BasicandAppliedEcology16(2015)335–346 Solidago canadensis impacts on native plant and pollinator communities in different-aged old fields Annamária Fenesia,b,∗, Csongor I. Vágásic,d, Monica Beldeanb, Rita Földesie, Levente-Pé ter Kolcsárb , Julie Te re sa Shapiroe ,f, Edina Törökb, Anikó Kovács -Hostyáns zkie aDepartmentofEcology,UniversityofDebrecen,Egyetemtér1,H-4032Debrecen,Hungary bHungarianDepartmentofBiologyandEcology,Babes¸-BolyaiUniversity,Republiciistreet42, RO-400015Cluj-Napoca,Romania cEvolutionaryEcologyGroup,HungarianDepartmentofBiologyandEcology,Babes¸-BolyaiUniversity, ClinicilorStreet5–7,RO-400006Cluj-Napoca,Romania dMTA-DE“Lendület”BehaviouralEcologyResearchGroup,DepartmentofEvolutionaryZoology, UniversityofDebrecen,Egyetemtér1,H-4032Debrecen,Hungary eMTAÖK,LendületEcosystemServicesResearchGroup,Alkotmányút2–4,H-2163Vácrátót,Hungary fSchoolofNaturalResourcesandEnvironmentandDepartmentofWildlifeEcologyandConservation, UniversityofFlorida,Gainesville,FL32611,USA Received27August2014;accepted12March2015 Available on line19M arch 2015 Abstract Secondarysuccessioninformerarablefields(i.e.oldfields)mightbealteredbythecolonizationofinvasivealienspecies, withpossiblecommunity-wideimpacts,hinderingtheabilityofoldfieldstobecomespecies-richcommunities.However,the effects of invasive species on local communities have rarely been addressed in the light of secondary succession. Therefore, we studied the impact of the highly invasive Solidago canadensis on plant and pollinator communities along a gradient of invasion severity in old fields with different ages (1–20 years since last ploughing) in Southern Transylvania, Romania. We askedwhethertheinvasionofS.canadensiscausesshiftsin(1)thecompositionanddiversityofplantcommunities,and(2) pollinator communities along the successional gradient. Further, we asked (3) to what extent the presence of S. canadensis affectedflowervisitationofnativeplantspeciesbypollinators.Accordingtoourresults,theinvasionreducedthenativeplant speciesrichnessthroughoutsuccession,althoughthemostprofoundnegativeeffectonplantdiversityandvegetationnaturalness wasexertedinoldersuccessionalcommunities.TheinvasionofS.canadensishadanegativeeffectontheabundanceofbees irrespectiveoftheoldfieldage;however,therewasnosimilarnegativeeffectonhoverflies.Nativeflowersexperiencedreduced visitation by wild bees, honey bees and hoverflies due to the augmented presence of S. canadensis. Therefore, the invasion of this perennial plant species diverts the trajectory of vegetation succession, alters the mutualistic links between the native elementsoftheseoldfields,andcausesanon-desiredalternativestablestatestobeinstalled. ∗Correspondingauthorat:Babes¸-BolyaiUniversity,HungarianDepartmentofBiologyandEcology,Republiciistreet42,400015Cluj-Napoca,Romania. Tel.:+40744916952;fax:+40264431858. E-mailaddress:[email protected](A.Fenesi). http://dx.doi.org/10.1016/j.baae.2015.03.003 1439-1791/©2015GesellschaftfürÖkologie.PublishedbyElsevierGmbH.Allrightsreserved. 336 A.Fenesietal./BasicandAppliedEcology16(2015)335–346 Zusammenfassung Die sekundäre Sukzession auf ehemaligen Feldern könnte durch die Besiedelung mit invasiven fremden Arten verändert werden, die möglicherweise die ganze Gemeinschaft beeinflussen, indem sie die Entwicklung hin zu artenreichen Gemein- schaftenerschweren.IndessenwurdendieEinflüssevoninvasivenArtenauflokaleGemeinschaftenselteninHinblickaufdie sekundäreSukzessionbetrachtet.WiruntersuchtendeshalbdenEinflussderhoch-invasivenKanadischenGoldrute(Solidago canadensis) auf Pflanzen- und Bestäubergemeinschaften entlang eines Gradienten unterschiedlich starker Invasion auf alten Feldern unterschiedlichen Alters (1 bis 20 Jahre seit dem letzten Pflügen) in Süd-Transsilvanien (Rumänien). Wir fragten, obInvasiondurchdieGoldruteVeränderungeninderZusammensetzungundDiversitätderPflanzen-undBestäubergemein- schaftenentlangdesSukzessionsgradientenhervorruft.Desweiterenfragtenwir,inwieweitdieAnwesenheitderGoldruteden Blütenbesuch an einheimischen Pflanzen beeinflusst. Die Invasion reduzierte den Artenreichtum der einheimischen Pflanzen über die gesamte Sukzession hinweg, aber der stärkste negative Effekt auf die Pflanzendiversität und die Natürlichkeit der VegetationerfolgteindenälterenSukzessionsgemeinschaften.DieGoldruten-InvasionhatteunabhängigvomAlterderFelder einennegativenEffektaufdieAbundanzderBienen,aberesgabkeinengleichartigenEffektaufdieSchwebfliegen.Einheimi- schePflanzenerfuhreneinenreduziertenBlütenbesuchdurchWildbienen,HonigbienenundSchwebfliegen,derdervermehrten AnwesenheitderGoldrute geschuldet war.DeshalbändertdieInvasiondurchdieseausdauernde Pflanzenart denVerlaufder Vegetationssukzession,ändertdiemutualistischenVerbindungenzwischendeneinheimischenArtenderehemaligenFelderund verursachtdieEtablierungeinesunterwünschten,alternativenstabilenZustandes. ©2015GesellschaftfürÖkologie.PublishedbyElsevierGmbH.Allrightsreserved. Keywords: Canada goldenrod; Plant invasion; Plant–pollinator interaction; Fallow; Set-aside; Apis; Romania; Natura 2000; Biodiversity conservation Introduction invertebrate communities, such as pollinator insects (van Hengstum,Hooftman,Oostermeijer,&vanTienderen2014). Over the last few decades, various social and economic Formerstudiespresentedhighlycontroversialresults,repor- changes hav edr iven theaband onment ofext ensiv eareasof ting eith er pos itive (Bart omeus, Vilà, & San tamaría 2008) formerly arab le land s in several parts of the world. If the se or n egative effects (de Groot, Kl eijn & J ogan 2007; M oron´ areasare leftalo neand p ropagul esour ce sare availa ble ,sec- et al. 2009 ) on po llin ator ab undanc e and di versity . Inva- ondar ysu cce ssiont akes placeand diverses em i-naturalv ege- siv e p lants are reported t o influence the abundanc e and tationc anbeforme dins hortti me( Stoatee tal.2009).Fo rmer dive rsity of sm aller, mo stly solitary w ild bees (Apoi dea) arable lan ds (hereaft er oldfi elds) maysu pp ort farmla ndbio- oflimited di spersala bility(G athmann &T scharn tke2002), diversi tyand providev alua bleeco syste mservi ces(e.g.p olli- as well as hoverfli es (Syr phidae) and b umblebees, which nation)at the beginnin gofseco ndarysucc ession(T scha rntke, ge nerall yh avelarger foragingrang es(G reenleaf,Wi lliams, Batáry, & Do rmann201 1) andtheir conservatio nvaluemay Winfree,& Kre men2 007).The refore, invasivespe ciesmight increase in the later stages (Co rbet 1995). Howe ver, re ach- threaten bio diversity not o nly at the local lev el, but at the ingasta ge ofd ivers esecon daryve getation isnotfre efrom landscap e scale as w ell (Hejd a, Py sˇek, & Jaro sˇík 20 09). cha lle nges, as oldfield sarehighl ysusceptib le toin vasi onby Moreover, invas ive plan t specie s can a lso act ind irectly alienplants pe cies (Catf ord etal.2 012).Non-n at iveinvas ive on ecologi cal comm unitie s by we aken ing t he m utualistic speci esare likelyto colonize th efr eshlya bandonedc roplands lin ks between native plants an d their pol linat ors (Aizen, because of increa se dsoilnut rien tavaila bilityassoc iatedwith Mora les, & M orales 2008) by lurin g pollinato rs away previous fe rtilizer ap plic ations, a nd limited competition for from nat ive plants, e ither p arti ally (B askett, Eme ry, & resources (Standis h, Cramer, & Hob bs 200 8). Longer-li ved Rudg ers201 1)orco mpletel y(Bartom eusetal. 2008).T his alienspec iescanper sistuntil th elaters tageso fsuccession, expansio n of p lan t preference for pollina tio n c an resu lt in there fore per turb ing natu ral s ucce ssion pathw ay s and alter- fewerpolli nat orvis its,pervasiv ein terspecificp ollen trans fer ing the p attern of ve getation recovery ( Meiners, Pick ett, & (Bask ett et al. 2 011), and cons equently, de creased repro- Cad ena sso 200 2). By preven ting the e stablishme nt of la te- duction suc ces s and d ivers ity of the nat ive flora. D espite succession al spec ies, alien plant inv asion may ca use the these po ssible d etrim ental effe cts , the direct effect of inva- invaded habi tats to re main in the early suc cessio nal sta ges sive plant spe cies on sec ondary succ ession of o ld fields foralon gtime(C ra mer,Ho bb s,& Stan dish2008). is re lativel y under stud ied (Flory & Clay 201 0), wh ile the T h e est ablis hment of invasiv e p lants ma y influence not ef fects of in vasive specie s on th e pollin ator co mmuni ties only th e native flora an d the tr ajector y of its second ary along the successi onal tra ject ory has so fa r never been succe ssio n, but can also have an impa ct o n the related studie d. A.Fenesietal./BasicandAppliedEcology16(2015)335–346 337 ◦ Here, we focus on a large-scale biological invasion phe- annual temperature is 8 C (Dra˘gulescu 2003). The region nomenon in southern Transylvania, Romania. Large-scale hasbeendesignatedasNatura2000SiteofCommunityInter- abandonmentofarablelandshasoccurredoverthelasttwo est(Sighis¸oara-TârnavaMareNatura2000site,ROSCI0227) decades due to a significant class of absentee landowners because18habitatslistedintheEUHabitatsDirectiveAnnex and the decreasing profitability of traditional subsistence Icanbefoundinthisregion. agriculture (Fischer, Hartel, & Kuemmerle 2012). Solidago We examined the succession of vegetation and pollina- canadensis L. (Canada goldenrod) is the most successful torcommunitiesinthissiteusingachrono-sequenceofold invasivecolonizeroftheseoldfields,oftenbecomingdom- fields representing different stages of succession (space for inant even at the landscape scale. Its impact on the species time substitution, sensu Pickett 1989). We chose old fields richnessofvascularplantsandpollinatorshasalreadybeen with a known abandonment date in the vicinity of four studied in other countries (de Groot et al. 2007; Moron´ villages(Biertan,Laslea,Malâncrav,Richis¸;Fig.1A).Aban- et al. 2009). However, these studies analysed only the final donmentherecanbedefinedasthecessationofploughing, stage of invasion, when the invasive species had already though other uses and activities such as mowing or graz- establisheddominanceinthecommunity,andcomparedthis ing may be present in the fields. These old fields may also stage with non-invaded, semi-natural communities, repre- be infrequently burned by locals. The 36 fields examined senting only the two contrasting extremes of the invasion rangedinagefrom1to20yearssinceabandonment.These gradient. sites were also chosen to represent the whole range of S. To fill these gaps in our knowledge regarding the impact canadensisabundancefromhighlytolessinvadedareas(rel- of invasion on both plant and pollinator communities over ativecover0–80%).Mostsiteswerequitesmall(meanarea: thecourseofsecondarysuccessiononoldfieldsalongacon- 1.3ha, range: 0.08–2.7ha). The year of last ploughing, and tinuousgradientofinvasion,weinitiatedalarge-scalefield currentland-usepracticesweredeterminedbyinterviewing study.Weaimedtoanswerthefollowingquestions:(1)Does landowners, and based on that information, we categorized the invasion of S. canadensis cause shifts in the composi- the old fields as mown, grazed or without management. tion and diversity of plant communities over the course of To determine whether the sites had been recently burned, secondarysuccessionofoldfields?(2)Doesabundanceand or not, we looked for local signs of fire (burned trees and species richness of pollinators change due to the invasion shrubs, incinerated litter or grass tussocks) and asked local of the old fields? (3) To what extent does the presence of people. S.canadensisaffectflowervisitationofnativeplantspecies by pollinators? To answer these questions, we studied the effect of S. canadensis density (0–80% relative cover) in Vegetationsurvey 36 old fields with various ages (1–20 years since the last ploughing). Percentcoverofvascularplantspecieswitharesolutionof 1%wasvi sually es timatedw ithin three4 ×4m plotspers ite (Fig.1B).Weassigned0.5%tospecieswithacoversmaller Materialsandmethods than1%(speciesrepresentedbyonesmallindividualorby seed lings only). T he level of inv asio n was characteriz ed in Theinvasivespecies eachplotbytherelativecoverofS.canadensisandaveraged over thet hre epl otsatsit eleve l.T he surveyswe rec onducted S. canadensis is a rhizomatous, patch-forming perennial oncepersiteinMay–June2012. herb oftheAster ac ea e.Ithasbeco meanexceptio nallysuc- In ord er to a ccurately d epict compositional changes dur- cessf ul inva sive species s ince its intro du ction to Europ e in ing t he cou rs e of succe ssion, w e calculated the follow ing the18th century andisn oww ides preadthroug ho uttheco n- vari able sforea ch site: tinent (Weber 1998). In addition to its prolific vegetative propag ation(M eyer& Sc hmid199 9), S.c anaden sisreleases (1) Native vascular plant species richness and diversity. chemicalsth atinhibi tth egrowt h,germ in ationandsu rvivalof Specie s richness was c alculated by avera ging the num- nativeplan ts(A bhilas ha, Quintan a,Vivanco, &Jo shi2008 ), ber of s pecies in the three plots pe r site. The Sha nnon and ch ange t he soil com position by divertin g n utrien ts and dive rsi tyindex of eac hsite wasc alcu lated bas edonthe min erals(Z hang ,W ang,Qian,& Li2 009). proportio nalco ver ofea chs pecie s.S.canad ensisw as not includedint hecal cul ation . (2) Naturalness.Allplantspecieswereassignedtooneofthe Locationandstudysites three naturalness groups according to Sanda, Popescu, Doltu,andDoni¸ta˘(1983):(i)speciesofnaturalandsemi- Our study area was located in the middle section of the naturalhabitats,hereaftercalled“grasslandspecies”,(ii) Târnava Mare Valley in Southern Transylvania, Romania species common to both natural and ruderal communi- (Fig. 1A). Climate is classified as moderate continental. ties, hereafter “generalist species”, and (iii) species of Meanannualrainfallisapproximately650mm,whilemean ruderal communities, hereafter “ruderal species”. The 338 A.Fenesietal./BasicandAppliedEcology16(2015)335–346 Fig.1. MapofthestudyareawiththeNatura2000sitedelineated.Thefourvillages,wherethesamplingtookplaceareshowninblack(A). Schematicmapshowingsamplingdesign(B). proportionofeachcategorywascalculatedforeachsite Pollinatorsampling andusedinsubsequentanalyses. (3) Functional guilds. The proportion of the main func- Wesampledpollinatorinsectsin22sites(outof36sites tional groups was taken into account by distinguishing withbotanicalsurveys).Weselectedthesesitestocoverthe (i)graminoidspeciesbelongingtothePoaceae,Cyper- whole range of successional age (1 to 20 years) and low to aceaeorJuncaceae,(ii)legumespeciesoftheFabaceae highdensitiesofS.canadensiscover.Thesiteswereatleast and(iii)forbs,i.e.herbaceousplantsbelongingtoother 250mfromeachother.Pollinatorsweresampledalongtwo families.Asthepresenceoftreeandshrubspeciesnever 100mtransectsperstudysite;eachtransectassignedwasat exceeded 1% average cover at site level, we excluded least15mfromthefieldedgeand15mapartfromeachother thosespeciesfromthecalculationofthisvariable. (Fig.1B).Wesurveyedthepollinatorstwice:firstduringthe A.Fenesietal./BasicandAppliedEcology16(2015)335–346 339 peakfloweringseasonofindigenousvegetation(21–25July For vegetation data, the dependent variables were plant 2012)andsecondduringthepeakfloweringofS.canaden- speciesrichness,diversity,naturalnessandfunctionalguilds, sis (19 –23 Augus t 2012) . Tr ansec t counts w er e p erformed all calc ulated fro m the da ta collected in the three 4 ×4m between 9 a.m. and 6 p.m. under favourable weather con- plots per site. We included old field age, the proportional ditions with little wind and few clouds at most. A pair of cover of S. canadensis and field area (in ha) as continu- observerswalkedalongeachtransectfor20minandrecorded ous fixed variables, land-use type (mown, grazed, without allinsectsactivelypollinating,flyingorperchingontheveg- management),anthropogenicfire(burned,unburned)asfixed etation. In the case of active pollination, the visited plant factors, and village (4 levels) as a random term. In addi- specieswasalsorecorded.Pollinatorswerehand-nettedand tion, the pairwise interaction between fixed variables and preservedin96%ethanolforlateridentificationatthespecies proportionalcoverofS.canadensiswasalsoincludedinthe level. models. Wedistinguishedfourpollinatorgroups:honeybee(Apis For pollinator variables (abundance and species richness mellifera L.), bumblebees (Bombus spp.), other wild bees of pollinator groups and the number of visited flowers (Apoidea: Collettidae, Melittidae, Halictidae, Megachili- except S. canadensis) land-use, fire, sampling period (July dae, Andrenidae, Apidae except Bombus spp. and honey vs. August) were included as fixed factors. The propor- bee), and hoverflies (Syrphidae). Bumblebees and other tional cover of S. canadensis, field area, old field age, wild bees were analysed separately because these two species richness of flowering plants, the number of flowers groups have different biological traits in terms of floral of native species and the number of S. canadensis flow- require ments , flying a bilities and socia lity (Gath ma nn & ers (colle cted from the ten 1×1m qu ad rates placed along Tscharntke 2002; Greenleaf et al. 2007; Michener 2007), the pollinator transects) were built in as fixed covariates. and therefore different responses to landscape and local The two-way interactions between S. canadensis cover and scale environmental conditions were expected. Although the other five explanatory variables (except flower vari- there are some semi-social species and/or genera among ables) were also included in the models. We included the collected bees (e.g. some Halictus spp.), we use ‘soli- study site nested within village as a random effect for tary bees’ in the current paper for wild bees other than abundance and number of visited flowers; we used only bumblebees. village as a random effect for species richness of polli- Asthepresenceofpollinatorsdependsonthepollenand nators because data from the two transects per sites were nectarsupply,wegatheredadditionalbotanicalinformation: pooledduetolownumberofavailablespeciesatthetransect floweringplantspeciesandthenumberofflowersatspecies level. levelwere reco rdedatt en1 ×1 mquad rat esplace de quidis- Ineachcase,westartedwithsaturatedmodels(Appendix tantly along the same two transects per site (Fig. 1B). We A) and removed non-significant variables using backward countedthenumberofheadsinthecaseofAsteraceaespecies stepwise selection, except those that were part of a sig- andsimpleumbelsforApiaceae,butrefertobothasflowers nificant interaction. We extracted the contrasts for retained forthesakeofsimplicity. significant terms as follows: the contrasts for covariates or Wecalculatedseveralvariablestodetectpotentialchanges interactions between covariate and factors, based on differ- inthepollinatorcommunitiesduringsuccession: encesofleastsquaremeans,wereextractedusingthelsmeans function of the lsmeans package (Lenth 2013). For land- (1) Abundanceofeachpollinatorgroup(bumblebees,soli- use types, Tukey contrasts were performed using the glht tary bees, honey bee, hoverflies) per transect was function of multcomp package (Hothorn, Bretz, & Westfall calculated as the number of individuals per group per 2008).Contrastsforinteractionsbetweencovariatescannot transect; be computed, hence we drew regression surface plots with (2) Species richness of bumblebees, solitary bees and hov- the dependent variable and the interacting covariates. Nor- erflies;speciesdatafromthetwotransectspersitewere mality of the residuals was tested using the Shapiro–Wilk pooledbecauseoftherelativelylowspeciesnumbers; test and homoscedasticity of the residuals verified by plot (3) NumberofvisitedflowersotherthanS.canadensiswas diagnosis. Abundance variables were log-transformed prior calculatedasthetotalnumberofflowersvisited(actively to analyses to reach normal residual distribution. We cal- pollinated)byeachpollinatorgrouppertransect. culated effect sizes (partial r) and their 95% confidence intervals (Nakagawa & Cuthill 2007) to present the bio- logically meaningful magnitude by which the significant Statisticalanalyses explanatory variables influenced plant and pollinator abun- danceanddiversity. All statistical analyses were carried out in the R statis- Before building LMMs, we tested all predictor variables tical environment version 3.0.1 (R Core Team 2013). We formulticollinearitybycalculatingthevarianceinflationfac- builtseparatelinearmixed-effectsmodels(LMMs)foreach tor (VIF) using vif function of the fmsb package in R. VIF vegetation and pollinator variable using the nlme package values were <2 in all cases, so no multicollinearity was (Pinheiroetal.2013). detected in any of the performed models (Rogerson 2001). 340 A.Fenesietal./BasicandAppliedEcology16(2015)335–346 Table1. TheeffectsofSolidagocanadensiscover,ageoftheoldfieldsandland-use(M–mown,W–withoutmanagement)onvegetation characteristicsaccordingtothelinearmixed-effectsmodels.Positiveeffectsareindicatedby“+”andnegativeeffectsby“–”.Effectsizes (Partialr)andtheir95%confidenceintervalarealsoshown.p-Valuesofsignificanteffectsareinbold.Significantinteractionsbetweenthe covariatesS.canadensiscoverandoldfieldagehavenocontrastsbutareplottedonFig.2. df F p Contrasts Effectsize 95%C.I. Richnessanddiversity Plantspeciesrichness S.c anaden siscover 1,29 4.43 0.044 − 0.09 −0.24;0.43 Oldfieldage 1,29 32.88 <0.001 + 0.72 0.38;1.06 SCC ×O FA 1, 29 3.09 0.088 Diversity S.canadensiscover 1,29 20.43 <0.001 − 0.24 −0.09;0.58 Oldfieldage 1,29 6.07 0.019 + 0.61 0.27;0.95 SCC ×O FA 1, 29 15.19 <0.001 −0.58 −0.92; −0.24 Naturalness Grasslandspecies S.canadensiscover 1,29 0.01 0.956 Oldfieldage 1,29 53.51 <0.001 + 0.80 0.46;1.14 SCC ×O FA 1, 29 5.30 0.028 −0.39 −0.73; −0.05 Generalistspecies S.canadensiscover 1,29 0.94 0.339 Ol dfieldage 1, 29 21.43 <0.001 − −0.72 −1.06;−0.38 SCC ×O FA 1, 29 10.69 0.002 0.51 0.17; 0.86 Ruderalspecies Oldfi eldage 1,31 45.09 <0.001 − −0.76 −1.11;−0.42 Functionalguilds Graminoids S.canadensiscover 1,29 1.56 0.221 Ol dfieldage 1, 29 9.53 0.004 + 0.21 −0.12;0.55 SCC ×O FA 1, 29 5.78 0.022 0.40 0.06; 0.74 Legumes S.canadensiscover 1,27 5.42 0.027 + 0.52 0.18;0.86 Oldfieldage 1,27 8.10 <0.008 + 0.68 0.34;1.02 SCC ×O FA 1, 27 8.39 0.007 −0.48 −0.82; −0.14 Land -u se 1, 27 4.47 0.021 W>M −0.47 −0.81; −0.13 Forbs S.canadensiscover 1,30 12.30 0.001 − −0.50 −0.84;−0.16 Ol dfieldage 1, 30 38.31 <0.001 − −0.74 −1.09; −0.40 In addition, all dependent variables were tested for spatial Theproportionofgrasslandspeciesincreased,whilethat autocorrelation(seeAppendixA). ofgeneralistandruderalspeciesdecreasedwitholdfieldage. TheeffectofS.canadensiscoverwasnon-significantinall three naturalness groups. However, S. canadensis did sig- Results nifica ntly interac t with o ld field ag e, as the propo rtio n of grassland species decreased, and that of generalist species Speciesrichness,diversityandcompositionof increasedintheolderfields,whiletheabundanceofgeneralist plantcommunity speciesdeclinedinyoungerfieldswithhigherS.canadensis cover(T able1,F ig .2C–E). Theoldfieldage,theproportionalcoverofS.canadensis, In the case of the plant functional guilds, the proportion andthe irin terac tion swe reimportant source so f variationfor of gr ami noids an d l egum es increase d, whil e th at of forbs allv egeta tionvariabl es(Ta ble1).Pla ntspeci es richnessa nd de creasedwith oldfi eldage( Table1).D espite their tem poral div ersity incre ased with field age . The density of S. can an- decline, fo rbs w ere the dom inant gro up throu ghou t succes- densis w as negativ ely a ssocia ted with the num be r o f plant sion.Hig hS.c anade nsis coverwas assoc iatedwithin creased species irre spectiveof oldfieldag eand wit hthediv ers ityof propo rtion of legumesin young ersi tes,andtha tofg raminoid olderfie lds(Table1 ,F ig.2 Aan dB ). speciesino ld erfields, wh ilethepr oport iono ffor bs decreased A.Fenesietal./BasicandAppliedEcology16(2015)335–346 341 Fig. 2. Plant species richness (A), plant diversity (B), the proportion of three plant naturalness groups (grassland species (C), generalist species(D)andruderalspecies(E)),andtheproportionofthreefunctionalguilds(graminoids(F),forbs(G)andlegumes(H))plottedagainst theincreasingproportionalcoverofSolidagocanadensisandoldfieldage. irrespectiveofoldfieldage(Fig.2F–H).Land-usetypepre- sampling time and species richness of actually flower- dicted only the proportion of legumes, which was smaller ing plants (Table 2). The abundance of bumblebees was inmownandgrazedsitesthaninsiteswithoutmanagement decreasedbyS.canadensiscoverinJuly(F=6.78,p=0.018, (Table1). Fig. 3A) and marginally significantly increased in August (F=3.03, p=0.099, Fig. 3B). Bumblebees were present in higher abundance in July than in August, and enhanced Abundanceandspeciesrichnessofpollinators by both native and S. canadensis flower abundance. The abu ndan ceofho ney bee sdecreased withS. canadensisc over Theabundanceofsolitarybeeswassignificantlydecreased and old field age. The abundance of hoverflies showed the by S. c anadensis co ver in Ju ly (F =6 .33, p=0.02 2), but no opp osite patt ern w ith h igher value s i n August t han in J uly. suc h effect was found in Aug ust ( Table 2 , Fig. 3A ). T he Hoverfly abunda nce w as enh anced by the field area an d by abun dance of s olitary be es was influen ced by land -use, theabund anceofS.c ana densisflow ers (Ta ble2 ). 342 A.Fenesietal./BasicandAppliedEcology16(2015)335–346 Table2. TheeffectsofS.canadensiscover,oldfieldage,land-use(G–grazed,W–withoutmanagement),samplingtime(July,August) andseveralvegetationparametersontheabundanceandspeciesrichnessofdifferentpollinatorgroupsaccordingtothelinearmixedeffect mod els.Flo werabunda ncereferst oa llfl owersexcep tth oseofS .canaden sis .Positive effectsar eindica tedby“+” ,n ega tiveeff ectsby “−“. Effectsizes(Partialr)andtheir95%confidenceintervalarealsoshown.p-Valuesofsignificanteffectsareinbold. df F p Contrasts Effectsize 95%C.I. Abundance Solitarybees S.can adensiscover 1,15 4.19 0.059 (−) La nd-use 2, 15 3.80 0.046 W>G −0.64 −1.08;−0.19 Sampling time 1, 63 38.87 <0.001 Jul > A ug −0.58 −1.03; −0.13 Floweringplantspeciesrichness 1,63 21.81 <0.001 + 0.48 0.03;0.93 SCC×ST 1, 63 7.29 0.009 Ju ly:− 0.32 −0.13; 0.77 Bumblebees S.canadensiscover 1,17 1.14 0.301 Sa mplingtime 1, 62 31.48 <0.001 Jul>Aug −0.60 −1.05;−0.15 Flowerab unda nce 1, 62 9.18 0.004 + 0.28 −0.17; 0.73 S.cana densisflow erabundance 1, 62 4.53 0.037 + 0.05 −0.40; 0.50 SC C×ST 1, 62 12.21 0.001 Ju l−;Aug(+) 0.41 −0.04; 0.86 Honeybee S.ca nadensiscover 1,16 13.06 0.002 − −0.68 −1.13;−0.23 Ol dfieldage 1, 16 8.11 0.012 − −0.58 −1.03; −0.13 Hoverflies Area 1,17 4.94 0.040 + 0.54 0.09;0.99 Samp lingtime 1, 64 7.25 0.009 Ju l<Aug −0.04 −0.49; 0.41 S.canadensisflowerabundance 1,64 35.03 <0.001 + 0.59 0.15;1.04 Speciesrichness Solitarybees Samp lingtime 1,38 19.02 <0.001 Jul>Aug −0.52 −0.97;−0.07 Floweringplantspeciesrichness 1,38 21.30 <0.001 + 0.60 0.15;1.05 Bumblebees Samplingtime 1,39 4.25 0.046 Jul>Aug −0.31 −0.76;0.14 Hoverflies Area 1,37 12.22 0.001 + 0.50 0.05;0.95 Flowerabundance 1,37 14.35 0.001 + 0.58 0.13;1.03 S.cana densisflowerabundance 1, 37 6.07 0.019 + 0.38 −0.07; 0.83 The species richness of solitary bees was higher in July native species by solitary bees, honey bees and hoverflies thaninAugustandincreasedwiththenumberofflowering (Table 3, Fig. 4). In addition, we found a significant inter- plantspecies.Similarly,morebumblebeespecieswerefound action between sampling time and S. canadensis cover: the in July than in August. The species richness of hoverflies number flowers (excluding S. canadensis) visited by bum- increased with field area and the flower abundance of both blebees was slightly decreased by S. canadensis in July nativespeciesandS.canadensis(Table2). (Fig. 4A) but marginally significantly increased in August (Fig.4B). Flowervisitation Ingeneral,thenumberofflowers(excludingS.canadensis) Discussion visited by bumblebees and solitary bees was higher in July than in A ugust (Table 3). The flow er v isitat ion of ind ige- EffectsofS.canadensisinvasionon nous plant species by solitary bees and honey bees was different-agedoldfieldplantcommunities positively influenced by the number of flowering species. The abundance of flowers (excluding S. canadensis) had a Although it was already known that plant communities positive effect on the number of bumblebee-visited flow- invadedbyS.canadensismightsuffershiftsindiversityand ers(Table3).Further,theincreasingcoverofS.canadensis composition (de Groot et al. 2007; Moron´ et al. 2009), we had a significant negative effect on the flower visitation of are the first to report reduced species richness and altered A.Fenesietal./BasicandAppliedEcology16(2015)335–346 343 Fig. 3. Abundance of four pollinator groups (honey bee, bum- Fig.4. NumberofvisitedflowersexceptSolidagocanadensisplot- bleb ees , solitary be es, hove rflies) plo tted aga inst the incre asing teda ga instthein cre asingp roportio nalcov erofS.ca nadensisin July proportio nalcove rofS. canadensis before itsflowe ring inJuly(A) (A) andin Au gust(B).N otethaty-ax ishas lo g- scale.Statis tic ally andduringits mass blo om inginAug ust(B). No tethaty-a xis has log- sign ifica nt andmar gina llysig nific antres ults aremarke dwith*and scal e.Stati stic ally significant an dmarg inall ysig nific antres ults are (*)andwit hreg ressionlin es;NS–no n-signi fica ntregre ssion s. marke dwith*and (*)andwith regr essionlines ;NS–non- signific ant regressions. Table3. TheeffectsofS.canadensiscover,oldfieldage,samplingtimeandseveralvegetationparametersonthenumberofflowers(excluding S.canadensis)visitedbydifferentpollinatorgroupsaccordingtolinearmixedeffectmodels.Flowerabundancereferstoallflowersexcept th oseofS.cana densis .Po sitiveeff ectsareind icatedb y“+”,neg ati veeffe ctsby “−“.E ffectsiz es(Parti alr)andth eir95% c onfi dencei nterval arealsoshown.P-valuesofsignificanteffectsareinbold. df F p Contrasts Effectsize 95%C.I. Numberofvisitedflowers(excludingS.canadensis) Solitarybees S.can adensiscover 1,17 11.47 0.004 − −0.47 −0.92,−0.02 Sa mplingtime 1, 64 81.71 <0.001 Ju l>Aug −0.72 −1.17; −0.27 Flowering plan tspeciesrichness 1, 64 13.76 <0.001 + 0.42 −0.03; 0.87 Bumblebees S.canadensiscover 1,17 1.08 0.313 Sa mplingtime 1, 63 36.64 <0.001 Jul>Aug −0.54 −0.99;−0.09 Flowerab unda nce 1, 63 8.37 0.005 + 0.29 −0.16; 0.74 SCC× ST 1, 63 6.54 0.013 Ju l(−);Aug:(+) 0.31 −0.14; 0.76 Honeybee S.ca nadensiscover 1,17 15.74 0.001 − −0.63 −1.08;−0.18 Fl oweringpla ntspe ciesrichness 1, 65 3.70 0.059 (+ ) 0.23 −0.22; 0.68 Hoverflies S.canadensiscover 1,17 4.31 0.053 (−) −0.45 −0.90;0.01 344 A.Fenesietal./BasicandAppliedEcology16(2015)335–346 naturalness and functional composition in the context of assemblageovertime(Ebeling,Klein,Schumacher,Weisser, secondarysuccession. &Tscharntke2008),itisunderstandablewhytheabundance The early years of vegetation succession are usually of solitary bees and bumblebees declined in July, the pre- characterizedbyawidearrayofdisturbance-tolerantruderal bloomingperiodoftheinvader(seealsodeGrootetal.2007; species and fast colonizing generalist species (Ruprecht Moron´etal.2009;Lenda,Skórka,&Moron´2010).However, 2006). Solidago canadensis displaced the native generalist inAugust,themyriadofflowersproducedbyS.canadensis species in young successional fields, which might slow had a positive effect on the abundance of bumblebees and downthesecondarysuccession,asgeneralistspecieswould ontheabundanceandspeciesrichnessofthemostlygener- lead the community to the next stage of succession by alisthoverflies.Therefore,thisinvasivespeciesmightserve outcompeting disturbance-tolerant ruderal species (Catford asanalternativefoodresourceforcertainpollinatorsduring etal.2012).Solidagocanadensiscanmonopolizeresources itsfloweringperiod,especiallyforbumblebeesthathavestill andhinderthecolonizationofotherspeciesduetoitsquick persistent colonies until late summer/early autumn, and for capacity for vegetative expansion and a dynamic foraging late-emerginghoverflies(seealsodeGrootetal.2007).From strategy(Barthaetal.2014).Onlynativelegumeswereable thispointofview,S.canadensismayindirectlyactasfacili- to increase their importance during the early years of the tatorbymaintaininglargerpollinatorpopulationsthroughout succession,probablyduetotheirnitrogen-fixingability. thefloweringseason(Bjerknes,Totland,Hegland,&Nielsen Inoldersuccessionalfields,thespeciesturnoverdecreases, 2007). and biotic filters can be particularly important (Pickett, S. canadensis is considered one of the most important Cadenasso,&Bartha2001).Ifamonopolistspeciesbecomes plants yielding unifloral honey sources in Central Europe dominant (fast-growing clonal species tending to eliminate (Farkas&Zajácz2007).Therefore,ahighlyinterestingfind- other species, sensu Falinˇska 1991), it may outcompete the ingisthattheincreasingcoverofS.canadensissignificantly subordinate species and may have a strong negative impact decreased the abundance of honey bees, even during mass onlocaldiversity(Barthaetal.2014).S.canadensisseems flowering.Althoughhoneybeesareimportantpollinatorsof to be a monopolist species as it negatively influenced the S. canadensis, they are likely to begin visiting goldenrod speciesrichnessofforbsandthediversityoftheolderfields.It flowers only when the abundance of other flowering plant outcompetedspeciesbelongingtothesamefunctionalgroup species has declined (Gross & Werner 1983). As the tradi- (perennialnon-nitrogen-fixingforbs),butseemedtofacilitate tionally managed landscape we studied is a diverse mosaic orhavenoeffectongraminoidspecies.Fargione,Brown,and ofcroplands,oldfieldsandmanagedgrasslands(pasturesor Tilman(2003)provedthatthestrongestinhibitoryeffectsof hayfields),pollinatorscaneasilyfindotherpollensourcesin residentplantsareonintroducedplantsofthesamefunctional thevicinityoftheinvadedoldfields. guild. Our results indicate that this relationship may work Although plant species richness and composition went inverselyaswell:theinvasivespeciesmightout-competethe through considerable alteration along the 20 years of suc- nativesbelongingtothesamefunctionalgroup. cession, neither solitary bees nor hoverflies showed any Most of the grassland specialists are expected to enter difference among old fields of different age. Therefore, the the community at mid-successional stages, because they bimodal species richness distribution of pollinators along are adapted to colonize the few available microsites under secondary succession presented by Steffan-Dewenter and a closed canopy (Bartha et al. 2014). However, the Tscharntke (2001) could not be replicated for bees or for increased presence of disturbance-tolerant generalists and hoverflies in our study. This is because neither the number the decreased proportion of the grassland specialists in the offloweringspecies,northenumberofflowersdependedon invaded old fields underlines that if this invasive species is theageofoldfields(datanotshown).Oneexceptionwasthe abundant,secondarysuccessionslowsdown,orevenhaltsat honey bee, which seemed to prefer plant species typical of a prolonged earlier successional state (Cramer et al. 2008) thebeginningofsuccessioninyoungeroldfields. insteadofrecoveringtowardssemi-naturalgrasslandvegeta- tion.Similarresultswerefoundinmid-successionaloldfields invad edbyS .gigan teaa swell ,h avingverystrong neg ative Indirecteffectsofinvasion:Flowervisitationof impactonsuccession(Barthaetal.2014). nativespecies Plantsandtheirpollinatorsaretightlyintertwinedcompo- EffectofS.canadensisontheabundanceand nentsofe colo gica lcommunit ies, therefo retheadvers eeffect speciesrichnessofpollinators of S. canadensis might have further indirect repercussions as well. Alien plant species can have a strong effect on the The decreased plant species richness and especially the pollination success of native species, often competing with declineofforbsledtofoodresourcesbecomingmorelimited thembycausingeitherreducedpollinatorvisitationratesor forpollinatorsintheoldfieldswestudied(seealsodeGroot increasedheterospecificpollinationofnativeflowers.How- et al. 2007). As a more diverse and flower rich plant com- ever,insomecases,bothatlocalandlandscapescales,these munity can harbour a more diverse and stable pollinator often mass-flowering invasives may facilitate native plant
Description: