Can Niche Dynamics and Distribution Modeling Predict the Success of Invasive Species Management Using Biocontrol? Insights From Acacia longifolia in Portugal

e o-08-576667 Oc obe 14, 2020 Time: 16:43 # 1 ORIGINAL RESEARCH published: 19 Oc obe 2020 doi: 10.3389/ e o.2020.576667 Edi ed by: Rubén G. Ma eo, Au onomous Uni e si y o Mad id, Spain Re iewed by: Ai o Gas ón, Poly echnic Uni e si y o Mad id, Spain Tiago Vasconcelos, São Paulo S a e Uni e si y, B azil *Co espondence: Ma co Dinis [email p o ec ed] Joana R. Vicen e [email p o ec ed] Elizabe e Ma chan e [email p o ec ed] Hélia Ma chan e [email p o ec ed] Special y sec ion: This a icle was submi ed o Biogeog aphy and Mac oecology, a sec ion o he jou nal F on ie s in Ecology and E olu ion Recei ed: 26 June 2020 Accep ed: 25 Sep embe 2020 Published: 19 Oc obe 2020 Ci a ion: Dinis M, Vicen e JR, Césa de Sá N, López-Núñez FA, Ma chan e E and Ma chan e H (2020) Can Niche Dynamics and Dis ibu ion Modeling P edic he Success o In asi e Species Managemen Using Biocon ol? Insigh s F om Acacia longi olia in Po ugal. F on . Ecol. E ol. 8:576667. doi: 10.3389/ e o.2020.576667 Can Niche Dynamics and Dis ibu ion Modeling P edic he Success o In asi e Species Managemen Using Biocon ol? Insigh s F om Acacia longi olia in Po ugal Ma co Dinis1,2,3*, Joana R. Vicen e2*, Nuno Césa de Sá1,3,4, F ancisco A. López-Núñez3, Elizabe e Ma chan e3*and Hélia Ma chan e1,3* 1Cen e o Func ional Ecology - Science o People & he Plane , Ins i u o Poli écnico de Coimb a, Escola Supe io Ag á ia de Coimb a, Coimb a, Po ugal, 2Cen o de In es igação em Biodi e sidade e Recu sos Gené icos (CIBIO/InBIO), Uni e sidade do Po o, Po o, Po ugal, 3Cen e o Func ional Ecology - Science o People & he Plane , Depa men o Li e Sciences, Uni e si y o Coimb a, Coimb a, Po ugal, 4Ins i u e o En i onmen al Sciences, Leiden Uni e si y, Leiden, Ne he lands Biological in asions a e among he majo con empo a y h ea s o biodi e si y. Biocon ol has a long his o y as a sa e and e ec i e s a egy o he con ol o in asi e species in se e al wo ld egions, ye he li e his o y and ecological equi emen s o biocon ol agen s a e o en poo ly unde s ood. Species dis ibu ion models and assessmen s o niche o e lap and dynamics cons i u e a way o quan i y and compa e ecological niches and a e widely used o p edic ing biological in asions. While hese ools can also be aluable o p edic he e ec i eness o biocon ol p og ams and an icipa e whe he biocon ol agen s can es ablish on a eas a ge ed o con ol, hey emain unde u ilized o his pu pose. In his wo k, species dis ibu ion models and compa isons o niche dynamics a e used o p edic he success o he ongoing biocon ol p og am o Acacia longi olia in Po ugal, and po en ially along he Medi e anean Basin, using he Aus alian gall- o ming wasp T ichilogas e acaciaelongi oliae, p e iously eleased in Sou h A ica. Niches o bo h he in asi e plan and he biocon ol agen we e ound o be highly simila and o emain conse ed h ough he in oduc ion p ocess. Dis ibu ion models iden i y sui able clima ic a eas o A. longi olia in 19% o he Medi e anean Basin and p edic success ul es ablishmen o T. acaciaelongi oliae in 41% o he sui able a ea o A. longi olia, despi e excluding a ew loca ions whe e e y ecen es ablishmen occu ed. These esul s allow us o quan i y he isk o u u e A. longi olia in asion and po en ial success o biocon ol, as well as es ablish a compa a i e amewo k o simila p og ams being conside ed in o he egions o he wo ld dealing wi h A. longi olia in asions. Keywo ds: biological in asions, biocon ol, dis ibu ion modeling, niche conse a ism, Acacia longi olia, T ichilogas e acaciaelongi oliae F on ie s in Ecology and E olu ion | www. on ie sin.o g 1Oc obe 2020 | Volume 8 | A icle 576667 e o-08-576667 Oc obe 14, 2020 Time: 16:43 # 2 Dinis e al. Niche-Based Biocon ol Managemen o Acacia INTRODUCTION Biological in asions a e among he majo con empo a y h ea s o biodi e si y (IPBES, 2019). Speci ically, in asi e plan s cause changes in ecosys em se ices and unc ionali y, wi h nega i e impac s a di e en le els, e.g., a he le el o plan and soil communi ies and nu ien dynamics (Ma chan e e al., 2008a,b; Le Mai e e al., 2011;Vilà e al., 2011;Simbe lo e al., 2013; Zhang e al., 2019) which a e o en di icul o e e se o mi iga e wi hou ac i e es o a ion e o s (Le Mai e e al., 2011). Con ol o in asi e species is equen ly p ohibi i ely expensi e and labo -in ensi e (Ma ais e al., 2004). Unde s anding biological in asion p ocesses is essen ial o in o ming decision-making and op imizing he alloca ion o limi ed esou ces. Co ela i e app oaches based on geog aphic occu ence da a (such as species dis ibu ion models and mul i a ia e en i onmen al assessmen s o niche o e lap) a e aluable o quan i y and compa e ecological niches and a e widely used o p edic and imp o e managemen o biological in asions (e.g., B oennimann e al., 2007;Vicen e e al., 2010, 2011, 2016;Pe i pie e e al., 2012;Guisan e al., 2014). Biological con ol (he ea e biocon ol) has a long his o y (o e a cen u y) as a sa e and e ec i e s a egy o he con ol o in asi e plan species (Clewley e al., 2012) in se e al egions o he wo ld, such as Sou h A ica, Aus alia, o Uni ed S a es. Ye , he biocon ol o in asi e plan s was only ecen ly (less han a decade ago) ini ia ed in Eu ope (Shaw e al., 2018). Classical biocon ol consis s on he delibe a e in oduc ion o a na u al enemy o he a ge in asi e species in o a new egion in aded by he la e wi h he objec i e o mi iga ing i s compe i i e ad an age. Biocon ol agen s should be highly speci ic in o de o minimize in e ac ions wi h na i e biodi e si y in he egions whe e hey a e in oduced. They should also be able o es ablish iable popula ions in he in aded a eas. Conside ing his, i ollows ha he chances o success o a biocon ol agen a e likely o be imp o ed when he ecological equi emen s o i s success ul es ablishmen and long- e m pe sis ence a e simila o hose o i s a ge . These equi emen s, due o hei explici ly geog aphically-o ien ed and scenopoe ic na u e, all wi hin he G innellian de ini ion o he ecological niche (sensu Pe e son e al., 2011). Consequen ly, he simila i y in ecological equi emen s be ween species can be cha ac e ized using he ela ed concep o niche o e lap (sensu Wa en e al., 2008). Despi e his, e y ew s udies so a ha e explici ly compa ed biocon ol agen s and hei espec i e a ge s in e ms o niche o e lap (bu see Sun e al., 2017). Acacia longi olia is a Leguminosae na i e o sou heas e n Aus alia which has become highly in asi e in se e al exo ic loca ions. Cu en ly, in asi e popula ions o A. longi olia a e es ablished beyond i s na i e ange in sou h-wes e n Aus alia (Cos ello e al., 2000;Impson e al., 2011), as well as in New Zealand, Sou h A ica, Chile, Uni ed S a es (Cali o nia), Spain (Galicia), and Po ugal (EFSA Panel on Plan Heal h, 2015). La ge long-li ed seed banks and he abili y o ge mina e ollowing dis u bances (pa icula ly i e) a e common ai s in he Acacia genus which accoun o i s high in asi e abili y (Richa dson and Kluge, 2008;Passos e al., 2017). In asions by A. longi olia a e associa ed o changes in soil communi ies and nu ien cycles (Ma chan e e al., 2008a,b), dis u bance o plan (Cos ello e al., 2000;Ma chan e e al., 2003, 2015), and plan - gall communi ies (López-Núñez e al., 2017), nu ien and wa e cycling (We ne e al., 2010), and i e egimes. Many o hese e ec s end o pe sis a e emo al, hinde ing es o a ion e o s (Ma chan e e al., 2009, 2011a). Mechanical and chemical con ol o in asi e Acacia species is o en highly demanding in e ms o esou ces, ime and labo (Ma ais e al., 2004). Reduc ion o seed banks is pa icula ly p oblema ic, ye essen ial o he long- e m managemen o in asi e plan s wi h high seed p oduc ion (Richa dson and Kluge, 2008). Biological con ol is consequen ly o en p oposed as a aluable and indispensable asse o he managemen o Acacia in asions (Ma ais e al., 2004;Richa dson and Kluge, 2008;Wilson e al., 2011). T ichilogas e acaciaelongi oliae is a gall- o ming wasp endemic o Aus alia. Like all gall- o ming insec s, i is a highly speci ic endopa asi e which equi es success ul in e ac ion wi h i s hos o comple e he li e cycle, making i an ideal candida e o biocon ol p og ams due o he ex emely low isk o de imen al in e ac ions wi h species o he han i s p e e ed hos , and i s inabili y o expand beyond he dis ibu ion o i s hos (Dennill e al., 1993;Ma chan e e al., 2011b). Cu en ly, only wo p og ams using T. acaciaelongi oliae o con ol A. longi olia exis in he wo ld. I was i s in oduced in Sou h A ica in 1982 and 1983 (Dennill, 1985, 1988, 1990;Dennill and Donnelly, 1991; Dennill e al., 1993) whe e i quickly became es ablished h ough mos o he ange o A. longi olia, success ully dec easing he ep oduc i e po en ial (89–95%) and ege a i e g ow h (53%) o i s a ge and causing A. longi olia mo ali y in s ess ul si ua ions (Dennill, 1985, 1988;Dennill and Go don, 1990). Owing o he success o A. longi olia biocon ol in Sou h A ica, a simila p og am was ini ia ed in Po ugal (Ma chan e e al., 2011b), wi h he i s elease o T. acaciaelongi oliae in he wild aking place ecen ly, in 2015 (Ma chan e e al., 2017). As o 2018, ou popula ions o T. acaciaelongi oliae es ablished in Po ugal, and i is expec ed ha hey will main ain he cu en end o popula ion g ow h and expansion, e en ually eaching all a eas o he coun y in aded by A. longi olia (Ma chan e e al., 2017). While he p og am was ini ia ed in Po ugal, he agen may po en ially sp ead o o he egions o he Medi e anean Basin whe e adequa e en i onmen al condi ions a e a ailable. Unde s anding how he niches o a biocon ol agen and i s a ge species ela e in hei na i e dis ibu ion may be insu icien o p edic he success o biocon ol. The possibili y o niche shi s du ing in oduc ion in o new en i onmen s (Guisan e al., 2014) may p omo e niche di e gence be ween he wo species, c ea ing a misma ch in en i onmen al space which, when anspo ed o geog aphic space, can esul in inabili y o he biocon ol agen o es ablish in a eas a ge ed o con ol. This wo k aims o cha ac e ize he ealized niches o A. longi olia and T. acaciaelongi oliae and es he adequacy o T. acaciaelongi oliae as a biocon ol agen o A. longi olia in Po ugal as a unc ion o niche o e lap. To accomplish his, we cha ac e ize and compa e he ealized niches o he hos plan and o he biocon ol agen in hei na i e and in oduced anges. We also es he hypo hesis o niche shi s by compa ing ealized niches be ween na i e and in oduced anges o bo h species. F on ie s in Ecology and E olu ion | www. on ie sin.o g 2Oc obe 2020 | Volume 8 | A icle 576667 e o-08-576667 Oc obe 14, 2020 Time: 16:43 # 3 Dinis e al. Niche-Based Biocon ol Managemen o Acacia We hen p ojec he en i onmen al a ini ies o A. longi olia and T. acaciaelongi oliae in o geog aphical space using species dis ibu ion models, in o de o p edic he success o he biocon ol p og am p ima ily in Po ugal bu ul ima ely in he geog aphically and clima ically p oxima e Medi e anean Basin. MATERIALS AND METHODS Occu ence Reco ds A da abase o A. longi olia and T. acaciaelongi oliae occu ence eco ds was compiled om he Global Biodi e si y In o ma ion Facili y1, A las o Li ing Aus alia2, Sou h Aus alian Museum3 p o ided upon eques by Doc o Pe e Hudson, Aus alian Na ional Insec Collec ion4p o ided upon eques by Doc o Juani a Rod iguez, In asi e Plan s in Po ugal – In aso as.p 5 (Ma chan e e al., 2017), Sou he n A ica Plan In ade s A las (Hende son, 1999), li e a u e e e ences (Dennill, 1985, 1987; P insloo and Nese , 2007;Lado, 2008;Hen iksen e al., 2017, 2019); and da a p o ided by au ho s and collabo a o s (see sec ion “acknowledgmen s”). Da a collec ion o A. longi olia and T. acaciaelongi oliae was independen and no eco ds we e sha ed among he wo da ase s. Occu ence da a we e conside ed un il June 2018. A. longi olia eco ds we e included aking in o accoun he species sensu la o (i.e., including subspecies A. longi olia longi olia,A. longi olia lo ibunda and A. longi olia sopho ae, all conside ed hos s o T. acaciaelongi oliae). In June 2018, T. acaciaelongi oliae was s ill na owly dis ibu ed in Po ugal, and hus occu ence eco ds in his a ea we e no conside ed o subsequen analyses. Reco ds ou side he na i e a ea (sou heas e n Aus alia) and he in oduced a eas o in e es (Sou h A ica and Po ugal) we e excluded. Unce ain o duplica e eco ds we e also excluded. Reco ds wi h less posi ional accu acy han 10 km we e excluded o maximize geog aphic co e age while minimizing he loss o spa ial p ecision (Gu ié ez-Rod íguez e al., 2017). To educe po en ial sampling bias and spa ial au oco ela ion, a declus e ing app oach (e.g., Vale e al., 2016;Dinis e al., 2019) was pe o med using he R package ecospa (Di Cola e al., 2017), which consis ed o delimi ing a 10 km bu e a ound each eco d and, when wo o mo e bu e s o e lapped, emo ing one o he eco ds andomly. This p ocess was epea ed un il all emaining eco ds pe species had a dis ance o a leas 10 km among hem. Clus e ing was quan i ied be o e and a e he declus e ing ea men o ensu e an e ec i e dec ease in he le el o clus e ing (e.g., Dinis e al., 2019). This was done by calcula ing he Nea es Neighbo Index in A cMap 10.1 (En i onmen al Sys ems Resea ch Ins i u e, 2012). The inal da abase consis ed o 1493 and 140 occu ences o A. longi olia and T. acaciaelongi oliae, espec i ely (Supplemen a y Table 1). Occu ences we e hen 1h ps://www.gbi .o g/en/ 2h ps://www.ala.o g.au/ 3h p://www.samuseum.sa.go .au/ 4h ps://www.csi o.au/en/Resea ch/Collec ions/ANIC 5h p://www.in aso as.p u he subdi ided by geog aphical a eas o in e es (sou heas e n Aus alia, Sou h A ica, and Po ugal; Figu e 1). Bioclima ic Va iables and S udy A ea The na i e dis ibu ions o A. longi olia and T. acaciaelongi oliae we e ep esen ed by sou heas e n Aus alia. Sou h A ica and Po ugal we e selec ed o ep esen a eas o A. longi olia in asion because hey a e he a eas whe e biocon ol p og ams wi h T. acaciaelongi oliae a e ongoing. The Medi e anean Basin (sligh ly expanded o include also no he n Spain) was selec ed o model p ojec ions o wo easons. Fi s , i is he geog aphical a ea o which Po ugal is pa and A. longi olia occu s wi h in asi e po en ial in se e al coun ies in he egion (Spain, F ance, I aly, Lo enzo e al., 2010); a he same ime, i A. longi olia is p esen , his c ea es he possibili y o na u al dispe sion o T. acaciaelongi oliae. Second, clima ic simila i y wi h he global dis ibu ion o A. longi olia and T. acaciaelongi oliae (p edomina ely in a eas o Medi e anean clima e) is expec ed o maximize analogy o p edic o s. Nine een bioclima ic a iables ep esen ing a e age con empo a y condi ions (1970–2000) we e ob ained om Wo ldclim 2.06(Hijmans e al., 2005) a a spa ial esolu ion o 10 ×10 km, o ensu e consis ency wi h he minimum posi ional accu acy o he occu ence da a. O he scenopoe ic candida e a iables, such as opog aphy, we e conside ed bu ul ima ely excluded unde he assump ion ha hey would no be in o ma i e a his spa ial esolu ion, while a iables ela ed o soil composi ion and ege a ion co e /s uc u e we e excluded due o being dynamically linked o he p esence o he a ge species, consequen ly alling ou side he scope o he G innellian niche (Pe e son e al., 2011). The empo al ex en o he Wo ldclim da ase o e laps wi h he collec ion da es o mos occu ence eco ds used in his wo k, and was he e o e selec ed as an adequa e ep esen a ion o clima ic condi ions a ailable o he a ge species upon he ime o obse a ion. Backg ound and aining a eas we e de ined by clipping bioclima ic laye s o a 200 km bu e su ounding he minimum con ex polygon which includes all occu ence eco ds. Bu e size was selec ed aking in o accoun coa se dis ibu ional pa e ns, opog aphy, ege a ion zones, and he limi s o clima ic egions acco ding o he Köppen-Geige clima e classi ica ion (Köppen, 1900) and aims o ep esen he habi a s and a eas which he species may ha e easonably sampled in each egion (Ba e e al., 2011) and p e en biases in he sampling o backg ound da a, which a e known o occu when using o e ly b oad o es ic i e aining a eas (VanDe Wal e al., 2009;Ande son and Raza, 2010). This was done sepa a ely o each species in each ele an a ea o dis ibu ion, i.e.: o A. longi olia in i s na i e ange in sou heas e n Aus alia as well as he in asi e anges in Sou h A ica and Po ugal; and o T. acaciaelongi oliae in i s na i e and in oduced anges in sou heas e n Aus alia and Sou h A ica, espec i ely. Analyses we e also pe o med combining he mul iple na i e and in oduced/in aded anges o each species (Aus alia, Sou h A ica and Po ugal o A. longi olia, Aus alia and Sou h A ica o T. acaciaelongi oliae, 6h p://www.wo ldclim.o g F on ie s in Ecology and E olu ion | www. on ie sin.o g 3Oc obe 2020 | Volume 8 | A icle 576667 e o-08-576667 Oc obe 14, 2020 Time: 16:43 # 4 Dinis e al. Niche-Based Biocon ol Managemen o Acacia FIGURE 1 | S udy a eas and occu ence eco ds o Acacia longi olia and T ichilogas e acaciaelongi oliae used in his wo k. (A; op igh ): Po ugal; (B; bo om le ): Sou h A ica; and (C; bo om igh ): sou heas e n Aus alia. T ichilogas e acaciaelongi oliae eco ds om Po ugal a e me ely indica i e and we e no included in any analyses due o hei low numbe and ecen es ablishmen . Figu e 2). All ope a ions we e pe o med in A cMap 10.1 (En i onmen al Sys ems Resea ch Ins i u e, 2012). Niche O e lap Ecological niches we e compa ed be ween A. longi olia and T. acaciaelongi oliae, as well as be ween di e en na i e and in oduced a eas o each species’ dis ibu ion (Figu e 2). Fo A. longi olia, niches we e compa ed be ween he na i e ange in Aus alia and he wo in aded anges in Sou h A ica and Po ugal. Fo T. acaciaelongi oliae, compa isons ook in o accoun only he na i e and in oduced dis ibu ions in Aus alia and Sou h A ica, espec i ely. All h ee dis ibu ion a eas o A. longi olia we e compa ed, o a oid assump ions abou in oduc ion pa hways. Niches we e compa ed in mul i a ia e en i onmen al space using he PCA-en echnique (B oennimann e al., 2012). This app oach allows o pai wise compa isons o niches be ween g oups, and consis s o h ee s eps: (1) calcula ion o densi y o occu ences along a mul i a ia e en i onmen al space o bo h a ge g oups; (2) measu emen o he o e lap be ween densi y dis ibu ions o he wo g oups in en i onmen al space (niche o e lap), as de ined by Schoene ’s D me ic (Schoene , 1970); and (3) s a is ical es s o niche equi alency (whe he niche o e lap is cons an when andomizing occu ences among bo h g oups) and niche simila i y (whe he he niche o one g oup is mo e simila o he o he han would be expec ed by chance gi en he a ailable en i onmen al backg ound), using a pe mu a ion- based amewo k (Wa en e al., 2008;B oennimann e al., 2012). Rep esen a ion o niches as densi ies along en i onmen al space allows he quan i ica ion o di e ences in he densi ies o g oups, which can be exp essed as niche s abili y (p opo ion o niche B o e lapping wi h niche A), niche expansion (p opo ion o niche B non-o e lapping wi h niche A), and niche un illing [unique p opo ion o niche A (i.e., non-o e lapping wi h niche B); Guisan e al., 2014]. These me ics we e de eloped o compa e F on ie s in Ecology and E olu ion | www. on ie sin.o g 4Oc obe 2020 | Volume 8 | A icle 576667 e o-08-576667 Oc obe 14, 2020 Time: 16:43 # 5 Dinis e al. Niche-Based Biocon ol Managemen o Acacia FIGURE 2 | Wo k low o niche o e lap assessmen s and dis ibu ion modeling. Panel (A; op): Regional species dis ibu ion da ase s we e compa ed o assess niche conse a ism o Acacia longi olia and T ichilogas e acaciaelongi oliae. Niche o e lap among species was quan i ied (o e lap), es ed (equi alency and simila i y), and cha ac e ized (s abili y, expansion, and un illing) o assess ecological o e lap in en i onmen al space. Panel (B; bo om): Dis ibu ion modeling was used o p ojec ecological o e lap in o geog aphic space and p edic he po en ial abili y o T. acaciaelongi oliae o es ablish in a eas suscep ible o A. longi olia in asion in he Medi e anean basin. Ac onyms: AU: Aus alia, SA: Sou h A ica, and PT: Po ugal. na i e and exo ic niches, and ep esen he p opo ion (0–1) o he ealized niche ha emains cons an (s abili y), is gained (expansion), o is los (un illing) du ing in oduc ion/in asion. Applica ion ou side o an explici compa ison o na i e s in oduced anges (such as when compa ing wo un ela ed axa like A. longi olia and T. acaciaelongi oliae) equi es an a bi a y assignmen o each g oup o ei he niche A o B, which will al e he o de in which niche expansion and un illing a e calcula ed bu will no a ec he inal esul s o in e p e a ion. Backg ound a eas o each g oup we e spa ially delimi ed using he same a eas used o calib a ion o niche models. Fo each pai wise compa ison, backg ound a eas o he wo g oups unde compa ison we e me ged. En i onmen al space was de ined by he 19 bioclima ic a iables a ailable om Wo ldclim 2.0 (Hijmans e al., 2005), a a esolu ion o 10 ×10 km. Though se e al o hese a iables a e co ela ed in geog aphic space, he PCA-based app oach add esses his by c ea ing an o hogonal linea combina ion o he o iginal a iables. This app oach allows he inclusion o all a ailable bioclima ic a iables, maximizing he en i onmen al a iance unde conside a ion o niche assessmen s and minimizing assump ions abou a iable impo ance. Gi en wha li le is known abou he ecology o T. acaciaelongi oliae, we op ed o minimize such assump ions. Niche o e lap was calcula ed and niche equi alency and simila i y es s we e pe o med among egions o each species, and ac oss species. We also calcula ed me ics o niche s abili y, expansion and un illing using he same amewo k. Fo all es s among species, A. longi olia was conside ed as g oup A and T. acaciaelongi oliae as g oup B. All analyses we e pe o med in he R package ecospa (Di Cola e al., 2017). Species Dis ibu ion Models Based on he e idence o niche conse a ism o bo h A. longi olia and T. acaciaelongi oliae (see Resul s sec ion), species dis ibu ion models we e calib a ed using all a ailable eco ds o sou heas e n Aus alia, Sou h A ica and ( o A. longi olia) Po ugal. This allows maximiza ion o he numbe o occu ence eco ds used o ain he models and a be e sampling o ele an en i onmen al g adien s. To a oid o e pa ame iza ion, dis ibu ion models we e de eloped using a subse o he bioclima ic a iables used o niche o e lap analysis. This app oach p ecludes di ec compa ison be ween he wo me hods. Consequen ly, we ea he dis ibu ion models s ic ly as niche-based ools o p edic ing dis ibu ions a he han accu a e ep esen a ions o he niches, elying solely on he niche o e lap es s o any in e ences ega ding he dynamics and ela ionships o niches. Spa ial co ela ion be ween bioclima ic F on ie s in Ecology and E olu ion | www. on ie sin.o g 5Oc obe 2020 | Volume 8 | A icle 576667 e o-08-576667 Oc obe 14, 2020 Time: 16:43 # 6 Dinis e al. Niche-Based Biocon ol Managemen o Acacia a iables was assessed in A cMap 10.1 (En i onmen al Sys ems Resea ch Ins i u e, 2012), and mul icollinea i y was assessed by calcula ing he Va iance In la ion Fac o (VIF) using he usdm package in R (Naimi e al., 2014). When wo o mo e a iables we e highly co ela ed, p e e ence was gi en o he one wi h he mos biological sense. A se o i e and six sligh ly co ela ed (Pea son’s <0.7) and non-collinea a iables (VIF <5; e.g., Vicen e e al., 2013) we e selec ed o c ea e models o T. acaciaelongi oliae and A. longi olia, espec i ely, (Table 1). An ensemble modeling app oach was pe o med in he R package biomod2 (Thuille e al., 2009) using a combina ion o se en modeling echniques: Gene alized Linea Models (GLM), Gene alized Boos ing Models (GBM), Flexible Disc iminan Analysis (FDA), Classi ica ion T ee Analysis (CTA), Mul i a ia e Adap i e Reg ession Splines (MARS), Random Fo es s (RF), and Maximum En opy (MAXENT). Models we e c ea ed using 5 da ase s o 1000 pseudoabsences andomly dis ibu ed h ough he aining a eas, and 8 eplica es we e pe o med o each combina ion o pseudoabsence da ase and modeling algo i hm, o a o al o 280 models pe species. Each model was calib a ed se ing aside 30% o a ailable p esence eco ds o model e alua ion. The esul ing es ing da ase s we e used o e alua e model disc imina ion by calcula ing he a ea unde he cu e o he ecei e ope a ing cha ac e is ic cu e (AUC), and ensemble models ac oss all modeling echniques we e c ea ed o each species (e.g., Pe i pie e e al., 2017), using he mean o all models whe e AUC >0.7. Indi idual models we e p ojec ed o he Medi e anean Basin and bina y ensemble models we e c ea ed using he Maximized Sum Th eshold c i e ia (MST; Can o e al., 1999), implemen ed in biomod2 unde he “bina y.me h = ‘ROC”’ a gumen (Thuille e al., 2009). This me hod selec s he h eshold which maximizes he sum o sensi i i y and speci ici y. We op ed o use his app oach due o i s good gene al pe o mance in compa a i e s udies and pa icula ly o i s supe io pe o mance wi h low-p e alence da ase s and abili y o minimize omission e o s, which a e gene ally mo e cos ly in conse a ion applica ions such as iden i ying a eas unde isk o biological in asion (Liu e al., 2005;Jiménez-Val e de and Lobo, 2007). The impo ance o en i onmen al a iables o each model was de e mined by a e age pe mu a ion impo ance (Phillips e al., 2006). RESULTS Niche Cha ac e iza ion Niche o e lap among A. longi olia was o e all low be ween he h ee egions (less han 0.5, wi h 0 co esponding o no o e lap and 1 o comple e o e lap), wi h he g ea es o e lap occu ing be ween A. longi olia in Sou h A ica and Po ugal (0.416; Table 2). Niche equi alency, i.e., whe he niche o e lap emains cons an when andomizing occu ences among g oups, was ound in all pai wise es s excep o A. longi olia in Aus alia s. Po ugal and Sou h A ica s. Po ugal. Niche simila i y, i.e., whe he g oups a e mo e simila han expec ed conside ing a ailable backg ound, was ound in all pai wise es s. Simila niche dynamics we e ound o all A. longi olia compa isons, wi h p edominan niche s abili y (>0.89), analogous niches ac oss egions and e y li le expansion in o no el a eas o en i onmen al space du ing in oduc ion. Niche un illing, i.e., a eas o en i onmen al space p esen in o iginal ange bu no a e in oduc ion, anged be ween 0.2 and 0.26 o all pai s excep Sou h A ica s. Po ugal, which had e y li le un illing (<0.01). Fo T. acaciaelongi oliae in Aus alia s. Sou h A ica, niche un illing was o simila magni ude o A. longi olia (0.25), bu niche expansion was much mo e p onounced (0.56), wi h a co esponding dec ease in niche s abili y (0.44; Table 2). Fo compa isons be ween A. longi olia and T. acaciaelongi oliae, niche o e lap anged be ween 0.53 (in Aus alia) and 0.85 (in Sou h A ica). The niches o he wo species we e ound o be equi alen and simila i y was no ejec ed in any pai wise es . Niche s abili y was highe han 0.96 in all compa isons, wi h co esponding low alues o niche expansion om A. longi olia o T. acaciaelongi oliae (<0.04). Niche un illing om A. longi olia o T. acaciaelongi oliae was lowes in Sou h A ica (0.05) and highes in Aus alia (0.23; Table 2). All signi ican (p- alue <0.01) niche simila i y es s co esponded o alues o o e lap g ea e han he dis ibu ion TABLE 1 | Va iable ID and minimum, maximum, and mean alues ac oss model aining a eas o Acacia longi olia and T ichilogas e acaciaelongi oliae o ecogeog aphical a iables used in model c ea ion. Acacia longi olia T ichilogas e acaciaelongi oliae Min Max Mean Min Max Mean BIO2 Mean diu nal ange (C◦) 5.28 18.04 12.37 BIO3 Iso he mali y (%) 33.02 62.67 48.95 36.18 63.39 49.77 BIO4 Tempe a u e seasonali y 201.10 634.90 426.80 BIO6 Min empe a u e o coldes mon h (C◦)−6.15 13.42 3.18 BIO9 Mean empe a u e o d ies qua e (C◦) 0.50 25.03 15.93 BIO13 P ecipi a ion o we es mon h (mm) 20.00 299.00 91.38 BIO14 P ecipi a ion o d ies mon h (mm) 1.00 119.00 26.93 2.00 119.00 25.85 BIO15 P ecipi a ion seasonali y 8.55 87.14 39.22 BIO19 P ecipi a ion o coldes qua e (mm) 11.00 842.00 174.60 Va iables a e ep esen ed only o he species in which hey we e u ilized o model c ea ion. F on ie s in Ecology and E olu ion | www. on ie sin.o g 6Oc obe 2020 | Volume 8 | A icle 576667 e o-08-576667 Oc obe 14, 2020 Time: 16:43 # 7 Dinis e al. Niche-Based Biocon ol Managemen o Acacia o pseudo eplica es, indica ing ha he g oups a e mo e simila han expec ed conside ing he a ailable backg ound. Species Dis ibu ion Models All 280 models pe species had AUC alues abo e 0.7 and we e hus included in he ensembles. AUC alues o A. longi olia models anged be ween 0.79 and 0.94, wi h a weigh ed a e age o 0.90 ±0.02 (±SD). AUC o models o T. acaciaelongi oliae anged be ween 0.71 and 0.92, wi h a weigh ed a e age o 0.85 ±0.04 (±SD; Table 3). Tempe a u e seasonali y (BIO4), minimum empe a u e o he coldes mon h (BIO6), and p ecipi a ion o he coldes qua e (BIO19) we e he mos impo an a iables o he A. longi olia models, wi h ma ginal con ibu ions om he emaining a iables. Va iable impo ance o T. acaciaelongi oliae was less s aigh o wa d, wi h all a iables con ibu ing somewha o he models, ye iso he mali y (BIO3) and mean diu nal ange o empe a u e (BIO2) we e he a iables wi h he mos explana o y powe (Table 3). Sui able a eas o A. longi olia we e p edic ed h ough he A lan ic coas o he Ibe ian Peninsula and Mo occo, in some a eas ex ending up o 280 km inland (Figu e 3). Sui abili y o A. longi olia was also iden i ied in no he n Spain, sou he n F ance, some a eas o he I alian and Balkan peninsulas (pa icula ly along he Ad ia ic coas ), mos o he Medi e anean coas o No h A ica and he Middle Eas and all Medi e anean a chipelagos. T. acaciaelongi oliae is p edic ed o ha e sui able clima ic condi ions in mos o he Ibe ian coas line (excep o a gap o app oxima ely 300 km in no he n Po ugal), he en i e y o he coas line o Mo occo, mos o he Medi e anean a eas o Libya, Egyp , Is ael, and Pales ine and a eas in he Ad ia ic coas and in he islands o Cyp us, Sicily and he Aegean sea (Figu e 3 and Supplemen a y Figu e 1). In o al, 19% o he o al a ea o he Medi e anean basin was p edic ed o be sui able o A. longi olia, and 41% o he a ea iden i ied as sui able o A. longi olia is also p edic ed as sui able o T. acaciaelongi oliae. Cu en ly es ablished popula ion o T. acaciaelongi oliae we e all co ec ly iden i ied as sui able by he modeling app oach used, wi h he excep ions o he popula ion loca ed in he ci y o Coimb a (Figu e 3 and Supplemen a y Figu e 1) and a ew e y ecen ly obse ed popula ions in he no he n coas (no shown). DISCUSSION Niche Dynamics o A. longi olia and T. acaciaelongi oliae While he e a e di e ences in niche dynamics be ween di e en anges o A. longi olia and T. acaciaelongi oliae, he esul s a e o e all consis en wi h niche conse a ism du ing he in asion/in oduc ion p ocess, espec i ely. E idence o niche simila i y on all cases sugges s ha obse ed di e ences in niches a e ela ed o di e ences in habi a a ailabili y and/o in oduc ion pa hways be ween Aus alia, Sou h A ica and Po ugal. Compa isons be ween Aus alia and Sou h A ica e ealed con as ing pa e ns o he wo species, wi h A. longi olia in asion being associa ed wi h niche s abili y, and T. acaciaelongi oliae in oduc ion being associa ed wi h s ong niche expansion. I ’s unlikely ha his e lec s ac ual expansion o T. acaciaelongi oliae in o new en i onmen s in TABLE 2 | Niche o e lap and dynamics be ween all pai s o na i e (AU: Aus alia) and in oduced (SA: Sou h A ica; PT: Po ugal) Acacia longi olia (Al) and T ichilogas e acaciaelongi oliae (Ta). Pai s PC1 PC2 Niche o e lap (D) Equi alency (sig = non-eq.) Simila i y A- >B Simila i y B- >A Expansion S abili y Un illing Acacia longi olia AU-PT 44.27 20.19 0.18 0.001*0.052 0.07 0.084 0.916 0.257 AU-SA 37.24 26.64 0.33 0.263 0.012 0.009*0.102 0.898 0.211 SA-PT 45.01 19.67 0.416 0.001*0.011 0.013 0.06 0.94 0.004 T ichilogas e acaciaelongi oliae AU-SA 37.24 26.64 0.157 0.987 0.141 0.134 0.561 0.439 0.254 Among species AlSA-TaSA 35.02 27.08 0.846 0.604 0.004*0.005*0.039 0.961 0.053 AlAU-TaAU 48.96 22.37 0.526 0.94 0.094 0.084 0.001 0.999 0.227 AlAll-TaAll 39.71 24.08 0.556 1 0.09 0.09 0.013 0.987 0.106 Compa isons a e made wi hin each species and be ween he wo species conside ing he na i e (AU) and in oduced (SA) anges whe e bo h occu . Tes s be ween he wo species conside ing he ull ange whe e hey occu (AU, SA and PT o A. longi olia; AU and SA o T. acaciaelongi oliae) a e also p esen ed (AlAll-TaAll). Pe cen age o explained a ia ion o he wo i s p incipal componen (PC1 and PC2), niche o e lap (Schoene ’s D), p- alues o equi alency and simila i y es s (p <0.01, signi ican alues ma ked wi h *) and indices o niche expansion, s abili y and un illing a e p esen ed. All es s o niche simila i y p esen ed null dis ibu ions lowe han he obse ed o e lap, indica ing niche simila i y. TABLE 3 | Numbe o occu ence eco ds (N), a e age (and SD) A ea Unde he Cu e (AUC) and a e age (and SD) pe mu a ion impo ance o each ecogeog aphical a iable ac oss all 280 species dis ibu ion model eplica es o Acacia longi olia and T ichilogas e acaciaelongi oliae. ∗Indica e highes con ibu ing a iables. N AUC BIO3 BIO4 BIO6 BIO9 BIO14 BIO19 Acacia longi olia 1493 0.897 (0.022) 0.045 (0.033) 0.238* (0.089) 0.178* (0.089) 0.072 (0.052) 0.042 (0.051) 0.350* (0.094) N AUC BIO2 BIO3 BIO13 BIO14 BIO15 T ichilogas e acaciaelongi oliae 140 0.846 (0.040) 0.501* (0.109) 0.527* (0.113) 0.251 (0.153) 0.343 (0.296) 0.261 (0.236) F on ie s in Ecology and E olu ion | www. on ie sin.o g 7Oc obe 2020 | Volume 8 | A icle 576667 e o-08-576667 Oc obe 14, 2020 Time: 16:43 # 8 Dinis e al. Niche-Based Biocon ol Managemen o Acacia FIGURE 3 | Spa ial o e lap be ween model p edic ions o sui abili y o Acacia longi olia and T ichilogas e acaciaelongi oliae in he Medi e anean egion. T ichilogas e acaciaelongi oliae model was cons ained o a eas o p edic ed sui abili y o A. longi olia o p e en classi ica ion o sui abili y in a eas whe e he species canno pe sis due o lack o i s hos plan . Bo om: Popula ions o T. acaciaelongi oliae al eady es ablished in Po ugal as o 2018 Sou h A ica, as i s dependency on A. longi olia would equi e he hos plan o do he same, which is ejec ed by he e idence. Ra he , non-clima ic ac o s such as bio ic in e ac ions a e likely o cons ain he dis ibu ion o T. acaciaelongi oliae in Aus alia, whe e he species is na i e and as such su e s highe le els o pa asi ism and compe es wi h o he bud- eeding insec s (Nese , 1984). Despi e p esen , he lowe le el o such in e ac ions in Sou h A ica (Manongi and Ho mann, 1995; Seymou and Veld man, 2010) may ha e allowed he species o expand in o p e iously unoccupied a eas o niche space which a e none heless con ained wi hin he ealized niche o A. longi olia. The e idence o niche expansion may also e lec he low numbe o occu ence eco ds o T. acaciaelongi oliae in Aus alia, which hus ail o ully ep esen he ealized niche. Gall- o ming insec s, pa icula ly in na i e egions whe e hey coexis wi h hei hos s o en in equilib ium, a e equen ly o e looked and unde s udied o ganisms, despi e he impo ance o se e al species as biocon ol agen s (Dennill, 1988; Ha is and Sho house, 1996) o pes s (Gil-Tape ado e al., 2018; Fe acini e al., 2019). This esul ed in a ela i ely educed da ase in Aus alia, ulne able o spa ial biases, which a e known o a ec niche es ima es (G aham e al., 2008;Sánchez-Fe nández e al., 2011). Rega dless, he e idence o niche conse a ism allowed us o combine he T. acaciaelongi oliae occu ence eco ds o Aus alia and Sou h A ica o he de elopmen o niche models, minimizing he e ec o spa ial biases. The e y high niche o e lap be ween A. longi olia and he gall- o me T. acaciaelongi oliae is consis en wi h he high speci ici y o T. acaciaelongi oliae wi h i s hos plan (Dennill e al., 1993;Ma chan e e al., 2011a). The ealized niche o T. acaciaelongi oliae is ound o be almos ully con ained wi hin he niche o A. longi olia on all egions whe e he wo species co-occu ( e lec ed in high niche s abili y and nea ze o niche expansion). In Sou h A ica, niches o he wo species we e ound o be mo e simila han expec ed gi en he a ailable en i onmen al backg ound and niche un illing was ound o be he lowes among all in e species compa isons. This sugges s a nea o al coloniza ion by T. acaciaelongi oliae F on ie s in Ecology and E olu ion | www. on ie sin.o g 8Oc obe 2020 | Volume 8 | A icle 576667 e o-08-576667 Oc obe 14, 2020 Time: 16:43 # 9 Dinis e al. Niche-Based Biocon ol Managemen o Acacia o all en i onmen s whe e A. longi olia occu s, which has been epo ed by some au ho s (J. Ho mann, pe sonal communica ion). I also indica es ha he a o emen ioned expansion o T. acaciaelongi oliae in Sou h A ica in he absence o subs an ial nega i e bio ic in e ac ions ook place exclusi ely wi hin he ange o A. longi olia, highligh ing he high speci ici y o T. acaciaelongi oliae which is a key ac o in i s eliabili y as a biocon ol agen . Aus alia had he lowes niche o e lap and highes niche un illing among all in e species compa isons, highligh ing ha T. acaciaelongi oliae is no known o occu (o epo ed) in many o he en i onmen s whe e A. longi olia is p esen in i s na i e ange. Again, his likely e lec s non- clima ic in luences hinde ing he abili y o T. acaciaelongi oliae o success ully colonize all A. longi olia habi a s. Implica ions o he Managemen o A. longi olia Including Biocon ol The p edic ed sui able a ea o A. longi olia in he Medi e anean basin a exceeds he cu en dis ibu ion. Besides en i onmen al condi ions, in oduc ion his o ies a e known o g ea ly in luence species dis ibu ions (Donaldson e al., 2014). One o he main pa hways o A. longi olia in oduc ion in Po ugal was o sand binding o coas al dunes (Ma chan e e al., 2003). This coas al in oduc ion, associa ed wi h habi a connec i i y and dispe sal cons ain s, may accoun o he cu en species dis ibu ion which is no known o ex end as a inland as p edic ed by ou model (wi h a ew excep ions in some pa icula loca ions). This hypo hesis seems likely when conside ing ha A. longi olia (sensu la o) is known o occu much u he inland in i s na i e dis ibu ion han in Po ugal (Cou e al., 2020). Likewise, in Sou h A ica, whe e A. longi olia was in oduced also o s abilize inland slopes (Dennill and Donnelly, 1991), he species is p esen as a inland as 500 km, demons a ing ha i can become es ablished a om coas al clima e gi en he adequa e means o dispe sal. Rega dless, we canno dismiss he possibili y ha ac o s no included in he models, such as bio ic in e ac ions o non-clima ic abio ic a iables, may cons ain he dis ibu ion o A. longi olia beyond he model’s p edic ion. The cu en model p edic s isk o A. longi olia in asion in mos o he Medi e anean and, especially, A lan ic coas line o he Medi e anean Basin egion. A eas o cu en A. longi olia in asion in he egion besides Po ugal include no h-wes e n Spain (Communi y o Galicia), which is co ec ly p edic ed as clima ically sui able. A. longi olia is cu en ly p esen , e en i no ye conside ed as in asi e, in se e al coun ies which a e p edic ed as ha ing sui able a eas o he es ablishmen o he species, namely F ance, I aly, G eece and Tu key (Akanil and Middle on, 2010;EFSA Panel on Plan Heal h, 2015). Ou esul s sugges ha pa icula ca e should be aken o p e en A. longi olia in asion in hese egions whe e he species is al eady p esen wi hin an a ea o sui able en i onmen . The model o A. longi olia iden i ied a small a ea o sui abili y in he sou he n Py enees Moun ains, a egion cha ac e ized by mon ane condi ions ha a e uncha ac e is ic o he species. Species dis ibu ion models ha e a limi ed abili y o ex apola e beyond he en i onmen al condi ions a ailable in he aining a ea and may p oduce unexpec ed esul s when en i onmen al homology be ween aining and p ojec ion a eas is no assu ed (Eli h and Lea hwick, 2009). Despi e ou e o s o maximize homology by es ic ing he model p ojec ions o he Medi e anean Basin, moun ain anges in he egion a e likely o include ex eme condi ions which lead o ex apola ion, esul ing in less eliable p edic ions in hese a eas. T ichilogas e acaciaelongi oliae is p edic ed o ind sui able clima e in a signi ican pa o he po en ial A. longi olia dis ibu ion a ea. Cu en a eas o A. longi olia in asion whe e T. acaciaelongi oliae is no p edic ed o ind sui able condi ions a e loca ed in he coas line o no he n Po ugal and Galicia, as well in he mos inland a eas o Po ugal. Ne e heless, he exis ence o a well-es ablished popula ion o T. acaciaelongi oliae in he Po uguese ci y o Coimb a and e y ecen (2019; López-Núñez e al., in p ep.) es ablishmen in no he n a eas o Po ugal (classi ied as unsui able by he model) indica es ha he app oach used in his wo k may no be su icien o ully cha ac e ize he niche o his species. Pa ial inconsis encies amongs modeling ou pu s and ield obse a ions occu in o he s udies (e.g., Gallien e al., 2012;Fischbein e al., 2019) o se e al easons. In ou s udy, ag eemen be ween model p edic ions and known dis ibu ions in he Medi e anean Basin was highe o A. longi olia han o T. acaciaelongi oliae, likely e lec ing he di e ence in numbe and quali y o occu ence eco ds be ween he wo species. Mo eo e , his may be explained by he ac ha he biocon ol agen was only ecen ly (2015) in oduced in Po ugal (Ma chan e e al., 2017), s ill no ha ing enough ime o sp ead and es ablish ac oss he sui able a ea. Fu he mo e, he li e his o y o T. acaciaelongi oliae is cha ac e ized by a nea - o al exis ence as an endopa asi e, and as such he mic oen i onmen al condi ions and in e ac ions wi h he hos plan a e likely o be be e p edic o s o he species’ abili y o o m iable popula ions han mac oclima ic ac o s. Despi e he iden i ica ion o ecogeog aphical a iables ha ha e a limi ing e ec on T. acaciaelongi oliae, he en i onmen al igge s equi ed o comple ion o his species’ li e cycle a e s ill poo ly unde s ood. T. acaciaelongi oliae has shown o be sensi i e o such igge s, as he change in hemisphe e (and consequen empo al misma ch be ween he species’ li e cycle and seasonal queues) upon in oduc ion o Po ugal esul ed in low es ablishmen success in he i s yea s (Ma chan e e al., 2017, López-Núñez e al., in p ep.). Iden i ying hese igge s is undamen al o p ope ly assess he long- e m obus ness o e ec i e biocon ol o A. longi olia media ed by T. acaciaelongi oliae. Ul ima ely, mechanis ic app oaches o modeling which in eg a e mic ohabi a and bio ic in e ac ions be ween hos and pa asi e as well as an explici unde s anding o how hese ac o s con ibu e o he li e cycle o T. acaciaelongi oliae (Palhas e al., in p ep.) a e expec ed o be e app oxima e he undamen al niche and, when in eg a ed wi h co ela i e app oaches such as he one p esen ed in his wo k, allow o mo e accu a e p edic ions. Despi e he limi a ions desc ibed abo e, he high niche o e lap be ween he wo species shows p omise ega ding he e ec i eness o T. acaciaelongi oliae o biocon ol o F on ie s in Ecology and E olu ion | www. on ie sin.o g 9Oc obe 2020 | Volume 8 | A icle 576667