herbivores associated with arundo donax in california growth with 25% stem mortality in infested...

138

1 MarineScienceInstitute,UniversityofCalifornia,SantaBarbara,CA93106,USA.

2 Department of Biology, Eastern Connecticut State University,Willi-mantic,CT,USA.

3 USDA-ARS European Biological Control Laboratory, Montpellier,France.

Correspondingauthor:T.L.Dudley<[email protected]>.*Theseauthorscontributedequallytothiswork.©CABInternational2008

Herbivores associated with Arundo donax in California

T.L. Dudley,1,* A.M. Lambert,1, 2,* A. Kirk3 and Y. Tamagawa1

Summary

TheOldWorldgrass,Arundo donaxL.(giantreed),isaseriousinvaderofCaliforniariparianareas,anditspurportedecosystemimpactsledtoitsconsiderationasatargetforbiologicalcontroldevelop-ment.However,theherbivorecomplexintheArundoadventiverangehasnotbeencharacterized,sothereislittleinformationregardinginsectsthatmayhinderbiologicalcontroleffortsbyinterferingwiththereleaseofnewagentsorthatcouldbepromotedasaugmentativebiologicalcontrolagentsiftheyhaveasubstantialimpactonthetargetweed.HerewereporttheresultsofsurveysinCalifornia,with emphasis on three presumably non-indigenous insects that inflict significant damage to the host. Oneisashoot-boringwasp,Tetramesa romana(Walker)(Hymenoptera:Eurytomidae),witharangelimitedtosouthernCaliforniaandthatdamagesshootsgenerallylessthan1cmindiameter.Ashootfly, Cryptonevrasp.,isalsoassociatedwithshootdamageandoftenmortalityofsecondarystems.Athirdherbivore,theaphidMelanaphis donacis (Passerini),iswidespreadinthesouthernandcentralpartsofthestatebuthaslessapparentimpacttothehost.T. romanaandCryptonevrasp.arecur-rentlycandidatesforbiologicalcontroldevelopmentandintroductionfromoverseaslocations.TheirestablishedpresenceinCaliforniasuggeststhateffortscouldberevisedtofocusondocumentationofhost ranges and impacts under field rather than in quarantine conditions, in anticipation of future re distributioninNorthAmerica.

Keywords: biologicalcontrol,Cryptonevra,Melanaphis donacis,Tetramesa romana,giantreed;herbivore.

IntroductionArundo donaxL.(giantreed)maybethemostdestruc-tive invader of California riparian areas, displacingnative vegetation, transpiring excessive groundwater,posing erosion and wildfire risks and providing poor wildlife habitat (Bell, 1994; Dudley, 2000; Kisner,2004). Few arthropods appear to be associated withArundoinCalifornia(HerreraandDudley,2003;Kirket al., 2003), and most are using it as opportunisticstructuralhabitatratherthanasafoodsource.Avarietyofherbivoresisfoundintheregionoforigin(fromtheMediterranean Basin and across southern Asia), andthelevelofherbivoreimpactsisconsideredtobemuch

greaterthaninCaliforniaorotherareaswhereArundoisinvasive(Kirket al.,2003).Classicalbiologicalcon-trolisbeingdevelopedusingseveralcandidateagentsfromEurasia(KirkandWidmer,2004),butimplemen-tationisnotanticipatedforatleastseveralyears.

A standard element of a biological control pro-gramme should involve documentation of herbivoresattackingthetargetweedinitsadventiverangetode-termineifnewagentsareneededorifeffectsofexist-ingherbivorescanbeenhanced,aswellastoevaluatethepotentialforinterferencewithintroductionofnewagents (Harris, 1975; Olckers and Hulley, 1995).Aspartof theprogramme tobuild theecological frame-workfor justifyingArundobiologicalcontrol,wearecharacterizingandcomparingArundo herbivores andassociatedplantcondition inCaliforniaandSouthernEurope.Earlyinvestigationsindicatedthatseveralnon-indigenous insects, including a shoot-boring wasp ofthefamilyEurytomidae,arepresentinCalifornia.Thewasp was subsequently identified as Tetramesa romana(Walker), a widespread Mediterranean species anda primary candidate in the Arundo biological controlprogramme. Members of this genus are highly host

139


specific (Al-Barrak et al., 2004), which makes themparticularlysuitableforbiologicalcontrol.

Ourinitialobjectiveistodeterminethegeographicextentof thisandothercurrentArundoherbivores inCalifornia and to quantify their efficacy against this hostweed.Thelargerobjectiveistoincreasetheimpactofexistingherbivoresthroughaugmentativemeasuresand/ortodistributethemmorewidelytoprovidemoreextensivehost suppression.Wearecurrentlyevaluat-ing their host specificity and appropriateness for mass rearingandredistribution.

Materials and methodsToexamineArundoherbivoredistributionsinCalifornia,weconductedmonthlyinsectsurveysofArundo standson the Santa Clara River and less frequent (once ortwiceduringstudyperiod)extensivesurveysofArundo-infestedareasthroughoutthesouthwesternUS(Table1).Linetransects100mlongwereestablishedwithinArundo vegetation, and samples were collected from0.5m2quadratsplacedat10-mintervalsalongtransectlines.AsofApril2007,a totalof994Arundo shootswithin all plots surveyedwere cut at the soil surfaceandbundledtogetherfortransportbacktothelabora-tory, where they were stored at 8°C until processed.Primaryormainshootsandsideshootsorsecondaryshoots were examined separately. Shoot lengths anddiameterswererecorded,andshootswerevisuallyex-aminedforherbivoresandevidenceofherbivoredam-age.Thesewerethensplitlengthwiseandexaminedforinternalfeeders.Afterdissection,shootsweredried(2daysat55°C)todeterminedryweightbiomass.Allre-coveredinsectsweresenttotheUSDAEuropeanBio-logicalControlLaboratory inMontpellier,Franceforidentification, and voucher specimens were deposited intheSantaBarbaraMuseumofNaturalHistory,Santa

Barbara,CA.Plotswerealsousedtodetermineshootdensityandbiomassperunitareaforanalysisofplantgrowth differences between native and introducedrangesandevaluationofimpactsofbiologicalcontrolagentsafterreleaseandestablishment.

Plant use by T. romana (oviposition, feeding andpupation sites) andpotential impactsof infestationonArundowereevaluated.Wefocusedonthisspeciesbe-causeit isoneof theprimaryagentsbeingtestedasapotential biological control (Kirk andWidmer, 2004).Shootlength,basaldiameterandbiomassofshootsandsideshootsinfestedwithT. romana,werecomparedwiththose of uninfested plants. Main shoot and side shootdatawereanalyzedseparatelyusingStudent’sttest.

Results and discussionNonativeinsectherbivoreswerefoundusingArundoas a significant food source, in contrast to prior sur-veys that shownumerous arthropodsusing this plantfor non-consumptive purposes (Herrera and Dudley,2003). Two non-native insects and another unidentified (butpotentiallynon-native)insectwererecoveredfromArundoshootsduringsampling.Anaphid,Melanaphis donacis(Passerini),wasfoundthroughoutthesamplingrange (except very northern California) with greatestabundanceincoastalArundopopulations.Aphidsfeedprimarilyon theapical shootsand lessmature,distalleaves,andalthoughreachinghighpopulationdensitiesinearlyspringinsomelocations,onlyminordamagewas observed on plants in one location. CoccinellidswereabundantonArundo shoots severalweeks afterpeakaphiddensitiesandmayhavebeenresponsibleforthedecreaseinaphidsbylateMay-aphidswereoftennotpresentduringsamplinginJulythroughDecember.Wedidnotquantifyaphiddensitiesduringsurveysforseveralreasons;therewasalackofanyvisibleaphid

Table 1. Samplinglocations(riversystems)bycounty,samplingintensity,andinsectsrecordedinCalifornia.

County # Samplingsites Insects Alameda 1 Melanaphis donacis (Passerini) Humboldt 2 None Imperial 2 Melanaphis donacis Inyo 2 None Kern 1 None LosAngeles 4 Melanaphis donacis,Tetramesa romana (Walker) Mendocino 2 None Monterey 5 Melanaphis donacis Orange 2 Melanaphis donacis,Tetramesa romana Riverside 3 Melanaphis donacis,Tetramesa romana SanBernardino 3 Tetramesa romana(onesite) SanDiego 5 Melanaphis donacis,Tetramesa romana SanLuisObispo 2 Melanaphis donacis SantaBarbara 7 Melanaphis donacis,Cryptonevrasp.,Tetramesa romana SantaClara 1 Melanaphis donacis Ventura 8 Melanaphis donacis,Cryptonevrasp.,Tetramesa romana Yolo 1 Melanaphis donacis

140

XIIInternationalSymposiumonBiologicalControlofWeeds

feedingeffect,aphidswereoftendislodgedfromstemsduringstemsamplingandcollection,aphidmovementreducedaccuratemeasurements,andthetimeandlabor-intensivenatureofcountingaphidswouldhavesubstan-tiallyreducedourabilitytoadequatelysampleArundopopulationsthroughoutthestate.

Dipteran larvaewere recovered fromover80%ofArundo shoots in one area on the floodplain terrace in theSantaClaraRiver,VenturaCounty,CA.Weinfre-quentlyfoundsimilardamagethroughoutthisriversys-temandthenearbyVenturaRiver.Severallarvaefeedtogether in theuppernodesonmainshootsofplants,andfeedingdamageresultedinatypical‘witchesbroom’shootgrowthwith25%stemmortalityininfestedshoots(Fig.1).The‘witchesbroom’shootgrowthwasevidentinmostshootscoveringa2-haradius.Wedidnotexam-inethesedamagedshootsforbacterialdisease,whichcancausethistypeofdeformity;however,weassumethatfeedingdamageontheprimaryshootpromotedsecond-aryshootproduction,assimilarlyoccursinstemswithwind-damaged primary shoots.The gregarious larvaewereassociatedwith‘hourglass’shootdamage(Fig.2)andmaybeaninquilinespeciesthatfeedsonmicrobesthat colonize damage from another fly species (A. Kirk, personal observations). Chloropid flies (Cryptonevra

sp.)intheMediterraneanregionprimarilyattackdevel-opingcanesuptoabout75cminlengthandcausesimi-lar‘hourglass’damagetowhatweobserved.Inaddition,itoranotherspeciesofChloropidaeattackssideshootsand/orleadingshootsontallercanesandareoftenfol-lowedbyseveralinquilinespecies-upto18inquilinespp.havebeenrecordedfromsouthernFrance(A.Kirk,unpublisheddata).Damagepreventsshootsfromelon-gatingduringgrowthandresultsinstuntedshootsanda‘witchesbroom’appearancefromincreasedsideshootproduction(Fig.1).Wearefocusingoursamplingef-fortsduringthespringgrowingperiodwhenherbivoresproducing the initial shoot damage may be present.Cryptonevra sp. is a candidate for potential biologi-calcontrolintroduction(TracyandDeloach,1998),sofurtherdocumentationofitsdistributionandimpactsinNorth America and verification that the species pres-enthereisthesameastheMediterraneantaxonbeingtested by US Department ofAgriculture-AgriculturalResearchService(USDA-ARS)areessentialforfutureagentselection.

The shoot-boring wasp, T. romana, was collectedfrom Arundo populations throughout southern Cali-fornia (Fig.3). Itsexitholesandgall-likeformationsproducedduringlarvalfeedingareevidentonprimary

Figure 1. Arundo donax L. shoot infested with Cryptonevra sp. in the Santa Clara River, CA (arrow). Note the atypical ‘witches broom’ appearance of the infested shoot relative to the other, uninfested shoots.

141


and secondary shoots.We have observed ovipositionon shoots both in greenhouse cultivation and at field sites(Fig.4).Shootinfestationwasvariable(range:0%to80%)withameanof23.1±4.4%(±SE)shootsin-festedacrossallsitessampled(n=994).However,inMarch2007,weobservedabout2.5haofanArundostandneartheSantaClaraRiver(VenturaCo.)withca.

99%ofsideshootsinfestedandkilled.Arecentsurveyatthissamelocationsuggeststhatinfestationlevelsofcurrent-yearshootsarestillhigh.

Waspdensitiesarehighestonsmallerdiametermainshoots(primarilynewshoots)andsideshootsoflarger,matureshoots(Fig.5).Waspdensitiescanexceed35individualson amain shoot and six individualson a

Figure 2. Arundo donax L. shoot damage caused by Cryptonevra sp. feeding. The arrow points to the characteristic ‘hourglass’ damage.

Figure 3. Distribution of Tetramesa romana (Walker) (Hymenoptera: Eurytomidae) in California as of April 2007.

142


sideshoot,butaverage4.9±3.6individualspermainshootand3.4±1.9persideshoot.Infestedmainshootswere shorter and smaller in diameter thanuninfestedshoots(shootheight,t=3.02,df=315,p=0.003;shootdiameter, t=3.93,df=315,p<0.001),butbiomasswas not significantly different between infested and un-infestedshoots.Infestedsideshootswerethinnerthanherbivorefreeshoots(t=2.0,df=197,p=0.05).Side

shoot height and biomass were not significantly af-fectedbywaspinfestation.InFrance,T. romanainfestsshoots over a broader diameter range (A. Kirk, per-sonalobservation);Arundoshootsarealso,onaverage,thinnerandshorterthaninNorthAmericanpopulations(A.Lambert,unpublisheddata).

Biomass reduction in infested shoots could not beinferred from thesedata for two reasons: (1) extreme

Figure 4. Tetramesa romana (Walker) (Hymenoptera: Eurytomidae) ovipositing in an Arundo side shoot. Bar represents 2mm.

Figure 5. Distribution of emergence holes of Tetramesa romana (Walker) over the stem density range (mm) of Arun-do donax L. main and side shoots.

143


variability in measures of morphological character(shoot length, diameter and biomass) even with sub-stantial replication and (2) wasps colonize primarilynew shoots and secondary shoots, which have lowerbiomass than mature and main shoots, respectively(Spencer et al., 2006). Therefore, wasp impact maybe less noticeable on young shoots relative to that ofan herbivore attacking the larger mature shoots. ThefrequencyofdamagebutlackofsubstantialimpactofT. romanatoArundopopulationsinoursurveysdoesnotnecessarily reflect an absence of impact for this species. Itisanticipatedthatthisherbivore,inassociationwithotheragentssuchas therhizome-feedingscale insect,Rhizaspidiotus donacis (Leonardi), may inflict greater damagetohostplants(A.Kirk,personalobservations).

Further surveys are necessary to accurately deter-minethefullextentandlimitsoftheT. romanadistri-butionanditspotentialtoinfestagreaterproportionofshoots. For example, as our sampling size increases,the relationship between wasp density and main andsideshootmortalitybecomesstronger.Wearecurrentlyevaluatingabove-andbelowgroundbiomassrelation-ships of infested and uninfested plants in a commongardenexperimenttoelucidatetheeffectofherbivoryonresourceallocation.Experimentsarealsobeingcon-ductedbytheUSDA-ARStodeterminetheimpactofT. romanaonwaterusebyArundo(J.Goolsby,USDA,Weslaco,TX,personalcommunication).

GeneticcomparisonsareunderwayincollaborationwithD.KazmerandW.Jones,USDA-ARS,Sidney,MO,andEuropeanBiologicalControlLaboratory(EBCL),usingnewlyeclosedfemalescollectedthroughoutCali-forniaandmatchingwithwaspscollectedintheMedi-terranean region todetermine theoriginofCaliforniaT. romana.Itisunclearatthistimewhetherthiswaspisarecent introductioninCaliforniaandisstill in itsestablishmentphase,inwhichcaseitspotential(future)rangemaybemuchgreater.Alternatively,ifT. romanahasbeenestablished fora longperiod, thendispersalmayhaveachievedthefullextentofitspotentialrange,and the likelihood for redistribution is poorer.WaspshavebeenfoundthroughoutthewatershedsofVenturaCounty,whereasstands inothercounties,particularlyat the northern and southern extents of its apparentrange,areoftencolonizedmoreheterogeneouslywithuninfectedstreams inmoderatelyclose (e.g.<10km)proximity tocolonizedstands.Thismay indicate thatthecolonizationprocessis,asyet,incomplete.Patchydistributionscouldalso represent reducedphysiologi-calcompatibilitywithclimateregimesatthelimitsofthe T. romana distribution. Such a distribution in thewarm- temperate coastal region of California is notdissimilartootherorganismsintroducedforbiologicalcontrolofotherCaliforniapests,e.g.parasitoidAphytisspp. against the citrus red scale (Rosen and DeBach,1978, 1979). Furthermore, it is not yet clear whethertheremayhavebeentheco-introductionofT. romanaparasitoids(A.Kirk,unpublisheddata)andsubsequent

population regulation, which could also explain thevariationinattackrateobservedoverT. romana’sdis-tributioninCalifornia.

The unintentional introduction of T. romana andotherapparentspecialistherbivorestoNorthAmericaprovides the opportunity to conduct host-specificity trials and other experimental studies in the field, which will provide instructive ecological information lesseasily obtained under quarantine conditions. Plantgrowth and insect behavior are often constrained oratypicalinhighlycontrolledenvironments(Blosseyet al., 1994a,b), so results in open-field settings are ex-pectedtobemorerepresentativeofnaturalresponses.We suggest that an Arundo biological control pro-gramme based on development of ‘California’ T. ro-manaandCryptonevrasp.asaugmentativebiologicalcontrolagentsforlocalpopulationenhancementand/orre-distributioninNorthAmericashouldbecarriedoutinparallelwithforeignimportationandfurtherquaran-tinetesting.Host-rangetestingshouldbecontinuedtovalidate host specificity prior to agent movement, but conducting thisworkwith insectsalreadyestablishedinNorthAmericawouldenhancetheecologicalvalid-ity of the tests and conserve financial resources. This surveyofCaliforniaArundoherbivoresalsovalidatestheadviceofHarris(1975)andothersthatbiologicalcontrol be developed as a logical progression, withevaluationofweedecologyandassociatedherbivoresintheinvasiverangebeforeimportingneworganisms.

AcknowledgementsWe thank J. ten Brinke, K. Kennedy,V. Frankel andT. Lemein for field assistance and R. Sforza and two anonymousreviewersforeditorialassistance.Weap-preciatetheinformationandstudysiteaccessprovidedbyTheNatureConservancySantaClaraRiverProject,particularlyE.J.RemsonandC.Cory.Financialassis-tancewasprovided,inpart,bytheSantaClaraRiverTrusteeCouncil/US-FWSgrantno.81440-5-G021andthe University of California Integrated Pest Manage-ment Program grant no. 2006-34439-17024 (USDA-CSREESno.2006-34439-17024).

ReferencesAl-Barrak M., Loxdale, H.D., Brooks, C.P., Dawah, H.A.,

Biron,D.G.andAlsagair,O.(2004)Molecularevidenceusing enzyme and RAPD markers for sympatric evolu-tioninBritishspeciesofTetramesa(Hymenoptera:Eury-tomidae).Biological Journal of the Linnaean Society83,509–525.

Bell, G. (1994) Biology and growth habits of giant reed(Arundodonax).In:Jackson,N.E.,Frandsen,P.,Douthit,S. (eds) Arundo donax Workshop Proceedings, Ontario,CA.pp.1–6.

Blossey, B., Schroeder, D., Hight, S.D. and Malecki, R.A.(1994a) Host specificity and environmental impact of the weevil Hylobius transversovittatus, a biological control

144


agentofpurpleloosestrife(Lythrum salicaria).Weed Sci-ence42,128–133.

Blossey, B., Schroeder, D., Hight, S.D. and Malecki, R.A.(1994b) Host specificity and environmental impact of two leaf beetles (Galerucella calmariensis and G. pusilla)forbiologicalcontrolofpurpleloosestrife(Lythrum sali-caria).Weed Science42,134–140.

Dudley, T.L. (2000) Arundo donax L., In: Bossard, C.C.,Randall,J.M.,Hoshovsky,M.C.(eds)Invasive Plants of California’s Wildlands. University of California Press,Berkeley,CA,pp.53–58.

Harris,P. (1975)General approach tobiological controlofweedsinCanada.Phytoprotection56,135–141.

Herrera,A.and.Dudley,T.L.(2003)InvertebratecommunityreductioninresponsetoArundo donaxinvasionatSonomaCreek.Biological Invasions5,167–177.

Kisner,D.A.(2004)The effect of giant reed (Arundo donax) in riparian areas of Camp Pendleton Marine Corps Base, California.MScThesis,SanDiegoStateUniversity,CA.

Kirk,A., Widmer, T., Campobasso, G., Carruthers, R. andDudley, T. (2003) The potential contribution of naturalenemiesfromMediterraneanEuropetothemanagementof the invasiveweed Arundo donax (Graminae;Arundi-nae). Proceedings of the California Invasive Plant Coun-cil Symposium7,62–68.

Kirk,A.A.andWidmer,T.L.(2004)Biological control of Gi-ant Reed (Arundo donax) an invasive plant species in the USA. USDA-ARS Petition toAPHIS forTechnicalAd-visory Group. European Biological Control Laboratory,Montpellier,France.

Olckers T., and Hulley, P.E. (1995) Importance of prein-troductionsurveysinthebiologicalcontrolofSolanumweedsinSouthAfrica.Agriculture, Ecosystems and En-vironment52,179–185.

Rosen,D.andDeBach,P.(1978)Diaspididae.In:Clausen,C.P.(ed.)Introduced Parasites and Predators of Arthro-pod Pests and Weeds: a World Review. United StatesDepartment ofAgriculture,Agricultural Handbook 480,pp.78–129.

Rosen, D. and DeBach P. (1979) Species of Aphytis of the World.DrW.Junk,TheHague,TheNetherlands.801p.

Spencer,D.F.,Liow,P.,Chan,W.,Ksander,G.G.andGetsin-ger,K.D.(2006)EstimatingArundo donaxshootbiomass.Aquatic Botany84,272–276.

Tracy J.L. and Deloach. J. (1998) Suitability of classicalbiological control of giant reed (Arundo donax) in theUnited States. In: Bell, C.E. (ed.) Proceedings of the Arundo and Saltcedar Workshop. University of Cali-fornia Cooperative Extension Service, Holtville, CA,pp.73–109.

herbivores associated with arundo donax in california growth with 25% stem mortality in infested...

Documents