapman &Hall ,Ne w York, 343-351. Sheldrake, R., 1988.Th epresenc e ofth epast . Fontana, London, 391pp . Sheldrake, R., 1991. Therebirt h ofnature .Banta m Books,Ne w York, 217pp . Sinnige,C.A.M. ,Tamis , W.L.M., Klijn, F., 1992.Indelin g van debodemfaun a in ecologische soortengroepen. Report 80,Centru m voor Milieukunde, Rijksuniversiteit Leiden, 127pp . Smeding, F.W., 1993.Onkruidvegetatie s ind ebiologisch e landbouw. Stratiotes 7,29-37 . Smeding,F.W. , 1995.Protoco lNatuurplan . Mededelingen van deVakgroe p Ecologische Landbouw 002.95,Landbouwuniversitei t Wageningen, 137pp . Smeding,F.W. , Booij, C.J.H., 1999.Effec t of field margin management on insectivorous birds, aphids and their predators in different landscapes.Aspect s ofApplie d Biology 54, 367-373. Smeding,F.W. , Booij, C.J.H., 2001. Insectivorous breedingbird s on organic farms in relation to crop area, ecological infrastructure and arthropods. Chapter 4,Ph.D. Thesis , Wageningen University, 51-74. Smeding, F.W., Lock, C.A.M.,Booij , C.J.H.,2001a . Shifts in functional group composition of ground dwelling arthropods inrelatio n to ecological infrastructure and conversion to organic farming. Chapter 3, Ph.D.Thesis ,Wageninge n University, 31-50. Smeding,F.W. , Maassen, A.H.T.M., Bruggen, A.H.C.van , 2001b.Functiona l group compositions of vegetation dwellling arthropods inrelatio n to ecological infrastructure and time since conversion to organic farming. Chapter 2,Ph.D .Thesis , Wageningen University, 5-29. Smeding, F.W., Joenje, W., 1999.Farm-Natur e Plan: landscape ecology based farm planning. Landscape andUrba n Planning 46, 109-115.

125 Smeding,F.W. , Snoo, G.R. de,2001 . Aconcep t of food web structure in organic arable farming systems. Chapter 6,Ph.D .Thesis ,Wageninge n University, 83-99. Snoo, G.R. de, 1995.Unspraye dfield margins :implication s for environment, biodiversity and agricultural practice -Th e dutchfield margi n project inth eHaarlemmermeerpolder .Ph.D . Thesis,Rijksuniversitei t Leiden, 205pp . Snoo, G.R de,Chaney , K., 1999.Unspraye dfield margin s- what arew e trying to achieve? Aspects ofApplie d Biology 54,1-13 . Snoo, G.R. de,Canters ,K.J. , 1990. Side effects ofpesticide s on terrestrial vertebrates. CML report 35,Centru m Milieukunde, Rijksuniversiteit Leiden, 147pp . Steffan-Dewenter, I., Tscharntke, T., 1997.Earl y succession ofbutterfl y and plant communities in set-asidefields. Oecologi a 109,294-302 . Sotherton, N.W., 1984.Th e distribution and abundance ofpredator y arthropods overwintering on farmland. Annals of applied biology 105, 423-429. Sotherton, N.W., 1985.Th edistributio n and abundance ofpredator y Coleoptera overwintering infield boundaries . Annals of applied Biology 106, 17-21. Stoat, C, 1999.Th e influence offield boundar y structure onbreedin g territory establishment ofwhitethroa t Sylviacommunis andyellowhamme rEmberiza citrinella. Aspect s of Applied Biology 54, 125-130. Stobbelaar, D.J., Mansvelt, J.D. van (eds),2000 .Criteri a for sustainable landscape development. Agriculture, Ecosystems &Environmen t 77,no s 1-2, 178pp . Strong,D.R. , Maron, J.L., Conners,P.G. , 1996.To p downfrom underground ? The underappreciated influence of subterranean food webs on aboveground ecology. In: G.A. Polis & K.O.Winemiller , Food webs:integratio n ofpattern s and dynamics. Chapman & Hall,Ne w York, 170-178. Sunderland, K.D., 1991. The ecology of spidersi ncereal s inEurope .Proc . 6th Int. Symp. Pests andDisease s of Small Grain Cereals and Maize, Halle/Saale, Germany, 1,269-280 . Sunderland, K.D., Bilde, T.,Nijs , LJ.M.F. den, Dinter, A.,Heimbach , U., Lys,J.A. , Powell, W., Toft, S., 1996a.Reproductio n ofbeneficia l predators andparasitoid s in agroecosystems inrelatio n tohabita t and food availability. Acta Jutlandica LXXI:2 , series 10,Aarhu s University Press,Denmark , 117-156. Sunderland, K.D.,Topping , C.J., Ellis, S., Long, S., Laak, S. van de,Else , M., 1996b. Reproduction and survival oflinyphii d spiders,wit h special reference to Leptyphantes tenuis(Blackwell) . ActaJutlandic a LXXI:2 ,natura l science series 10,Aarhu s University Press, Denmark, 81-96. Svensson, B.,Lagerlof , J., Svensson, B.G., 2000.Habita t preferences of nest-seeking bumble bees (Hymenoptera: Apidae) in an agricultural landscape. Agriculture Ecosystems and Environment 77,247-25 5 Swift, M.J., Anderson, J.M., 1993.Biodiversit y and ecosystem functioning in agricultural systems. In: E.D. Schulze &H.A . Mooney (eds),Biodiversit y and Ecosystem Function, Springer Verlag, Berlin, 15-41. Theunissen, J., 1994.Intercroppin g infield vegetabl e crops:pes t management by agroecosystem diversification: anoverview . Pesticide Science, 42,65-68 . Theunissen, J., 1997.Applicatio n ofintercroppin g inorgani c agriculture. Biological Agriculture &Horticultur e 15, 251-260.

126 Theunissen, J., Kohl, J., 1999.Knelpuntanalys e van ziekten enplage n in vollegrondsgroente- en akkerbouwgewassen in de ecologische teelt.Not a 14,Plan t Research International, Wageningen, 24pp . Thomas, C.F.G.,Green , F.,Marshall , E.P.J., 1997.Distribution , dispersal, andpopulatio n size ofth e ground beetles,Pterostichus melanarius (Illiger) andHarpalus rufipes (Degeer ) (Coleoptera, Carabidae) infield margi n habitats.Biologica l Agriculture &Horticultur e 15, 337-352. Toft, S., 1996.Indicator s ofpre y quality for arthropod predators. Acta Jutlandica LXXI:2 , natural science series 10,Aarhu s University Press,Denmark , 107-117. Topping, C.J., 1997.Predictin g the effect of landscapeheterogeneit y on the distribution of spiders in agroecosystems using apopulatio n dynamics driven landscape-scale simulation model.Biologica l Agriculture &Horticultur e 15, 325-336. Topping, C.J., Sunderland, K.D., 1996.Estimatin g themortalit y rate of eggs and first free- living instarLepthyphantes tenuis(Aranae :Linyphiidae ) from measurements of reproduction and development. ActaJutlandic a LXXI:2 ,natura l science series 10,Aarhu s University Press,Denmark , 57-68. Tscharntke,T. , 1997.Vertebrat e effects onplant-invertebrat e food webs.In :A.C .Gang e& V.K. Brown (eds),Multitrophi c Interactions inTerrestria l Systems,Blackwel l Sciencific Publications, Oxford, 277-297. Tscharntke, T., Gathmann, A., Steffan-Dewenter, I., 1998.Bioindicatio n using trap-nesting bees andwasp s and their natural enemies: community structure and interactions.Journa l of Applied Ecology 35, 708-719. Tscharntke, T., Greiler, H.J., 1995.Insec t communities, grasses and grasslands. Annual Reviews of Entomology 40,535-558 . Unwin, D.M., 1981. Ake y toth e families ofBritis h Diptera. Field studies 5,513-553 . Vandermeer,J. , 1995.Th e ecological basis of alternative agriculture. Annual Reviews Ecol. Syst. 26,201-224. Vandermeer, J., Noordwijk, M. van, Anderson, J., Ong, C, Perfecto, Y., 1998.Globa l change and multi-species agroecosystems: concepts and issues.Agriculture , Ecosystems and Environment 67, 1-22. Venhorst, B.,Smeding , F.W., Booij, C.J.H.,(i nprep. ) Distribution ofhoverflie s (Syrphidae) in awhea tfield i n relation to distance to the field margin. Vereijken, P., 1989.Researc h on integrated arable farming and organic mixed fanning inTh e Netherlands. In:P .Vereijke n &D.J . Roye,Curren t status ofIntegrate d Arable Farming Systems Research inWester n Europe, IOBC/WPRSBulletin ,41-50 . Vereijken, P., 1997.A methodica l wayo fprototypin g integrated and ecological arable farming systems (I/EAFS) in interaction with pilot farms. European Journal of Agronomy 7,235-250 . Vereijken, P. (ed), 1998.Improvin g anddisseminatin g prototypes.Progres s report4 , Research network for EUan dAssociate d Countries onIntegrate d and Ecological Arable Farming Systems, AB-DLO,Wageningen , 55pp . Vereijken, P.,Kloen , H., enVisser , R., 1994.Innovatieprojec t Ecologische Akkerbouw en Groenteteelt, eerste voortgangsrapport. Rapport 28,AB-DLO ,Wageningen , 95pp . Vickermann, G.P., Sunderland, K.D., 1975.Arthropod s in cereal crops:nocturna l activity, vertical distribution and aphid predation. Journal ofApplie d Ecology 12,755-766 . 127 Visser, A.J.,2000 .Prototypin g on farm nature management -a synthesis of landscape ecology, development policies and farm specific possibilities. Aspects ofApplie d Biology 58, 299- 304. Wakeham-Dawson, A., Szoszkiewicz, K., Stern,K. , Aebischer, N.J., 1998.Breedin g skylarks Alaudaarvensis o n Environmentally Sensitive Area arablereversio n grass in Southern England: survey-based and experimental determination of density. Journal of Applied Ecoloy 35,635-648 . Wardle,D.A , Giller, K.E. &Barker , G.M., 1999.Th eregulatio n and functional significance of soilbiodiversit y in agroecosystems. In:D.Woo d & J.M. Lenne (eds), Agrobiodiversity: characterization, utilization andmanagement . CABIpublishing , Oxon/New York, 87-122. Weber, G., Franzen, J., Buchs W., 1997.Beneficia l Diptera infield crop s with different inputs ofpesticide s and fertilizers. Biological Agriculture &Horticultur e 15,109-122. Werff, P.A.va n der, 1991. Alleen een andere wetenschap kan zorgen voor een duurzame landbouw. Landbouwkundig tijdschrift 103(9) ,25-27 . Westerman, P.R., Werf, W.va n der,Hofman , A.,Vet , L.E. (inprep. ) Therelativ e importance ofvertebrate s and invertebrates in seedpredatio n on cerealfields. Westhoff, V., Bakker, P.A., Leeuwen, C.G. van, Voo,E.E .va n der, 1970.Wild e Planten I. UitgaveNatuurmonumenten , Amsterdam. White,T.C.R. , 1978.Th e importance of arelativ e shortage of food in animal ecology. Oecologia 33, 71-86. Wilson, J.D.,Browne , S.J., 1993. Habitat selection andbreedin g success of skylarksAlauda arvensiso n organic and conventional farmland. BTOResearc h report 129,4 8pp . Wilson, J.D.,Evans ,J. , Browne, S.J., King, J.R., 1997.Territor y distribution and breeding success of skylarksAlauda arvensis o n organic and intensive farmland in southern England.Journa l ofApplie d Ecology 34, 1462-1478. Winemiller,K.O. ,Polis , G.A., 1996.Foo d webs:wha t can they tell usabou t theworld ?In : G.A. Polis & K.O.Winemiller , Food webs: integration ofpattern s and dynamics. Chapman &Hall ,Ne w York, 1-22. Wingerden, W.K.R.E., Booij, C.J.H., 1999.Biodiversitei t enonderdrukkin g van ziektene n plagen: strategieen engraadmeters . IBN-rapport 413,Instituu t voor Bos-e n Natuuronderzoek/Instituut voor Plantenziektenkundig Onderzoek, Wageningen. Wise,D.H. , Snyder, W.E., Tuntibunpakul, P.,Halaj , J., 1999.Spider s in food webs of agroecosystems:theor y and evidence. The Journal of Arachnology 27,363-370 . Wolfert, J., 1998.Decisio n support for mixed ecological farming systems based onphysica l product flows. In:H .va n Keulen, E.A. Lantinga &H.H . van Laar, Mixed farming systems inEurope .APMinderhoudhoeve-reek s nr.2,Landbouwuniversitei t Wageningen, 207-212. Wolfert, J., Goewie, E.A.& Beulens, A.J.M., 1997.Dynami c product flow model for amixe d ecological farm. In:H .Kure , I. Thysen &A.R . Kristensen (eds),Proceeding s ofth e first European Conference for Information Technology inAgriculture , 15-18Jun e 1997.Th e Royal Veterinary and Agricultural University, Copenhagen, Denmark, pp. 199-204 Wolfert, J., (inprep. ) (Dynamic product flow models for decision support in farming systems).Ph.D .Thesis ,Wageninge n University. Wood, D.,Lenne ,J.M. , 1999.Agrobiodiversity : characterization, utilization and management. CABIpublishing , Oxon/New York, 490pp .

128 Wratten, S.D., Thomas, C.F.G., 1990.Far m scalespatia l dynamics ofpredator s and parasitoids in agricultural landscapes. In:R.G.H . Bunce & D.C. Howard (eds), Species dispersal in agricultural habitats. Belhaven Press, London. Zimmerman, J. Btichs,W. , 1996.Managemen t of arablecrop s and its effect onrov e beetles (Coleoptera: Staphylinidae), with special reference toth e effects of insecticide treatments. Acta Jutlandica LXXI:2 ,natura l science series 10,Aarhu s University Press, Denmark, 183-194.

129 Summary

Species diversity of indigenous plants and animals on farmland is declining dramatically due to agricultural intensification. Biodiversity loss in agriculture affects both economic and nature conservation values. One important solution for thereversa l ofbiodiversit y loss inth e agriculture of industrialised countries couldb eth e development of farming systems that are economically based on utilisation ofbiodiversit y and that also harbour conservation worthy species.Thi s idea iscompatibl e with the concept ofmulti-specie s agroecosystems. Development of organic farming systems is apossibl e implementation ofthi sconcept ; crop areas managed according to this conceptma yb e complemented by anadjacen t ecological infrastructure (E.I.)- Afoo d web approach is an appropriate method for investigations into the higher integration levels of ecosystems andma y therefore contribute to research onmulti - species farming systems.Progres s in scientific understanding of food webs is currently expanding into agroecology. A food web approach alsoprovide s aframewor k for ordering the scattered information on farmland species andhabitats .

A simplified set ofhypothese s withregar d to food web structure on organic farms served as the starting point for thethesis :i twa s expected that theduratio n oforgani c management, extensive crop rotation aswel l as animprove d E.I. (i.e.enlarge d area and latemowed) ,woul d relatepositivel y to the abundance of arthropods inbot h non-pest herbivorous and detritivorous primary subsystems. Thevarie d plant tissues in improved E.I.woul d particularly boost thenon-pes t herbivorous functional groups. Consequently predator and parasitoid numbers onth e farm would be supported. An accumulated high invertebrate density onth e farm would support insectivorous vertebrates,particularl y birds.Th e objectives ofth ethesi s were:t o take acomprehensiv e food web approach inrelatio n to the farm and E.I. management; to link component knowledge on species and farm operations to food web theories; andt opoin t out implications for multi-species farm development. The study area involved farms that were mainly situated inth e relatively uniform landscape of Flevoland. Thispolde r has aconcentratio n of organic arable farms, including farms with improved E.I.

Empirical investigations were concentrated onparticula r sections ofth e farmland community, namely ground dwelling arthropods, vegetation dwelling arthropods and insectivorous birds. Functional group compositions (e.g. detritivores, herbivores and predators) within these sections were assessed. Field studies indicated that allth e factors included inth e above hypothesis affected abundance and functional group composition ofbot h arthropods and birds.However , the effects differed from what was expected.

Inth e crop area (i.e.whea t crops),K-herbivor e abundance was related to farms that hadbee n recently converted and r-herbivore abundance was related to farms of alonge r duration with anintensiv e crop rotation. Both the observed herbivore increases might be explained bya combination of crop susceptibility to theherbivor e group, and limited predation pressure (Chapter 2). Therewas , accordingt o thehypothesis , some evidence for detritivore increase related to extensive crop rotation on long duration farms and for detrital subweb-connected predator increase inrelatio n to more extensive farms (Chapters 2 and 3).Predato r abundance in thecro p canopy wasparticularl y related tohig h K-herbivore numbers on recently converted farms. Thehighes t species diversity was found on long duration farms whichha d anintensiv e croprotatio n (Chapter 2).Th e farms which had anextensiv e crop rotation, compared with the other farms, had alo wpredato r number inth e crop canopy, whereas the parasitoids had intermediate numbers. The density ofbirds ' territories seemed tob e enhanced by the crop areas of those long duration organic farms that had an intensive crop rotation, this being possibly related to an increased herbivory (Chapter4) .

131 An increased arthropod abundance was found inth evegetatio n ofth e improved E.I.Th e increased functional groupsinclude d bothpredator s anddetritivore s (Chapter 2). However, K-herbivore numbers didno t increase inth e improved E.I.whe n they were compared toth e traditionally managed E.I. (Chapter 2 and 3);perhap sbecaus e theprolonge d vegetative development of frequently mown vegetation might extend the availability ofnutritiv e grass tissue. Also,th e abundance ofth e epigeicpredators , inparticula r ofth e quantitatively important group ofpredaciou sbeetles ,wa s not related to an improved E.I. (Chapter 3). Observations suggested that theirnumber s wererelate d to the crop area conditions. Open, frequently mown E.I.vegetatio n enhanced themor e specific ground dwelling functional groups,whe n compared to improved E.I. Theterritoria l density ofinsectivorou s birdswa s shown tob epositivel y related to thehig h arthropod abundance inth eE.I . (Chapter4) .

Field studiesprovide d some evidence for thedispersa l offunctiona l groups, abundant inth e E.I.int o thecro p area. However, the effects of crop conditions on the arthropod abundance in the crop areawer e observed to offset the influence ofth e E.I. for some groups. Contrary to expectations, observations suggested the opposite effects regarding flying arthropods,whic h werepossibl y transported intoth eE.I. , and epigeicpredaciou s beetles (Chapters 2an d3) . Bird studies (Chapter 4)reveale d positive correlations between bird functional groups anda combination ofcro p area and E.I. characteristics. Thisma y suggest thatbot h thecro p areaan d theE.I .habitat s are important tothes ebirds .Thi sma y also hold true for birdstha t typically live inth eE.I. ,becaus e acertai nproportio n ofthei rpre y could originate in the crop area.

In the discussion section of each empirical studyobservation s were linked to theoryi n food web literature.Fiel d observations provided some indications for indirect effects of subsidised detrital food chains onherbivor e abundance and consequently onbir d abundance, aswel l as possible effects of intraguild predation on arthropod functional group composition. However extrapolation ofth eresult s obtained inFlevolan d to other geographical regions mayrequir ea further examination of functional group assemblages, and ofprimar y resources related to weeds ort o semi-natural habitats othertha n herbaceous habitat (Chapters 5an d 8).

Aproposa l for adescriptiv e ortopologica l farm food web isdraw n from field observations as well asfro m references in literature (Chapter 6). Important themes in the food web theory weretentativel y applied tothi spreliminar y model in explaining differences between localised farm food web structures andho wthe y arerelate d toth e farm and/orecologica l infrastructure management. Predictions aremad e for four different farm food web structures that express four extremes oftw o environmental gradients which correspond toth e length of organic duration andth e amount/qualityo f theE.I . Empirical observations areuse d to illustrate the different food web structures. However, further investigation isnecessar y to confirm the hypotheses put forward.

The implications ofbot h empirical aswel l asconceptua l studies are that the organic duration and the amount/qualityo fth e ecological infrastructure maybot hpositivel y contribute to improving ecosystem services andt o aimsbase d onnatur e conservation. However, an optimisation ofth e farm food webwit hregar d to ecosystem services mayru n counter to nature conservation values, asi nth ecas e of enhanced epigeic predators whichwoul d depress herbivorous food items for birds. Nevertheless anincrease d understanding ofth e farm food web and itsmanagemen t islikel y to show more clearly thepossibilitie s for the development ofmulti-specie s agroecosystemstha t integrate ecosystem services andnatur e conservation goals. Theoutcom e ofthi s kind of applied ecological research is likely to improve thebasi so f pragmatic protocols for farm-nature planning exemplified in Chapter 7.

132 Samenvatting

De soortsdiversiteit end e aantallen vanwild ediere n enplante n in landbouwgebieden gaan achteruit ten gevolge van intensivering van de landbouw. Dit verlies aanbiodiversitei t heeft nadelen voor zowel economieal snatuurbescherming . Aanwending vanecosystem servicesi n verband metbiodiversiteit , zoukunne nbijdrage n aan eenverminderd einze t van bestrijdings- middelen,meststoffe n enfossiel e brandstoffen. Dezeecosystem servicesworde n mogelijk door eenklein e groepva n essentiele soorten verzorgd. Echtervanui t hetoogpun t vannatuur ­ bescherming isbred e stimulering vanbiodiversitei t gewenst omdat landbouw veelruimt e gebruikt en specifieke leefgemeenschappen omvat.Ee nmogelijk e oplossing voorhe t geschetste probleem ishe t ontwikkelen van landbouwsystemen diebiodiversitei t benutten en instandhouden endu s landbouw ennatuurbeschermin g ophe t niveau van een landbouw- bedrijf verenigen.Verschillend e agroecologen hebben dit themauitgewerkt . De Amerikaanse onderzoeker Vandermeer noemt dergelijke landbouwsystemen 'multi-species agroecosystems' (Vandermeer etal., 1998);di tbegri pva t dedoelstellin g vandez ethesi s samen.

Onderzoek dienstig aanhe tontwikkele n vanmulti-species agroecosystems i nd e praktijk verlangt een specifieke opzet. Watbetref t het object ishe tbelangrij k dathe t onderzochte systeem degewenst e specifieke complexiteit bezit omdat degezocht eprocesse n mogelijk alleen in een dergelijke context optreden. Vanuit deze overweging is gekozen omo p biologische landbouwbedrijven onderzoek te doen enme tnam e opbiologisch ebedrijve n met randomd epercele n een ecologische infrastructuur (E.I.) die verbeterd is,d.w.z .vergroo t qua oppervlak en ecologisch beheerd. Watbetref t demethod e ishe tbelangrij k dathe t onderzoek kenmerkenmee t ophe t systeemniveau.Vanui t dezeoverwegin g is gekozen vooree n voedselwebbenadering, overeenkomstigd e aanbevelingen van Gezondheidsraad (1997)me t het oogo pecosysteemanalyse .Uitgangspunte n enuitwerkinge n van debenaderin g zijn te vinden inhe t gezaghebbende werk vanPoli s& Winemille r(1996) . In agroecosystemen (ind e gematigde streken) verkeert voedselwebonderzoek nog in eenoverwegen d exploratieve fase metbijvoorbeel d fundamentele studies aanhe t detritus web van DeRuite r etal. (1993,1995 ) enhypothese-ontwikkelend e artikelen van Wiseet al. (1999) ,Brussaar d (1998) enTscharntk e (1997).

Het onderzoeksdoel voor ditproefschrif t was drieledig: -he t opstellen van een descriptief ('topologisch') voedselweb voor eenbiologisc h akkerbouwbedrijf, datzowe lhe tonder - alsbovengronds e systeem omvat enda t geschikt is voor het leggen van relaties met handelingen van deboer ; -he t leggen van verbanden tussen componentkennis (diersoorten, teeltmaatregelen en biotoopbeheer), en ecologische theorie vanuit devoedselwebbenadering , om zodoende aanwijzingen tevinde n voor effecten opbedrijfsnivea u van voedselwebinteracties; - zoekrichtingen aangeven voor het ontwikkelen vanbedrijfssysteme n diehe tbenutte n van 'ecosystem services'e nnatuurbeschermin g combineren.

Dehypothes e van het onderzoek bestaat uit een serie,vanui t de literatuur onderbouwde, deel- hypothesen tenaanzie n van devoedselwebstructuur .Dez ehypothes e was dat ero plan g (geleden) omgeschakelde bedrijven met eenverbeterd e E.I. een relatief grotedichthei d isva n mesofauna enmacrofaun a van deprimair e functionele groepen. De detritivoren worden bevorderd door een extensieve vruchtwisseling met granen enkunstweide n dievee l organische stof achterlaten, enoo k door organische mest.D eherbivoren , enme t name niet- plaagsoorten, hebben relatiefhog e dichtheden door deaanwezighei d van veel verschillende gewassen enwa t talrijkere onkruiden. DeE.I .bevorder t vooral de aantallen herbivoren vanwegehaa r grotere variatie aanplantenweefsels , inclusiefbijvoorbeel d bloemen enzaden . 133 Bij detoenam eva n deprimair e functionele groepen enhu n voedselbronnen spelen cumulatieve effecten overd ejare n mee. Detoegenome n dichtheid van beide primaire subsystemendraag t het hogeretrofisch e niveau van deongewerveld e predatoren en parasitoiden. De verhoogde dichtheden van deze groepen werken deexponentiel e toename vanplaagherbivore n tegen. Vervolgens isd e totaledichthei d van zowel depredatore n alsd e primaire groepen vanbelan g alsvoedse l voor insectenetende vogelso p hetbedrijf . De dichtheid vanpredatoren , parasitoiden envogel s wordt dusveronderstel d toe teneme ni n verband met omschakelingsduur enE.I.-kwaliteit .

Het onderzoek bestond uit veldonderzoek enliteratuurstudie . Het veldonderzoek was vrijwel beperkt totFlevolan d omdat landbouwbedrijven daar eenovereenkomstig etopografi e en historie hebben. Ookbevind t zich inFlevolan d eenconcentrati e van ongeveer 75 biologische akkerbouw- en vollegrondsgroenteteeltbedrijven, waaronder bedrijven met een verbeterde ecologische infrastructuur.

Inhe t veldonderzoek werden waarnemingenverrich t aan epigeische geleedpotigen door middel van vangbekers,vegetatiebewonend e insecten doormidde lva n een zuigapparaat, en insectenetende vogelsdoo r middel vanterritoriumkarteringen .Ongewervelde n werden verzameld intarwevelde n eni n de ernaast gelegen slootkanten. Vogels en omgevingsvariabelenwerde n gemeten op bedrijfsniveau.

Deresultate n van develdonderzoeke n lieten, overeenkomstigd ehypothese , effecten zien op dehoeveelhede n van functionele groepen, van deduu r van debiologisch e bedrijfsvoering, intensiteit vand evruchtwisselin g end eomvan g enhe t maaibeheer vand eE.I .

Ind egewasse n bleken herbivoren echter het talrijkst tezij n oppa s omgeschakelde bedrijven met eenweelderi g gewas;typisch e plaagherbivoren (bladluizen) hadden de hoogste dichtheden op langomgeschakeld e bedrijven met een intensieve vruchtwisseling (Hoofdstuk 2).Waarneminge n aan detritivoren wezen welmee ri nd e richting van dehypothese . Detritivoren (vliegjes enmugjes )ware no p langomgeschakeld ebedrijve n met een extensieve vruchtwisseling relatief talrijk (Hoofdstuk 2). Ook epigeische predatoren (o.a. loopkevers en spinnen),di ewaarschijnlij k vooral verbonden zijn met detritivore ketens, lekentalrijke r te zijn in systemen met een extensievere teeltwijze (Hoofdstuk 3).Predatore n inhe t bladerdek vanhe t gewasbleke n geassocieerd te zijn met herbivoren; dezeware n talrijk in weelderige gewassen vanpa s omgeschakelde bedrijven enware n divers qua soorten op lang omgeschakeldebedrijve n met een intensievevruchtwisselin g (Hoofdstuk 2).

Het verbeterdebehee r in deE.I .had , overeenkomstig dehypothes e een sterk positief effect op dedichthei d van arthropoden enme t namepredatore n ind evegetati e (Hoofdstuk 2). Detritivoren waren echterd e talrijkste primaire groep;herbivore n waren weliswaar talrijk maarhadde n eenovereenkomstig e dichtheid inE.I .me t het traditionele klepelbeheer. De talrijkste epigeische groepen vanpredator ekever s enbaldakijnspinne n vertoonden eengewas - affiniteit (Hoofdstuk 3).Andere , minder talrijke groepen, zoalswolfspinne n waren meer geassocieerd metklepelbehee r dan met het verbeterde beheer. Verbeterde E.I.ha d dus (ind e zomer)voo r epigeische soorten weinig betekenis. Waarnemingen ind e gewas-E.I. gradient, lietenzie n dathog e dichtheden ind eE.I . kunnen uitstralen ind erichting va n het gewas.D e omstandighedeni nhe t gewas waren echter overheersend enkonde n zulke effecten teniet doen ofaa nhe t zicht onttrekken,bijvoorbeel d in gevalva n detoenam e vanherbivore n en predatoren inee nweelderi g gewas.

Dedichthei d van broedvogels bleek geassocieerd te zijn met zowel bedrijfs/gewaskenmerken, alsme t E.I.kenmerke n (Hoofdstuk 4). Watbetref t het bedrijf, 134 suggereren deresultate n dat langomgeschakeld e biologischebedrijve n met een intensieve vruchtwisseling, dehoogst e vogeldichtheid hadden.Mogelij k bevordert de periodieke toename vanplaagherbivore n devoedselbeschikbaarheid . Watbetref t de E.I.werde n duidelijke verbanden gevonden tussen vogeldichtheid enverhoogd e dichtheid van arthropodengroepen ind evegetati e van deE.I .

Ind e discussies van ieder empirischonderzoe k (hoofdstukken 2, 3e n4 ) zijn deresultate n in verband gebracht met voedselwebtheorieen. Develdwaarneminge n gaven indicaties voorhe t optreden van een indirect negatief effect van toegenomen detritivore voedselketens, via bevordering vanpolyphag epredatoren , op dedichthei d vanherbivoren . En ook indicaties voormogelijk e interacties tussenpredatorengroepen . Verder onderzoek is echter noodzakelijk.

Inhoofdstu k 6i s opbasi s van literatuur enveldonderzoe k eentopologisc h voedselweb opgesteld. Devariati e in destructuu r van dit voedselweb isvoorspel d aan dehan d van vier extreme bedrijfstypen dietwe e gradienten representeren watbetref t deduu r van biologische bedrijfsvoering enE.I .kwaliteit . Aanvullend isno g onderscheid gemaakt tussen lang omgeschakeldebedrijve n metverbeterd e E.I., watbetref t demat e vanherbivori e inhu n gewassen. Thema's dienaa r voren kwamen ind ediscussie s van de empirische onderzoeken (hoofdstukken 2, 3e n 4)werde n ondergebracht inhe t grotere verband van het voedselweb op bedrijfsniveau. Overeenkomstigd e oorspronkelijke hypothese, is deaanvoe r van voedselbronnen van onderaf inhe t voedselweb alsbepalen d aangewezen ('bottomup ' processen). Mogelijk zijn echter ookterugkoppelinge n vanbove n naarbenede n inhe t voedselweb belangrijk ('top down' effecten). Zulke effecten treden op alsepigei'sch e predatoren, bevorderd door detritivoreprooien , herbivore prooisoorten onderdrukken. Een complete onderdrukking vanplage n enoo k van andereherbivore n kan echter ongunstig zijn voor dichtheden van insectenetende vogels omdatdez e groep inbelangrijk e mate gevoed wordt viaherbivor e voedselketens ophe t landbouwbedrijf. Ecosystemservices e n natuurbelang zijn dan tegenstrijdig. De studie laatzie n dat eenvoedselwebbenaderin g veelbelovend is;hierme e is onderzoek mogelijk aand e effecten van interacties inhe t voedselweb, inclusief gevolgen voor landbouw ennatuurdoelen . Eventuele tegenstrijdigheid kan de aanleiding zijn tezoeke n naar alternatieven ter optimalisatie van beide doelen. Verder onderzoek is echter zeer noodzakelijk.

Naast dedri ehoofdstukke n over veldonderzoek inFlevolan d enhe thoofdstu k methe t conceptuele model, ise r een empirisch hoofdstuk 5da t een studie toont van diergroepen op tweebiologisch e bedrijven inFlevolan d entwe e inhe trivierengebied . Inbeid e landschappen was eenbedrij f met veel en eenbedrij f met weinig E.I.Dez ewaarneminge n tonen echter vooral het contrast tussen de landschappen. Extrapolatie van deFlevolands e gegevens naar andere landschappen ismogelij k maarverlang t mogelijk uitbreiding van de functionele groepen en eenbete r zicht op debijdrage n aandez e functionele groepen (inclusief prooien en predatoren) van bijvoorbeeld houtigebiotope n en onkruiden.

Hoofdstuk 7beschrijf t eenprocedur e voorhe t opstellen vanbedrijfsnatuurplannen . Dit instrument ispragmatisc h en streeft naarz o goedmogelijk e ontwerpen opbasi sva n ecologische kennis,veldbiologi e enpraktisch e mogelijkheden. Door het expliciet makenva n doelen encriteri a isgetrach t omd e stappen van waarneming, viawaarderin g naar beslissing teverheldere n enoo k teversnellen . Devraa g inhoeverr e eenbedrijfsnatuurpla n effectief is voor het bevorderen vanpredatore n ofbijzonder e dieren,wa s een aanleiding tot de voedselwebstudies. Verder onderzoek opbasi s van dit proefschrift kan ind e toekomst bijdragen aan adviesinstrumenten tenbehoev e van multispeciesfarming systems. 135 Curriculum Vitae

Frans Wouter Smeding wasbor n on the4 th September 1962i n Meppel, TheNetherlands . He graduated from theAthenaeu m of the Stedelijk Lyceum inZutphen , in 1981, and subsequently studied Biology atth e Wageningen Agricultural University. He specialised in vegetation science,plan t ecology andwee d science aswel l as alternative methods in agriculture andhorticulture . He achieved his MSc inMarc h 1989,includin g a qualification for first degree education. In 1989-1990h e contributed to the setting up of unsprayedfield margin research atth e department ofVegetatio n science,Plan t ecology and Weed science.I n November 1990h e started towor k atth e department ofEcologica l Agriculture whereh e participated in ongoing research and international education programs withreferenc e to holistic approaches. Heha s specialised there, since 1992,o nth e integration of agriculture and nature. In collaboration with the Dutch agricultural extension service, he developed in 1994-1995 aprotoco l for farm-nature plans.Thi swor k was followed by explorative research on farm food webs in relation to themanagemen t of crop andfield margin s (1996-2001).H e iscurrentl y employed as lecturer by theWageninge n University and Research Centre, atth e Biological Farming Systemsgroup ,Departmen t ofPlan t Sciences.

137