©EntomologicaFennica.11October2006 Carabid (Coleoptera) assemblages in the Scottish uplands: the influence of sheep grazing on ecological structure LornaJ.Cole,MegL.Pollock,DuncanRobertson,JohnP.Holland&DavidI.McCracken Cole, L. J., Pollock, M. L., Robertson, D., Holland, J. P. & McCracken, D. I. 2006:Carabid(Coleoptera)assemblagesintheScottishuplands:theinfluenceof sheepgrazingonecologicalstructure.—Entomol.Fennica17:229–240. AsaresultoftheEuropeanCommunityCommonAgriculturalPolicyreformin 2005itispredictedthatlivestockgrazingintheScottishuplandswillbecomeless intensive.Ateachoftwouplandresearchcentres,twolarge(>40ha)plotswere establishedtoinvestigatetherelationshipbetweengrazingintensityintheScot- tishuplandsandbiodiversity.Oneplotwasgrazedintensivelybysheepwhilethe otherwasgrazedextensively.Groundbeetlesweresampledbypitfalltrappingto determinetheinfluenceofgrazingpressureontheecologicalmake-upofground beetleassemblages.Grazingintensitydidnotsignificantlyinfluencecarabiddi- versity.However,grazingintensity,altitudeandmoisturedidinfluencethecara- bidecologicalassemblagestructureatbothlocations.LargeflightlessCarabus speciesweremoreabundantinextensivelymanagedplotsthanintensivelyman- agedplotsatbothlocations.Itislikelythattheselong-living,relativelyimmobile beetleswerefavouredbythegreaterstabilityofthevegetationstructureintheex- tensivelygrazedplots.Monitoringtheecologicalassemblagestructureprovides amoresensitiveapproachthandiversityindiceswhencomparingtheimpactof grazingandagriculturalmanagementbutisalsorobustenoughtoallowdirect comparisonsbetweendifferentgeographicallocations. L.J.Cole,D.RobertsonandD.I.McCracken,LandEconomyandEnvironment ResearchGroup,ScottishAgriculturalCollege,Auchincruive,Ayr,KA65HW,U. K.;Email:[email protected] M.L.PollackandJ.P.Holland,HillandMountainResearchCentre,Scottish AgriculturalCollege,KirktonFarm,Crianlarich,FK208RU,U.K. Received2November2005,accepted26May2006 1.Introduction which is determined by historical support pay- ments received by each farm. Thelink between The European Common Agricultural Policy subsidies and production has consequently bro- (CAP) has recently been reformed resulting in ken,andasaresultagriculturalmanagementwill major changes in the way farming in the EU is no longer focus simply on maximising produc- supported (Leguen de Lacroix 2003). Subsidies tion.Consequently,itispredictedthatagriculture which were previously production orientated willbecomemorelinkedtomarketdemandsand (e.g. based on the number of grazing animals), thatchangesin agriculturalsubsidieswillresult have been replaced by a Single Farm Payment in, among others, changes in grazing manage- 230 Coleetal. (cid:127) ENTOMOL.FENNICAVol.17 ment.InScotland,itispredictedthatgrazingin- nature frequently disturbed, and several studies tensity in the Scottish uplands will be signifi- haveshownthatlargeflightlesscarabidsaremore cantly reduced as a result of the CAP reform abundantinextensivelymanagedhabitats(Blake (Cook2004).Thereisconcernthatthiswillhave etal.1994,1996,Riberaetal.2001).Changesin anadverseimpactonuplandvegetationandthe grazingpressureandmowingintensity notonly wildlifeassociatedwithgrazedopenhabitats. affectthedisturbancefrequencyofahabitatbut Changestolivestockgrazingpracticesinflu- alsothevegetationstructureandpreyavailability, encethevegetationstructureandcompositionat whichinturnmayinfluencetheecologicalstruc- bothpatchandlandscapelevelsandthisinturn ture of carabid assemblages. Extensively man- alters ground beetle assemblages and diversity agedgrasslandistypicallyassociatedwithricher, (Luff&Rushton1989,Dennisetal.1997).Cara- morestructurallydiversevegetationwhencom- bidbeetlesareoneofthemostcommonfamilies paredtointensivelymanagedgrassland(Marriott of surface-active arthropods in the agricultural 2004),andconsequentlymayfavourherbivorous landscapeandarethoughttobepotentialindica- carabids such as Harpalus, Bradycellus and tors of disturbance, as they are ubiquitous and Amara species. Intensively managed grassland, easilysampledusingpitfalltraps,aresensitiveto ontheotherhand,hasbeenfoundtofavourCol- environmental factors, and are morphologically lembola specialists as a result of an increase in and ecologically diverse (Lövei & Sunderland prey abundance and a more open vegetation 1996,Niemeläetal.2000).Thesebeetleshavean structurefacilitatingthesevisuallyhuntingpred- importantroleinthefarmlandecosystem,asthey ators(Riberaetal.2001). arenotonlyimportantpolyphagouspredatorsof Cole et al. (2002) classified ground beetles awiderangeofagricultural pests (Thiele1977, intosevenecologicalgroups.Theprimaryobjec- Lövei&Sunderland1996),butalsoformanim- tiveoftheworkreportedherewastousetheeco- portant dietary component for many decreasing logicalgroupsofColeetal.(2002)todetermine farmland birds (Holland & Luff 2000). While whethergrazingpressureintheScottishuplands monitoringtheinfluenceofgrazingintensityon influencestheecologicalmake-upofgroundbee- theabundanceofdifferentspeciesisvaluablein tleassemblages. This study thus focused on the its own right (Eyre et al. 1989, Rushton et al. influenceofgrazingpressureintheScottishup- 1989, Purvis et al. 2001), findings at different lands on the ecological structure of carabid as- geographicallocationsarerarelydirectlycompa- semblages and evaluated the robustness of this rableduetodifferencesintheindividualspecies approachformonitoringenvironmentalimpactat presentateachlocation.Monitoringregimesthat differentgeographicallocations. focus on the ecological structure of a beetle as- semblagearelesslikelytobesensitivetothere- quirements of individual species and conse- 2.Materialandmethods quentlymaybemorerobustthanregimeslooking atspeciescomposition(Willbyetal.2000).Fur- 2.1.Studysites thermore,byclassifyingassemblagesonthebasis oftheirecologywegainabetterinsightintotheir Fourlarge(>40ha)uplandgrazedplotswerees- ecological functioning (Whittaker 1975) which tablishedattwoScottishuplandresearchstations: mayinturnhelpustopredicthowanecosystem Kirkton (Perthshire) and Sourhope (The Bor- will respond to future disturbance (Diaz & Ca- ders).Thetworesearchcentreswereseparatedby bido1997). adistanceof184km.Twoadjacentplotsateach Life-history traits are strongly linked to the researchstationwereeachallocatedtooneoftwo frequency of habitat disturbance and large, im- sheepgrazingintensities:anextensivelightsum- mobilespecieswithalowfecunditytendtobefa- mer-onlygrazingandanintensivemoderateyear- vouredinundisturbedhabitatswhilesmall,mo- roundgrazing(Table1).Thegrazingintensities bilespecieswithahighfecunditytendtobefa- weresetbytakingintoaccountthetypeofforage vouredindisturbedhabitats(Southwood1988). available in each of the plots. Grazing regimes Intensively managed grassland habitats are by werestartedinJune2001atKirktonandOctober ENTOMOL.FENNICAVol.17 (cid:127) Grazingintensityandcarabidcommunitystructure 231 Table1.Grazingtreatmentsatthetworesearchcentres. Site Nat.GridRef. Plotsize Altitude Grazing Opento livestock Kirkton,Intensive NN370306 44.4ha 390–590m 1–2sheepha–1 Jan–Dec Kirkton,Extensive NN368303 40.8ha 390–600m 1–1.3sheepha–1 Jun–Oct Sourhope,Intensive NT846217 74.9ha 270–470m 3–3.7sheepha–1 Jan–Dec Sourhope,Extensive NT843211 49.7ha 260–420m 3sheepha–1 Jun–Sep 2002atSourhope.Theaimoftheintensivegraz- Using the HFRO sward stick (Barthram ingregimewastocreatearelativelyshort,homo- 1986)25swardheightsweresampledwithin25 geneous,less-tussockysward.Theaimoftheex- cmofeachsideofatransectlinecorrespondingto tensiveregimewastocreateastructurallydiverse the line of pitfall traps. Mean sward height (in swardwithdistincttussockandinter-tussockar- cm),swardheterogeneity(standarderrorofvege- eas. Atotalof40locations(i.e., 10withineach tation height), transect evenness (abundance of plot)weresampledin2003andafurther20loca- the dominant species), transect diversity (Shan- tions (i.e., 5 within each plot) were sampled in non-Wienerindex)andtransectspeciesrichness 2004.Eightofthe2003locationsatSourhopebe- werecalculated. cameencroachedwithbrackenandwereconse- A map was produced of the vegetation quently relocated in 2004. Sampling locations patchesina30mdiametercirclecentredonthe were chosen to represent the range of altitudes, middlepitfalltrap.‘Patch’wasdefinedasanarea vegetationtypesandstructureswithineachplot ofvegetationofsimilarspeciescompositionand andwereaminimumdistanceof8mapart. structure,distinctfromadjacentvegetation.The plantcommunity ofeachofthepatcheswasre- corded, and 25 or 50 sward heights were mea- 2.2.Invertebratesampling suredfromeachofthemainpatches.Thenumber ofpatches,numberofplantcommunities,within- Groundbeetlesweresampledusingpitfalltrap- patchspeciesrichness,patchmeanswardheight, ping.Trapsconsistedofplasticbeakers(75mmin patchvegetationheterogeneity(standarderrorof diameter and 100 mm deep) partly filled with patchvegetationheight),patchperimeterlength 100% monopropylene glycol as a killing agent andpatchheightratio(maximumpatchheightdi- and preservative. To prevent small mammals videdby theminimumpatchheight)weremea- from entering the traps and to limit damage by sured. livestock,a15mmmeshgridwassecuredover thetrapmouthwithametalstaple(Downieetal. 2000).Ateachsamplinglocationa16mrowof 2.4.Physicalattributes ninepitfalltraps(2minter-trapdistance)wases- tablishedatthebeginningofMay.Thetrapswere Duringthepitfallsamplingperiod,foursoilcores leftinsituforaperiodoffourweeksandthesam- (6 cm diameter and 10 cm deep) were taken at ple from each row of 9 traps was collected and randomfromeachlineofpitfalltrapsandthesoil pooledinJune. wassubjectedtostandardsoilanalysestodeter- mine: pH,% moisture content,% organic matter content, phosphorus availability (mg/l) and po- 2.3.Vegetationsampling tassium availability (mg/l). Information on the soil penetrability (blf/in2) was collected using a Ateachsamplinglocation,vegetationheightand soilpenetrometer.Inadditiontheslopeangle(de- species composition were measured between grees),theaspect(degreesfromNorth),andthe JuneandAugusteachyearattwospatialscales: altitudeofeachpitfallsamplinglocationwerere- thetransectlevelandthewiderpatchlevel. corded. 232 Coleetal. (cid:127) ENTOMOL.FENNICAVol.17 Table2.SpeciesinventoryfortheintensiveandextensivelygrazedplotsatKirktonandSourhope(E=extensive, I=intensive).Theecologicalgroupofeachspeciesisprovided(seefootnote). Species Ecologicalgroup Sour- Kirk- Sour Kirk- hope,E ton,E hope,I ton,I Calathusfuscipes Mediumnocturnalpredators X – X – Cychruscaraboides Mediumnocturnalpredators – X – X Harpalusrufipes Mediumnocturnalpredators X – – – Nebriabrevicollis Mediumnocturnalpredators X X X X Patrobusassimilis Mediumnocturnalpredators – X – X Patrobusatrorufus Mediumnocturnalpredators – X – X Pterostichusadstrictus Mediumnocturnalpredators X X X X Pterostichusaethiops Mediumnocturnalpredators X X X X Pterostichusmadidus Mediumnocturnalpredators X X X X Pterostichusmelanarius Mediumnocturnalpredators X – X – Pterostichusniger Mediumnocturnalpredators X X X X Carabusarvensis Carabusspp. – X – X Carabusglabratus Carabusspp. – X – – Carabusnemoralis Carabusspp. X X X – Carabusnitens Carabusspp. – X – X Carabusproblematicus Carabusspp. X X X X Carabusviolaceus Carabusspp. X X X X Amaraaenea Diurnalherbivores X – X – Amaracommunis Diurnalherbivores X X X – Amarafamiliaris Diurnalherbivores X – X – Amaralunicollis Diurnalherbivores X X X X Loricerapilicornis Collembolaspecialists X X X X Notiophilusaquaticus Collembolaspecialists X – – – Notiophilusbiguttatus Collembolaspecialists – X X X Agonumassimile Smallnocturnalpredators – X – X Calathusmelanocephalus Smallnocturnalpredators X – X – Pterostichusrhaeticus Smallnocturnalpredators X X X X Agonumfuliginosum Small/mediumpredators – X X X Agonumgracile Small/mediumpredators – – X – Agonummuelleri Small/mediumpredators – X X X Bembidionguttula Small/mediumpredators – – X – Bembidionunicolor Small/mediumpredators – – X – Clivinafossor Small/mediumpredators – X X X Elaphruscupreus Small/mediumpredators – X X X Elaphruslapponicus Small/mediumpredators – X – X Elaphrusuliginosus Small/mediumpredators – X – X Leistusrufescens Small/mediumpredators – – X – Pterostichusdiligens Small/mediumpredators X X X X Pterostichusstrenuus Small/mediumpredators X X X X Pterostichusversicolor Small/mediumpredators – X X X Trechusobtusus Small/mediumpredators – X X X Harpaluslatus Nocturnalherbivores – X – X Trichocellusplacidus Nocturnalherbivores X – – – Agonumpiceum Insufficientinformation – – X – Bradycelluscollaris Insufficientinformation – – X – Carabusgranulatus Insufficientinformation – X – X Elaphrusriparius Insufficientinformation – X – X Harpalusquadripunctatus Insufficientinformation X – – – Totalspeciesrichness 23 33 32 30 Totalcarabidabundance 1145 1207 1235 1436 ThesegroupscorrespondtoColeetal.(2002)ecologicalgroupsasfollows:Mediumnocturnalpredators–Group1,Carabusspecies–Group2,Diur- nalherbivores–Group3,Collembolaspecialists–Group4,Smallnocturnalpredators–Group5,Smalltomediumpredators–Group6,Nocturnal herbivores–Group7. ENTOMOL.FENNICAVol.17 (cid:127) Grazingintensityandcarabidcommunitystructure 233 2.5.Statisticalanalysis tionofthecompositionofgroundbeetleassem- blages,theyprovidelittleinformationontheeco- The9pitfalltrapscollectedateachlocationwere logical structure of the community. To obtain pooled prior to carabid identification. For each comprehensive information on the ecological sampling year the following measures of diver- makeupofthecarabidassemblagesateachloca- sitywerecalculated:thenumberofgroundbeetle tion,therelativeabundanceofgroundbeetleeco- species(S)tomeasurespeciesrichness,Berger- logicalgroups–asdescribedbyColeetal.(2002) Parkerdominanceindex(d)tomeasureevenness –wascalculatedforeachsamplinglocation(Ta- and Simpson’s diversity index (N2) to measure ble2).Fivespecies(ofthe48speciescollected) overalldiversity.Magurran(1988)recommends thatlackedsufficientecologicaldatawereomit- combiningdiversityindicestoobtainamoreac- ted from these analyses (Table 2). All omitted curatepictureofdiversitythansimplyconsider- specieswereextremelyrare,andconsequentlyit ingoneortwooveralldiversityindices.Priorto is unlikely that their omission influenced the the analyses, diversity indices were log-trans- analyses.Two-wayANOVAswereconductedon formedwhererequiredtoapproachnormalityof thearcsin-transformeddataforeachresearchsta- thecompareddistributions.ANOVAswerecon- tionindependentlytodetermineifgrazinginten- ducted(ateachresearchstationindependently)to sityandyearofsamplinginfluencedtheecologi- determinehowyearofsamplingandgrazingre- calmakeupofthegroundbeetleassemblages. gimeinfluencedgroundbeetlediversity.The20 Canonical Correspondence Analysis (CCA) newlocationssampledin2004,andthebracken was conducted for each research station inde- locations sampled in 2003 at Sourhope were pendently to determine the principal environ- omittedfromthisanalysis.Wheresignificantdif- mentalcomponentsdrivingtheecologicalassem- ferences were indicated by ANOVA, Tukey’s blagestructure.CCAswereconductedontherel- post hoc tests were applied to locate the differ- ativeabundanceofecologicalgroups.Atotalof ences. 20 continuous environmental variables and one Whilediversityindicesprovidesomeindica- categorical variable (Grazing Intensity) were Table3.EnvironmentalvariablestakenintoconsiderationinCCA. Environmentalvariable Description Moisture Soilmoisturecontent% Organicmatter Soilorganicmattercontent% pH SoilpH Phosphorus Availabilityofphosphorusinthesoil(mg/l) Potassium Availabilityofpotassiuminthesoil(mg/l) Penetrability Soilpenetrability(blf/in2) Aspect Aspectofthepitfalltraps(degreesfromnorth) Altitude Altitudeofthepitfalltraplocation(m) Slope Slopeofthepitfalltraplocation(degrees) Grazingintensity Grazingintensityofplot(intensiveorextensive) Transectheight Transect:meanswardheight(cm) Transectheterogeneity Transect:standarderrorofvegetationheight Transectrichness Transect:numberofplantspecies Transectevenness Transect:%abundanceofthedominantplantspecies Transectdiversity Transect:Shannon-Wienerindexofvegetation Patchratio Patch:maximumvegetationheightdividedbyminimumvegetationheight Numberofpatches Patch:numberofdifferentvegetationpatches Numberofcommunities Patch:numberofdifferentvegetationcommunities Patchrichness Patch:numberofspeciesinpatch Patchlength Patch:thetotalpatchedgelength Patchheterogeneity Patch:standarderrorofmeanpatchheight 234 Coleetal. (cid:127) ENTOMOL.FENNICAVol.17 considered in the analyses (Table 3). Environ- mentalvariableswerenormalisedbylogarithmic transformation where required. The eight brackenencroachedlocationsatSourhopewere omittedfromthisanalysisduetoinsufficientveg- etationdata.Aforwardselectionprocesswasuti- lisedandonlyvariablesfoundtobestatistically significant(atthe5%level)bytheMonteCarlo Permutation test were included in the analyses, thus reducing problems associated with multi- collinearity(terBraak&Šmilauer2002). Fig.1.Mean(+SD)relativeabundanceofCarabus speciesandCollembolaspecialists(Group2and4, 3.Results respectively)onintensivelyandextensivelygrazed plotsatKirktonandSourhope. Atotalof2,643carabidsconsistingof33species werecollectedfromKirktonand2,380carabids nificantly more species being recorded in 2003 consistingof36speciesfromSourhope(Table2). than2004(Table4b). At both research stations, large, immobile beetlesofthegenusCarabusweremoreabundant 3.1.Influenceofmanagementandsampling on extensively managed plots than intensively yearondiversityandecologicalgroup managedones(Table4a–b,Fig.1).Therelative abundance of Carabus species at Kirkton was Two-way ANOVAs (Table 4a–b) indicated that overtwotimeshigherintheextensiveplotthan grazing intensity did not influence the species the intensive plot, while at Sourhope the differ- richness (S), evenness (d) or overall diversity ence was even more profound with almost a 5 (N2) of the carabid assemblages at Kirkton or fold increase in relative abundance between the Sourhope.AtKirkton,theyearofsamplinghad intensiveandextensiveplots.Thisdifferencewas noinfluenceonanyofthediversityindicesstud- found despite Carabus species being predomi- ied, while at Sourhope a significant year effect nantlyautumnbreedersandhencepotentiallyless wasfoundforcarabidspeciesrichnesswithsig- active in spring when the samples were taken. Fig.2.Theecological make-upoftheinten- sively-andextensively- managedplotsat KirktonandSourhope. ENTOMOL.FENNICAVol.17 (cid:127) Grazingintensityandcarabidcommunitystructure 235 Table4.Resultsof2-wayANOVAforgroundbeetlediversityandrelativeabundanceofdifferentecological groupsatKirkton(a)andSourhope(b).LocationofsignificantdifferencesfoundusingTukey’sposthoctests(p <0.05).n/a=notanalysedduetoinsufficientdata. a)Kirkton Index/Group Treatment(df=1,36) Year(df=1,36) SpeciesRichness(S) F=1.53,p=ns F=0.26,p=ns Berger-Parker(d) F=0.95,p=ns F=2.60,p=ns Simpson's(N2) F=0.93,p=ns F=2.75,p=ns Mediumnocturnalpredators F=0.01,p=ns F=2.70,p=ns Carabusspp. F=7.17,p=0.01(Ext>Int) F=0.53,p=ns Diurnalherbivores n/a n/a Collembolaspecialists F=24.57,p<0.001(Ext<Int) F=1.07,p=ns Smallnocturnalpredators F=1.82,p=ns F=4.27,p<0.05(2003<2004) Small/mediumpredators F=0.48,p=ns F=3.20,p=ns Nocturnalherbivores F=0.08,p=ns F=0.12,p=ns b)Sourhope Index/Group Treatment(df=1,20) Year(df=1,20) Speciesrichness(S) F=0.01,p=ns F=5.34,P=<0.05(2003<2004) Berger-Parker(d) F=0.62,p=ns F=0.85,p=ns Simpson's(N2) F=0.11,p=ns F=0.33,p=ns Mediumnocturnalpredators F=0.09,p=ns F=0.05,p=ns Carabusspp. F=10.75,p<0.005(Ext>Int) F=1.35,p=ns Diurnalherbivores F=0.73,p=ns F=2.17,p=ns Collembolaspecialists F=0.00,p=ns F=1.58,p=ns Smallnocturnalpredators F=0.03,p=ns F=2.51,p=ns Small/mediumpredators F=0.87,p=ns F=1.57,p=ns Nocturnalherbivores n/a n/a The only other ecological group influenced by were well represented at Sourhope but largely grazingintensitywasCollembolaspecialists.At lackingatKirkton.Despitedifferencesintheeco- Kirkton,therelativeabundanceofthisgroupwas logicalmakeupoftheassemblagesatthetwore- fourtimeslowerintheextensivelymanagedthan search station, consistent trends in the relative intheintensiveplot(Table4a,Fig.1).AtSour- abundance of Carabus species were found be- hope, no difference was observed between the tweenextensivelyandintensivelymanagedplots twograzingtreatments(Table4b,Fig.1). (Fig.2). Therelativeabundanceoffunctionalgroups inthepitfalltrapsappearedtobefairlyconstant betweensamplingyearswithonlysmallnoctur- 3.2.Influenceofenvironmentalfactors nalpredatorsatKirktonbeinginfluencedbysam- ontheecologicalassemblagestructure plingyear.Thisgrouphadahigherrelativeabun- dancein2004thanin2003(Table4a,Fig.1).This CCAfortheSourhopedataproducedeigenvalues indicatesthattheecologicalassemblagestructure of 0.318, 0.18, 0.07 and 0.03 for axes 1–4, ac- was relatively consistent between sampling countingfor26.7%,15.7%,5.9%and2.7%ofthe years. total variation in ecological structure, respec- Cleardifferenceswerefoundintheecological tively (Fig. 3). CCA for the Kirkton data pro- make-upofcarabidassemblagesatKirktonand duced eigenvalues of 0.124, 0.070, 0.025 and Sourhope (Fig 2). Small nocturnal predators 0.004foraxes1–4,accountingfor22.9%,12.9%, dominated the community at Kirkton, while at 4.7%and0.7%ofthetotalvariationinecological Sourhopethecommunitywasdominatedbyme- structure, respectively (Fig. 4). The low eigen- dium sized nocturnal predators. Furthermore, valueswereexpected,astheordinationwascon- small to very small diurnal herbivore species ducted on only seven ecological groups as op- 236 Coleetal. (cid:127) ENTOMOL.FENNICAVol.17 Fig.3.CCAtriplotfortheKirktonecological-grouprelative-abundancedatashowingintensiveandextensive sites,carabidecologicalgroups(blacktriangles;seeTable2),continuousenvironmentalvariables(vectors)and intensityasacategoricalenvironmentalvariable(blacksquareboxnearorigin).Onlyenvironmentalfactorssig- nificantatthe5%levelareincludedinthegraph(seeTable5). Fig.4.CCAtriplotfortheSourhopeecological-grouprelative-abundancedatashowingintensiveandextensive sites,carabidecologicalgroups(blacktriangles;seeTable2),continuousenvironmentalvariables(vectors)and intensityasacategoricalenvironmentalvariable(blacksquareboxup-and-rightfromtheorigin).Onlyenviron- mentalfactorssignificantatthe5%levelareincludedinthegraph(seeTable5). posedtoalargenumberofindividualspecies.For AtKirkton,theCCAshowedaclearsepara- bothlocations,axes1and2accountedformostof tion of intensively vs. extensively grazed plots thevariationintheecologicalstructureofcarabid alongaxis2(Fig.3).AtSourhope,thedifference assemblages. between the intensively and extensively grazed ENTOMOL.FENNICAVol.17 (cid:127) Grazingintensityandcarabidcommunitystructure 237 Table5.InfluenceofenvironmentalvariablesoncarabidecologicalassemblagestructureasindicatedbyCCA. OnlyfactorsfoundsignificantintheMonte-Carlopermutationtest(p<0.05)areshown. Environmentalfactor Kirkton Sourhope Slope F=8.00,p<0.001 – Grazingintensity F=5.25,p<0.005 F=3.92,p<0.005 Altitude F=5.05,p<0.005 F=12.96,p<0.005 Moisture F=4.37,p<0.01 F=4.92,p<0.05 Phosphorusavailability F=3.18,p<0.05 – pH – F=4.39,p<0.05 Organicmatter – F=3.21,p<0.01 Transectdiversity – F=2.57,p<0.05 plotswasnotsoapparent(Fig.4).Theseparation pressurecontributedtothisdifferenceinvegeta- ofsamplinglocationsalongaxis1wasprimarily tionstructure. aconsequenceofaltitude.Thelower-altitudelo- cationsinboththeintensivelyandtheextensively grazedplotshadhigheraxis1scoresthanhadthe 4.Discussion higher-altitude locations. At Sourhope, sheep grazing was more intensive at the foot of both Grazing intensity in the uplands can influence plots where the vegetation was more nutritious carabidsatboththespeciesandassemblagelevel than on the poorer quality vegetation found at (Dennisetal.1997;Luff&Rushton1989).Habi- higheraltitudes. tataffinityofdifferentspeciescandifferbetween At Kirkton, five environmental variables various geographical locations. Species de- (Slope,GrazingIntensity,Altitude,Moistureand scribedbyThiele(1977)asbeingstrictlywood- Phosphorus Availability) significantly influ- landspecieshavealsobeenassociatedwithlong encedtheecologicalstructureofthegroundbee- grass,heathormoorlandintheU.K.(Luff1998). tle assemblage, while at Sourhope six environ- Consequently, findings at the species level are mentalvariables(Altitude,OrganicMatter,Graz- rarelytransferabletodifferentgeographicalloca- ing Intensity, Moisture, pH, Transect Diversity) tions.Diversityindices,beingunrelatedtotheac- did so (Table5). Threeenvironmental variables tualspeciespresent,mayprovideamorerobust influenced ecological structure at both research way of monitoring the influence of agricultural stations (i.e., Grazing Intensity, Altitude and managementacrossdifferentgeographicalloca- Moisture) indicating the potential of these vari- tions.Diversityindiceshavebeenwidelyusedin ablestobeusedaspredictorsofecologicalstruc- carabid studies (Kromp 1989, Luff & Rushton ture. 1989,Bhriainetal.2002,Shahetal.2003)with AtKirkton,vegetationattributesdidnotinflu- manystudiesindicatingahighercarabiddiversity encethecarabidecologicalassemblagestructure in extensively than intensively managed fields indicatingtheimportanceofsoilandtopography. (Luff&Rushton1989,Blakeetal.1996,Kromp AtSourhope,thediversityofthevegetationatthe 1989).Whiletheuseofdiversityindicesmaybe transect level, significantly influenced the eco- robust,muchoftheecologicalinformationislost. logicalstructure.AtSourhope,theloweraltitude Bysolelylookingatthediversityofacommunity samplinglocationsonbothplotsweredominated wemayrisktooversimplifythedataandconse- byAgrostisspecies(i.e.,A.caninaandA.capil- quentlylooseinformationthatmaybeapparentat laris), while the higher altitude locations were a finer scale. The use of ecological groupings dominatedbymosaicsofAgrostisspecies(i.e.,A. couldprovideausefulwayofretainingsuchin- caninaandA.capillaris)andNardusstricta.Itis formationwhileallowingrobustcomparisons. likelythatacombinationofaltitude,soilfactors Inthisstudy,grazingintensitydidnotsignifi- (e.g., soil pH and organic content) and grazing cantly influence carabid diversity but it influ- 238 Coleetal. (cid:127) ENTOMOL.FENNICAVol.17 enced the ecological structure, indicating that relative abundance of Collembola specialists at monitoringattheecologicallevelwasmoresen- Kirkton was related to an increase in the abun- sitive than monitoring diversity. Large species danceofprey. belongingtothegenusCarabushadahigherrela- Thevegetationdiversityofintensivelyman- tiveabundanceinextensivelymanagedplotsthan aged grassland tends to be more impoverished on intensively managed ones. These findings (bothintermsofthenumberofspeciesandstruc- wereconsistentnotonlybetweensamplingyears, turaldiversity)thanthatofextensivelymanaged butalsobetweenthetworesearchstations.Blake grassland(Marriott2004).Thisheldtrueforthe etal.(1994),Riberaetal.2001andTietze(1985) studysiteswherevegetationdiversity(measured also found that extensively managed grassland bytheShannon-Wienerindex)wasricherinthe habitatsfavouredlargercarabids,andColeetal. extensivelymanagedplotswhencomparedtothe (2002)foundthattheabundanceofCarabusspe- intensively managed plots. It may therefore be cies was greater in heather moorland and semi- predicted that diurnal and nocturnal herbivores natural grassland than on intensively managed wouldbemoreabundantintheextensivelyman- grass and arable land. Šustek (1987) found that agedplots.Thomasetal.(2001)foundthather- urbanisationhadasignificantinfluenceoncara- bivorouscarabids(i.e.,HarpalusandAmaraspe- bid body size with more urban areas having cies)wereassociatedwiththebotanicallydiverse higher numbers of small species while undis- fieldmarginsinintensivelymanagedagricultural turbedhabitatshadhighernumbersoflargesed- land,andthesespecieshavealsobeenshownto entaryspecies.Itwouldthereforeappearthatthe be associated with weed cover in arable crops relative abundance of large flightless carabids (Holland&Luff2000).Grazingintensity,how- (e.g.,Carabusspecies)mightprovidearobustin- ever,didnotsignificantlyinfluencenocturnalor dicator of habitat disturbance. It is likely that diurnal herbivores in this study. Herbivorous large,long-living,flightlessbeetlesarefavoured carabids tend to favour vegetation associated bythegreaterstabilityinvegetationstructurethat withdungpatches(Dennisetal.1997)andconse- accompaniesextensivegrazingpracticesandare quentlytheyarelikelytohaveapatchydistribu- less well adapted to the fluctuating conditions tionmakingaccuratesamplingdifficult. characteristic of intensively grazed sites. Most Despitetheecologicalstructureofthecarabid largeCarabusspeciesareautumn-breederswith assemblagesbeingquitedifferentatthetwoloca- over-wintering larvae and (at least in Scotland) tions, changes in the relative abundance of two-year breeding cycles (Ribera et al. 1999). Carabus species between extensive and inten- Immobilespecieswithlong-lifecyclesrequirere- sively managed plots were consistent. Monitor- sourcesthatarestableovertime. ingmethodsthatlookattherelativeabundanceof In agreement with the findings of other au- Carabusspeciesarethereforerobustand,atleast thors(Riberaetal.2001,Coleetal.2002),Col- inthisstudy,weremoresensitivethandiversity lembola specialists had a higher relative abun- indices. Thereisconsensusintheliteraturethat danceinintensivelygrazedlocationsatKirkton agricultural disturbance adversely influences whencomparedtoextensivelygrazedlocations. larger carabids and consequently it is possible The same trend was not observed at Sourhope. that the influence of management on Carabus Collembolaspecialistspreferamoreopensward species could simply be the result of their large asthisfacilitateshuntingbyvisualcues.Itispos- size.Whilesizemaybeasuitablesubstitutefor sible that intensive grazing resulted in a more ecological groupings, and indeed would be far openswardstructureatKirktonthanatSourhope simplertodetermineforspeciesnotintheorigi- hencefavouringCollembolaspecialists.Ouror- nalclassificationbyColeetal.(2002),itislikely dinations, however, indicated that vegetation that monitoring methods looking simply at size characteristicsdidnotinfluencethecarabideco- distributionswouldbelesssensitive.Forexam- logicalstructureatKirktonandgrazingintensity ple,itishighly unlikely thatmethodsanalysing andaltitudeappearedtobetheprincipalenviron- carabidassemblagessimply onthebasisofsize mental factors influencing Collembola special- distributionwouldhavepickeduptheinfluence ists. Consequently it is possible that the higher ofmanagementintensityonCollembolaspecial-