Uncovering the spatial pattern of invasion of the honeybee pest small hive beetle, Aethina tumida, in Italy

(1)

REVISTA

BRASILEIRA

DE

Entomologia

AJournalonInsectDiversityandEvolution

w w w . r b e n t o m o l o g i a . c o m

Biological

Control

and

Crop

Protection

Uncovering

the

spatial

pattern

of

invasion

of

the

honeybee

pest

small

hive

beetle,

Aethina

tumida,

in

Italy

Alessandro

Cini

a,∗

_,

_Ugo

_Santosuosso

b

_,

_Alessio

_Papini

a

a_Università_degli_Studi_di_Firenze,_Dipartimento_di_Biologia,_via_Madonna_del_Piano_6,_Florence,_Italy

b_Università_degli_Studi_di_Firenze,_Dipartimento_di_Medicina_clinica_e_{sperimentale,}_Largo_Brambilla_3,_Florence,_Italy

a

r

t

i

c

l

e

i

n

f

o

Articlehistory:

Received19September2018

Accepted27November2018

Availableonlinexxx

AssociateEditor:EduardoAlmeida

Keywords:

Apismellifera

DBSCAN

Invasivealienspecies

Pestspecies

Spatialdynamics

Smallhivebeetle

Notifiabledisease

a

b

s

t

r

a

c

t

Thefasttrackingofinvasionspatialpatternsofalienspeciesiscrucialfortheimplementationof pre-ventiveandmanagementstrategiesofthosespecies.Recently,ahoneybeepest,thesmallhivebeetle Aethinatumida(hereafterSHB),hasbeenreportedinItaly,whereitcolonizedmorethan50apiariesinan areaofabout300km2_._SHB_is_a_nest_parasite_and_scavenger_of_honeybee_colonies_native_of_Sub-Saharian

Africa.Likelybeinghelpedbytheglobalizationofapiculture,SHBunderwentseveralinvasionsinthelast twentyyears,causinglocallyrelevanteconomicimpact.Whilemanyfeaturesofitsbiologyhavebeen addressed,animportantknowledgegapconcernsthespatialinvasiondynamicsininvadedareas.Inthis paperwecoupledtwospatialanalysistechniques(geographicprofilingandadensity-basedspatial clus-teringalgorithm)touncoverthepossibleinvasionpatternofSHBinItaly.Weidentifiedtheporttown ofGioiaTauroasthemostlikelypointfromwhichSHBmayhavespreadandsuggestedthepossible successiveaxesofdiffusion.TheseputativediffusionpathssuggestthattheSHBspreadinsouthItaly mighthavebeenduetoamixofnaturaldispersalbetweencloseapiariesandlongerdistancemovement throughfaster,likelyhuman-mediated,communicationroutes.

).

Introduction

ThefasttrackingofthespatialpatternofInvasiveAlienSpecies (hereafterIAS) invasion innewareas rightaftertheirarrival is ofoverwhelmingimportanceinordertoadoptfruitfulpreventive strategiestoreducethespreadofIASintheinvadedareas, pre-vent possiblerecurrentintroductions and preparemanagement protocols(Hulmeetal.,2008).Aneffectiveandfastreactiontothe invasionisfacilitatedbytheidentificationofthepossiblespreading centre(s)andtheearlypost-invasiondiffusionpathways.

SHBisasub-Sahariancoleopteran,exotictoEurope,whichhas beendetectedinSouthItalysinceabout5years(Mutinellietal.,

2014;Palmerietal.,2015).Morerecently,newinvasionsofthis

pestwererecorded inBrazilwhere itappears tobestillatthe beginningofitsspread(Toufailiaetal.,2017).SHBisaparasiteand scavengerofhoneybeecolonies,andbothSHBlarvaeandadults feedonhoney,pollen,beelarvaeandalsoondeadadultbees(Ellis

etal.,2002;Neumannetal.,2016).Themainimpactonhoneybee

isrelatedtolarvaefeedingoncombs,whichsometimesmayresult

∗ Correspondingauthor.

E-mail:cini.ales@gmail.com(A.Cini).

ina complete collapseofthenest (Schmolke,1974).Additional damagesmayalsobeprovokedbySHBattackingstockedhoneybee products,suchascombsinstoragecontaininghoney(Elzenetal., 1999)and,potentially,spreadingofbeeviruses(Eyeretal.,2009). WhileSHBdoesnotrepresentarelevantpestspeciesinitsnative range,itcancausefrommoderatetobigimpactsininvadedareas

(NeumannandElzen,2004;Spiewoketal.,2007).Finally,SHBhas

beensuggestedtobeabletoexploitcoloniesofotherApidae,such asbumblebeesandstinglessbees(SpiewokandNeumann,2006;

Hoffmannetal.,2008).

SHBmovedrepeatedlyacrossinternationalborders,leadingto aworldwidediasporathatbroughtittocolonizeseveralstatesin theU.S.,Australia,Korea,SouthAmericaandalsoinvadingCanada, evenifinthelattercountryfastdetection,movementrestrictions andpossiblyalsoclimaticconditionsallowedtoavoid

establish-ment(Lounsberryetal.,2010;AlToufailiaetal.,2017;Leeetal.,

2017).

Europe (where SHB is a honeybeenotifiable pest, Commis-sionDecision2004/216/EC)faced theriskofaninvasionforthe firstdocumentedtimein2004,whenSHBwasdetectedin Portu-galduringqueenbeeimportcontrolinthelaboratory(Murilhas, 2004)andrapidlydestroyed.In September2014SHBwasagain reportedinEurope,havingbeenspottedinthreehoneybeenucleus

https://doi.org/10.1016/j.rbe.2018.11.005

(2)

coloniesin SouthItaly, nearthecoastlineof regionofCalabria, whereithasprobablyaccidentallyintroducedfromAfrica(Granato et al.,2017).SHB wasdetectedin 58 apiaries and one natural colony,spanningoveranareaofabout316km2 ₍_Mutinelli_et_al.,

2014).Newinfestedapiarieshavebeendetectedin2015,2016and

2017(http://www.izsvenezie.com/aethina-tumida-in-italy/), and

theinvasionisthusnotresolved.

Despiteresearcheffortsstronglyincreasedtheknowledgeon SHBduringthelastfewyears,manyknowledgegapsstillremain (Neumanetal.,2016).Amongthem,oneofthetopprioritiesisto understandtheinvasiondynamics,inparticularlythespatial pat-ternofdiffusionafterthearrivalinnewareas(NeumannandElzen,

2004).

Here,weproposeatwo-stepspatialanalysisinorderto iden-tifythepossiblepost-invasionspatialdynamicsofIASstartingfrom presence-onlyreports,usingtherecentinvasionofsmallhive bee-tleAethinatumidaMurray,1867(Coleoptera:Nitidulidae,hereafter SHB)asamodel.Thedetectionofspreadingcentres,inadditionto facilitatethepreventionofrecurrentIASintroduction,mayalso allowtheidentificationofthepossibleidiosyncraticfeaturesthat boostedtheinvasion(e.g.ecologicalrequirements).Startingfrom thespreadingcentre(s),diffusionmaythentakeplaceinseveral spatialways,beinginfluencedbytheecologicalneedsandlife his-torytraitsoftheIASaswellastheenvironmentalheterogeneityand possibleman-assisteddisplacement(Wilsonetal.,2009).Our anal-ysisaimsto(a)identifythemostlikelyspreadingcentre,(b)infer thelikelyearlypost-invasiondiffusionpathwaysofSHBinvasion inItalyand(c)provideamodelforotherinvasionsofA.tumidain othergeographicalareas.IASspreadingcentre(s)arethoseinvaded areas which more likely allowed the first successful establish-mentoftheIASandthankstowhichtheIASthenspreadtoother areas.

InordertoinferthepossibleinvasiondynamicsofSHBinItaly frompresencereports, wecoupledGeographicprofiling– here-afterGP–(Rossmo,2000),amodellingtechniquewidelyusedto inferspreadingcentreofinvasivespecies,togetherwitha density-basedspatial clustering algorithm (DBSCAN, Esteret al., 1996). Geographicprofiling(GP)isananalytictoolfortheidentification ofthegeometricaloriginoflinkedevents,whichhasbeenrecently appliedtoarangeofecologicalissues,amongwhichthe identifica-tionofthesourcepopulation(s)ofinvasivespeciesusingtheknown positionsoftheircurrentpopulations(Stevensonetal.,2012;Cini

etal.,2014;Faulkneretal.,2017;Papinietal.,2013,2017a,b),while

severalenhancementsofthemethodareunderstudyinorderto refinethereliabilityoftheanalysis(PapiniandSantosuosso,2017;

SantosuossoandPapini,2018).

TheDBSCANisaclusteringalgorithmdefinedasdensitybasedas itproducesclustersofpoints(onaplane)thataregroupedtogether accordingtotheconceptthateachpointoftheclustermustbeclose toatleastagivennumberpofotherpointsoftheclusterswithina givenradiusr(XuandWunsch,2005).DBSCANhasseveralfeatures thatseemadvantageousinbiological invasions,since itisquite robusttooutliers(Smithetal.,2002),minimallyaffectedbypoints orderinganditrequireslittleaprioriinformationaboutthedataset parametersvalues,suchasnumberandshapeoftheclusters(Smiti

andEloudi,2013),acasewhichistypicalofbiologicalinvasionsin

theirfirstyearsafterIASintroduction.

OuraimwastodrawthelikelyspatialpatternofSHBinvasionin Italy,thussettingthestageforfurtherimplementationofspecific monitoringprotocolsinthisarea.

Materialandmethods Datacollection

SHB presence data were retrieved from Mutinelli et al.

(2014)which providesthedata oftheNational Reference

Cen-treforApiculture(availableat

http://www.izsvenezie.it/aethina-tumida-in-italia/), collected during theapiary surveillance

pro-grammeafterthediscoveryofSHBinItaly.Briefly,asubsample oftheapiarypresentinaradiusof20and100kmfromthefirst reportwereinspected,andapiarywereconsideredinfestedifat leastonehiveshowedthepresenceofSHB,inanydevelopmental stage.Asgeoreferentiationofinfestedapiariesisnotfreely avail-ableduetoprivacyreasons,wegeoreferencedapiarysitedirectly fromthemap.Usingthisprocedureestimatedspatialaccuracywas 25metres,whichwasthusnotexpectedtoaffecttheresolution atwhichourspatialanalysiswasperformed(tenstohundredsof kilometres).Aswefocusontheveryearlypost-invasiondynamics inthearrivalzone(Calabriaregion)andweaimedat understand-ingpossibleaxesofdiffusionusingonlyearlypost-invasionsdata, wedidnottakeintoaccountthepresencereportfromSicily(one infected colonies in a 56-hives migratory apiary, likely due to humanhivetranslocation, Mutinelliet al.,2014), northe pres-encedatareferringto2015or2016surveys,anapproachalready used(Cinietal.,2014).Moreover,2015and2016datahavebeen collectedaftersanitarymeasures,suchashivedestruction,were performed,sothattheymightnotproperlyrepresenttheinitial spatialdynamicsofSHB.DataweremappedontheSouthItalymap (obtainedbytheopenaccesswebsitewww.openstreetmap.org). Identificationofthemostlikelyspreadingcentre:geographic profiling

GPusescoordinatesonamapoflinkedevents(e.g.locations whereanIAShasbeenreported)togenerateaprobabilitysurface, whichwillbesuperimposedontheoriginalmaptoproducethe so-calledgeoprofile.Suchgeoprofiles,basedonlyonpresencedata,do notprovidetheexactoriginoftheevents,butproducesdecreasing probabilitydensity,thusprioritizinggeographicalareas(Rossmo, 2000)wheremonitoringprotocolsshouldbeimplementedand tar-getedresearchescarriedout.WeusedthemodelforGeoprofiling analysisdescribedbyRossmo(2000)andmodified accordingto

Papinietal.(2013).Foreachpixelwithcoordinates(i,j)ofthe

tar-getarea,thescorefunction(p)iscalculatedasfollows(redrawn

fromStevensonetal.,2012):

Pij=k∗ c

n=1

⎡

⎢

⎣

(xi−xn)2+(yi−yn)2

f + (1−)∗B (f−g)

2B−

(xi−xn)2+(yi−yn)2

g

⎤

⎥

⎦

where: =

_1, _if_distance _> _B 0, otherwise

ForapointPofcoordinates(i,j)theformulasumsthe prob-abilityacrossallthelocationswheretheinvadingorganismwas found.Thismodelisbasedontwocomponents:adistancedecay function(inwhichtheprobabilityofapresencereportdecreases asdistance fromtheinvasion site increases)and a bufferzone ofradiusB,withinwhichthelikelihoodincreaseswhendistance

(3)

Fig.1. GraphicalexplanationoftheDBscanalgorithm.Itemswithinagivendistance

(radius)r(indicatedascircles)fromatleastagivennumberp(threeintheexample)

ofotherpointsareassignedtothesamecluster(i.e.thebluecirclesallbelongtothe

sameclusterbecauseeachbluecircleiswithinagivenradiusrofthreeotherblue

circles).Thetwogreenitemscannotbeassignedtothebluecluster,sincetheyare

withindistancerfromfewerthanthreepointsofthebluecluster.Thereditemis

positionedatadistancethatis≥todistanceAandthereforeitdoesnotbelongto

thebluecluster.

increases.Thedistance-decayfunctioncanbeexplainedby disper-salcostswhilethebufferzonemaybeduetohabitatunsuitabilityof invasionsites.Stevensonetal.(2012)suggestedthatitis conceiv-abletousegeneral,taxonorhabitatspecificvaluesforthemodel parametersincaseswherenodataonthetargetspeciesexist.In ourcase,wechosemodelparametersaccordingtopreviously pub-lishedanalysesoninvertebrateinvasivespecies,suchasStevenson

etal.(2012)andCinietal.(2014).Sincewedonothaveany

infor-mationaboutthepossiblerealvaluesofthecrucialbufferzone radius(B)forthis orrelatedspecies,and sinceitmightdepend onseveralfactors(suchasdispersal ability,habitat heterogene-ity,etc.),weselectedthevaluesobtainedfortheonlyavailable insectspeciesfromStevensonetal.(2012)(Bvalues:0.38–0.42), which,scaledtoourmapwasaboutB=10(itisnot straightfor-wardtoobtainthisdata,sinceStevensonet al.didnotindicate howlargeisBinkmortheresolutionoftheirmap).Inorderto evaluatetherobustnessofthemodelwhenBvaries,weranthe analysisusingBvaluesrangingfromB=5toB=20 (correspond-ingtoa rangeof about0.125–5km),which spanned acrossthe valuesfoundforotherterrestrialflyinginsects(Stevensonetal., 2012,adaptedtothedifferentmapscale).Forthispurposeweused thepythonscriptwrittenbytheauthors:Geoprof205csv.pyon Bitbucket(https://bitbucket.org/ugosnt/alandugo/).

Identificationofthepossiblediffusionpathways:clusteranalysis anddensitymaps

TheDBSCANalgorithmwasusedtoproduceclustersofpoints correspondingto observations on theused map. The points of a given cluster must beclose to at least a given number p of other points within a given radius r (Xu and Wunsch, 2005) (Fig.1).Inordertoseparatedifferentclustersofobservationswe usedthefollowingalgorithmsimplemented inpython: Kmeans (number of clusters assessed with Silhouette) (Python script Kmeanssil002.py)withthefollowingspaceVoronoitessellation asproposedindetailbySantosuosso andPapini(2016)andthe Density-basedspatialclustering(DBSCAN),asdefinedbyEsteretal.

(1996). DBSCANwas implementedin Python as dbscan1.0.0.py

andafterpreliminarytrialsweusedr=6(pixels),corresponding toabout100mand p=5 (numberofpoints tobefoundwithin theradius)asparameters.Finally,inordertoinferpossiblespatial invasiondynamics,wecoupledGPandDBSCANresultsbydrawing theputativeinvasiondirectionsconsideringasastartingpointthe

centreoftheidentified95%spreadingareasidentifiedbytheGP andthetipofeachdirectionarrowsasthecentreofmassofeach clusteridentifiedbyDBSCANmethod.

Results

TheGPanalysisidentifiedanareaofabout27km2_on_the_coast,

north/northeast ofthe townof GioiaTauroas the95%(in red) mostlikelyspreadingcentreofSHBinSouthItaly(Fig.2a)onthe basisoftheinfectedapiaries.Overlappingtheareaidentifiedby GPasthespreadingcentre,usingasinputsourcesinfectedapiary onlyorbothinfectedandnoninfectedapiary(nullmodel,Fig.2b), showedthatthesecondareaof95%probability(inred)oforigin, waslocatedinamorenorthernpositionwithrespecttothefirst (onlyinfectedapiaries),suggestingthattheareaidentifiedbyGP wasnotsimplyreflectingapiarydensity.Theintersectionbetween thetwoareasisshowninFig.2canditcorrespondstolessthan 4%ofthetotalnumberofredpixels,whichcanbeconsidereda significantdifference.GPresultsarerobust,asvaryingtheBvalue, theresultaboutthelocalizationoftheareaofhighestprobability ofspreadingcentredidnotvarymorethanfewpixels(datanot shown).

DBSCANclusterizationassigned the57 casesofinfected api-aries tothree clustersand identified17 apiariesasoutliers(i.e. notnecessarilybelongingtoanycluster).Twoofthethree clus-tersfallindeedattheborderofthespreadingcentreidentifiedby theGPapproach(compareFig.2withFig.3).Thethirdone(as pur-pleovalsinFig.3)isabout10kmsouthwardfromGioiaTauroand about15fromtheputativespreadingcentre,asidentifiedbyGP. Consideringthiscoupledanalysis,itseemsverylikelythatamain axisofSHBspread,pointingsouthward,canbeidentified(Fig.3a andb).

Discussion

Theidentifiedareaofabout27km2 _around_the_Port_of_Gioia

TauroasthemostlikelyspreadingcentreofSHBinItalysupports thestrictrelationshipofthisinvasionwiththepresenceofoneof themostimportantportsinSouthItaly.

Theposition ofthethree main clustersof SHB identifiedby the DBSCAN analysis showed that while two are close to the supposed spreading centre (tentatively identifiedwith GP), the thirdoneispositioned15kmsouthward.Thiscoupledapproach thussuggests(a)thataport,GioiaTauro,mightbetheentrance, and (b) that SHB might have had three main diffusion direc-tions,withdifferenteasinessofpenetrationalongdifferentspatial gradients.

Theimportanceofmarineportsaslikely pointsofIASentry hasbeenwidelyrecognized(Kelleretal.,2011;McCulloughetal., 2006).Travelers’baggage,cargoandtradeitemsmayeasily intro-duceexoticinvertebratesandplantweedsfromallovertheworld. Severalalieninsectsareindeedsuspectedofhavingspreadinto newareasdiffusingfrommarineporttown.Itisthecaseof wood-boringbeetles(Haack,2001)andofthesmallfruitpestDrosophila suzukii(Cinietal.,2014).Indeed,thepossibilitythatSHBmight alsoenterthroughportshasbeenproposedfortheUSAEast-coast andAustraliainvasions(Hood,2004)andwasalreadysuggested forItaly(Mutinellietal.,2014).OurGPapproachstronglysupports thissuggestion.

Thefindingof twovery closeclustersand onemoredistant southern cluster suggeststhat SHB Italian invasion might have proceededthankstoamixofleading-edgedispersal(i.e.diffusive spread from the edge of the invasion range, likely driven by naturaldispersal)andjumpdispersal(long-distancedispersalover substantial distances likely due to accidental human mediated

(4)

Fig.2.Identificationofthepossiblediffusioncentre.(a)The95%mostlikelyspreadingcentreisinredandcorrespondstotheresultsofthegeographicprofiling(GP)applied

onlytotheinfectedapiaries;(b)anullmodel,withGPresults(theredareas)basedonalltheapiariespresentinthearea,bothinfectedanduninfected.Theredareais

localizedinamorenorthernareawithrespectto(a);(c)intersectionoftheredareas(a)and(b).Sincetheintersectionislessthan5%(inourcasethedifferenceis4%orless)

ofthetotalredarea,wecanconsiderthedifferenceassignificant.Thebluepointscorrespondtouninfectedapiariesandredpointstoinfectedapiaries.(c)ispositionedin

themiddlebetween(a)and(b)tobettershowtheintersection.×300,000.

Fig.3. PossiblespreadingpathwaysfromtheidentifiedspreadingcentrearemappedintheinvadedregiononthebasisofthealgorithmDBSCAN.Thepinkandbluelines

representtheVoronoitessellationofthespaceaftertheKmeansanalysisforthreeclusters(checkedwithSilhouettemethod,seethetext).Thedashedarrowsconnectthe

centroidoftheclustersandshouldgivesuggestionsabouttheaxesofdiffusion.(a)ResultsoftheGPanalysis.(b)Representationoftherecordedobservations(infection

cases)ontheterritorywiththerepresentativesofthedifferentclustershighlightedwithdifferentcolours.Thebluepointscorrespondtouninfectedapiaries.×250,000.

transport), in agreement with previous findings (Hood, 2004;

Spiewoketal.,2008).WemustalsoconsiderthatwindsinCalabria

blowmostcommonlynorthwardfromsouthandsoutheastregions

(FratianniandAcquaotta,2017),andthethirdmostdistantcluster

ispositionedjustsouthtothefirsttwo.Hence,dispersalappears tobeintheoppositedirectionwithrespecttothesiroccowind. Moreover,intheareaofthethirdclusternomainroadispresent, henceapparentlythespreadingwasdue toshortdistancelocal movements.Theproposalofaspreadingmodelwithbeekeepers infesting their other apiaries through ‘unintentional transfer’ was called “distance and ownership model” (EFSA, 2015) and thismodelofspreadingwouldbesupportedbyourdata.Onthe contrarythe first two northern clustersare in industrial areas wherethehumanpresenceisverycommon:herethespreading occurredatrelativelylowdistance(lessthan1km)and maybe

due partly to thenatural movement of the pest and partly by accidentaltransportbyhumanactivity.

Policiestoreducetheriskoffurtherspreadingofthepestmaybe hencecentredmainlyontheregulationoftheactivityofbeekeepers proposingguidelinesaimingtoreducethetransferofmaterialfrom onehivetoanotherortotransporthivesfromoneareatoanother. Aswithanymodellingapproach,GPhassomelimitations.First, itsreliabilityreflectstheaccuracyofpresencedata.Unfortunately, standardizedpresencedataareverydifficulttoobtainforinvasive speciesrightaftertheirinvasion.IntheSHBcasepresentedhere, contrarytowhatusuallyhappenswithmanyIAS,thepestspecies wasalreadyrecognizedasahoneybeenotifiablepest (Commis-sionDecision2004/216/EC),sothatveterinaryservicesrapidlyput inplaceastandardizedsurveyinawideareaaroundthefirstSHB reports(Mutinellietal.,2014),whichresultedinareliabledatabase,

(5)

thusprovidingstrengthtoourmodellingapproach.Asecond pos-siblesourceofbiasedsamplingisrepresentedbytheabilityofSHB tosurviveonnon-honeybeerelatedresources,suchasrottenand fermentingfruitsorotherspecies(suchasbumblebees,Hoffmann etal.,2008),whichmightrepresentarefugeforSHBallowing fur-therre-infestations(NeumannandElzen,2004).However,sofar noSHBhasbeendetectedonrottenfruitsinafieldinvestigation carriedoutintheinfestedareabetween2014and2015suggesting thatfermentingfruitsdonotrepresentapossiblealternativefood resourceforSHBinCalabria(Mutinellietal.,2015).However,ithas beenshownthatthesealternativeresourcesarelesspreferredand lesssuccessfulfoodsources,thatalsoreduceSHBfitness(Ellisetal., 2002),sothatweareconfidentthattheuseddatasetrepresenta reliablepresencedataset.

Despite these possiblelimitations, we thus believethat our resultsprovideafirstreliablehypothesisofSHBinvasiondynamics inItaly,supportingthehypothesisofaseaportmediatedinvasion

(Mutinellietal.,2014forItaly,andHood,2004forUSandAustralia),

andfurtherhighlightingtheroleofseaportsasIASintroduction sites(Macketal.,2000).TheVoronoitessellationandthedashed lineslinkinggroupsofobservations(seeFig.3a)shouldrepresent anestimateofthespreadingdirection,whilethedistancefromthe hypothesizedpointoforigin,relatedtotheguessedtimeofarrival ofSHB,shouldgiveanideaaboutthespeedofthediffusionitself intheterritory.

Moregenerally,webelieveourresultsshowthatthecouplingof GPandDBSCANmightrepresentapromisingandrapidapproach tousethesedatatoprioritizeareas,insideIASinvadedranges,in whichtofocusfuturemonitoringandmanagingefforts.

Thisanalysismaybeusedefficientlyinareaswheretheinvasion isstillatthebeginningasinBrazil(AlToufailiaetal.,2017),inorder tobetterunderstandthecausesoftheinvasionanditsspreading directionsinordertoundertakesuitableactions.

Compliancewithethicalstandards

Thisarticledoesnotcontainanystudieswithhumanorother animalparticipantsperformedbyanyoftheauthors.

Authorcontributionstatement

AC,USandAPconceivedresearch.ACcollectedthedata.AC,US andAPanalyzedthedata.ACwrotethemanuscript.Allauthors readandapprovedthemanuscript.

Conflictofinterest

Theauthorsdeclarenoconflictsofinterest. References

AlToufailia,H.,Alves,D.A.,Benà,D.deC.,Bento,J.M.S.,Iwanicki,N.S.A.,Cline,A.R.,

Ellis,J.D.,Ratnieks,F.L.W.,2017.Firstrecordofsmallhivebeetle,Aethinatumida

Murray,inSouthAmerica.J.ApiculturalRes.56(1),76–80.

Cini,A.,Anfora,G.,Escudero-Colomar,L.A.,Grassi,A.,Santosuosso,U.,Seljak,G.,

Papini,A.,2014.TrackingtheinvasionofthealienfruitpestDrosophilasuzukii

inEurope.J.PestSci.87(4),559–566.

EFSAPanelonAnimalHealthandWelfare(AHAW),2015.Survival,spreadand

estab-lishmentofthesmallhivebeetle(Aethinatumida).EFSAJ.13(12),4328.

Ellis,J.D.,Neumann,P.,Hepburn,R.,Elzen,P.J.,2002.Longevityandreproductive

successofAethinatumida(Coleoptera:Nitidulidae)feddifferentnaturaldiets.J. Econ.Entomol.95(5),902–907.

Elzen,P.J.,Baxter,J.R.,Westervelt,D.,Randall,C.,Delaplane,K.S.,Cutts,L.,Wilson,

W.T.,1999.Fieldcontrolandbiologystudiesofanewpestspecies,Aethina

tumidaMurray(Coleoptera,Nitidulidae),attackingEuropeanhoneybeesinthe WesternHemisphere.Apidologie30,361–366.

Ester,M.,Kriegel,H.P.,Sander,J.,Xu,X.,1996.Adensity-basedalgorithmfor

dis-coveringclustersinlargespatialdatabaseswithnoise.In:Simoudis,E.,Han,J., Fayyad,U.(Eds.),Proceedingsofthe2ndInternationalConferenceonKnowledge DiscoveryandDataMining.AAAIPress,pp.226–231.

Eyer,M.,Chen,Y.P.,Schäfer,M.O.,Pettis,J.,Neumann,P.,2009.Smallhivebeetle,

Aethinatumida,asapotentialbiologicalvectorofhoneybeeviruses.Apidologie 40(4),419–428.

Faulkner,S.C.,Verity,R.,Roberts,D.,Roy,S.S.,Robertson,P.A.,Stevenson,M.D.,Le

Comber,S.C.,2017.Usinggeographicprofilingtocomparethevalueofsightings

vstrapdatainabiologicalinvasion.Divers.Distrib.23(1),104–112.

Fratianni,S.,Acquaotta,F.,2017.TheclimateofItaly.In:Soldati,M.,Marchetti,M.

(Eds.),LandscapesandLandformsofItaly.WorldGeomorphologicalLandscapes. Springer,Cham.

Granato,A.,Zecchin,B.,Baratto,C.,Duquesne, V.,Negrisolo,E.,Chauzat, M.P.,

Mutinelli,F.,2017.IntroductionofAethinatumida(Coleoptera:Nitidulidae)

in the regions of Calabria and Sicily (southern Italy). Apidologie 48 (2), 194–203.

Haack, R.A., 2001. Intercepted Scolytidae (Coleoptera) at US ports of entry:

1985–2000.Integr.PestManage.Rev.3–4,253–282.

Hoffmann,D.,Pettis,J.S.,Neumann,P.,2008.Potentialhostshiftofthesmallhive

beetle(Aethinatumida)tobumblebeecolonies(Bombusimpatiens).InsectesSoc. 55(2),153–162.

Hood,W.M.,2004.Thesmallhivebeetle,Aethinatumida:areview.BeeWorld85

(3),51–59.

Hulme,P.E.,Bacher,S.,Kenis,M.,Klotz,S.,Kühn,I.,Minchin,D.,Nentwig,W.,Olenin,

S.,Panov,V.,Pergl,J.,Pyˇsek,P.,Roques,A.,Sol,D.,Solarz,W.,Vilà,M.,2008.

Grasp-ingattheroutesofbiologicalinvasions:aframeworkforintegratingpathways intopolicy.J.Appl.Ecol.45(2),403–414.

Keller,R.P.,Drake,J.M.,Drew,M.B.,Lodge,D.M.,2011.Linkingenvironmental

con-ditionsandshipmovementstoestimateinvasivespeciestransportacrossthe globalshippingnetwork.Divers.Distrib.17(1),93–102.

Lee,S.,Hong,K.J.,Cho,Y.S.,Choi,Y.S.,Yoo,M.S.,Lee,S.,2017.Reviewofthesubgenus

AethinaErichsons.str.(Coleoptera:Nitidulidae:Nitidulinae)inKorea,reporting recentinvasionofsmallhivebeetle,Aethinatumida.J.Asia-PacificEntomol.20 (2),553–558.

Lounsberry,Z.,Spiewok,S.,Pernal,S.F.,Sonstegard,T.S.,Hood,W.M.,Pettis,J.,

Neu-mann,P.,Evans,J.D.,2010.WorldwidediasporaofAethinatumida(Coleoptera:

Nitidulidae),anestparasiteofhoneybees.Ann.Entomol.Soc.Am.103(4), 671–677.

Mack,R.N.,Simberloff,D.,MarkLonsdale,W.,Evans,H.,Clout,M.,Bazzaz,F.A.,2000.

Bioticinvasions:causes,epidemiology,globalconsequences,andcontrol.Ecol. Appl.10(3),689–710.

McCullough,D.G.,Work,T.T.,Cavey,J.F.,Liebhold,A.M.,Marshall,D.,2006.

Intercep-tionsofnonindigenousplantpestsatUSportsofentryandbordercrossings overa17-yearperiod.Biol.Invasions8(4),611–630.

Murilhas,A.M.,2004.AethinatumidaarrivesinPortugal.Willitbeeradicated?

EurBeeNewsletter2,7–9.

Mutinelli,F.,Federico,G.,Carlin,S.,Montarsi,F.,Audisio,P.,2015.Preliminary

inves-tigationonotherNitidulidaebeetlesspeciesoccurringonrottenfruitinReggio Calabriaprovince(south-westernItaly)infestedwithsmallhivebeetle(Aethina tumida).J.ApiculturalRes.54(3),233–235.

Mutinelli,F.,Montarsi,F.,Federico,G.,Granato,A.,MaroniPonti,A.,Grandinetti,G.,

Chauzat,M.P.,2014.DetectionofAethinatumidaMurray(Coleoptera:

Nitiduli-dae.)inItaly:outbreaksandearlyreactionmeasures.J.ApiculturalRes.53(5), 569–575.

Neumann,P.,Elzen,P.J.,2004.Thebiologyofthesmallhivebeetle(Aethinatumida,

Coleoptera:Nitidulidae):gapsinourknowledgeofaninvasivespecies. Apidolo-gie35,229–247.

Neumann,P.,Pettis,J.S.,Schäfer,M.O.,2016.QuovadisAethinatumida?Biologyand

controlofsmallhivebeetles.Apidologie47,427–466.

Palmeri,V.,Scirtò,G.,Malacrinò,A.,Laudani,F.,Campolo,O.,2015.Ascientificnote

onanewpestforEuropeanhoneybees:firstreportofsmallhivebeetleAethina tumida,(Coleoptera:Nitidulidae)inItaly.Apidologie46(4),527–529.

Papini,A.,Santosuosso,U.,2017.Snow’scaserevisited:newtoolingeographic

profilingofepidemiology.Braz.J.Infect.Dis.21(1),112–115.

Papini,A.,Signorini,M.A.,Foggi,B.,DellaGiovampaola,E.,Ongaro,L.,Vivona,L.,

Santosuosso,U.,Tani,C.,Bruschi,P.,2017b.Historyvs.legend:retracinginvasion

andspreadofOxalispes-capraeL.inEuropeandtheMediterraneanarea.PLoS One12(12),e0190237.

Papini,A.,Mosti,S.,Santosuosso,U.,2013.Trackingtheoriginoftheinvading

Caulerpa(Caulerpales,Chlorophyta)withgeographicprofiling,acriminological techniqueforakilleralga.Biol.Invasions15(7),1613–1621.

Papini,A.,Rossmo,D.K.,LeComber,S.C.,Verity,R.,Stevenson,M.D.,Santosuosso,U.,

2017a.Theuseofjackknifingfortheevaluationofgeographicprofiling

reliabil-ity.Ecol.Informatics38,76–81.

Rossmo,D.K.,2000.GeographicProfiling.CRCPress,BocaRaton,FL.

Santosuosso,U.,Papini,A.,2018.Geo-profiling:beyondthecurrentlimits.A

pre-liminarystudyofmathematicalmethodstoimprovethemonitoringofinvasive species.Russ.J.Ecol.49(4),362–370.

Santosuosso,U.,Papini,A.,2016. Methodsfor geographicprofiling of

biologi-calinvasionswithmultipleoriginsites.Int.J.Environ.Sci.Technol.13(8), 2037–2044.

Schmolke,M.D.,1974.AStudyofAethinatumida:TheSmallHiveBeetle.University

ofRhodesia,Rhodesia,pp.178.

Smith,R.,Bivens,A.,Embrechts,M.,Palagiri,C.,Szymanski,B.,2002.Clustering

approachesforanomaly-basedintrusiondetection.In:Proceedingsofthe Intel-ligentEngineeringSystemsThroughArtificialNeuralNetworks.ASMEPress,pp. 579–584.

Smiti,A.,Eloudi,Z.,2013.SoftDBSCAN:improvingDBSCANclusteringmethodusing

(6)

Spiewok,S.,Neumann,P.,2006.Infestationofcommercialbumblebee(Bombus impa-tiens)fieldcoloniesbysmallhivebeetles(Aethinatumida).Ecol.Entomol.31(6), 623–628.

Spiewok,S.,Duncan,M.,Spooner-Hart,R.,Pettis,J.S.,Neumann,P.,2008.Smallhive

beetle,Aethinatumida,populationsII:dispersalofsmallhivebeetles.Apidologie 39(6),683–693.

Spiewok,S.,Pettis,J.S.,Duncan,M.,Spooner-Hart,R.,Westervelt,D.,Neumann,P.,

2007.Smallhivebeetle,Aethinatumida,populationsI:infestationlevelsof

hon-eybeecolonies,apiariesandregions.Apidologie38(6),595–605.

Stevenson,M.D.,Rossmo,D.K.,Knell,R.J.,LeComber,S.C.,2012.Geographicprofiling

asanovelspatialtoolfortargetingthecontrolofinvasivespecies.Ecography 35,1–12.

Wilson,J.R.,Dormontt,E.E.,Prentis,P.J.,Lowe,A.J.,Richardson,D.M.,2009.

Some-thinginthewayyoumove:dispersalpathwaysaffectinvasionsuccess.Trends Ecol.Evol.24(3),136–144.

Xu,R.,Wunsch,D.,2005.Surveyofclusteringalgorithms.IEEETrans.Neural