C

CH

a

b

c

U

RA

A

fsaasplmwp

Z

ARfnLijczm

S

h1

Basic and Applied Ecology 17 (2016) 106–114

olony–colony interactions between highly invasive ants

leo Bertelsmeiera,∗, Sébastien Olliera, Amaury Avrila, Olivier Blightb,ervé Jourdanc, Franck Courchampa

Ecologie, Systématique & Evolution, UMR CNRS 8079, Univ. Paris Sud, Orsay Cedex 91405, FranceEstación Biológica de Donana, Consejo Superior de Investigaciones Científicas, 41092 Sevilla, SpainInstitut Méditerranéen de Biodiversité et d’Écologie marine et continentale (IMBE), Aix-Marseille Université,MR CNRS – IRD – UAPV, Centre IRD Nouméa, BP A5, 98848 Nouméa Cedex, New Caledonia

eceived 1 April 2015; accepted 14 September 2015vailable online 21 October 2015

bstract

Among invasive species, ants are a particularly prominent group with enormous impacts on native biodiversity and ecosystemunctioning. Globalization and on-going climate change are likely to increase the rate of ant invasions in the future, leading toimultaneous introductions of several highly invasive species within the same area. Here, we investigate pairwise interactionsmong four highly invasive species, Linepithema humile, Lasius neglectus, Pheidole megacephala and Wasmannia auropunctata,t the whole colony level, using a laboratory set-up. Each colony consisted of 300 workers and one queen. The number ofurviving workers in the competing colonies was recorded daily over 7 weeks. We modelled the survival of each colony duringairwise colony interactions, using a nonlinear model characterizing the survival dynamics of each colony individually. Theeast dominant species was P. megacephala, which always went extinct. Interactions among the three other species showed

ore complex dynamics, rendering the outcome of the interactions less predictable. Overall, W. auropunctata and L. neglectusere the most dominant species. This study shows the importance of scaling up to the colony level in order to gain realism inredicting the outcome of multiple invasions.

usammenfassung

Unter den invasiven Arten sind Ameisen eine ganz besonders auffällige Gruppe mit enormen Auswirkungen auf heimischerten und Ökosystemfunktionen. Die Globalisierung und anhaltender Klimawandel werden wahrscheinlich in der Zukunft dieate an Ameiseninvasionen erhöhen, was zu simultanen Einführungen von mehreren invasiven Arten in der gleichen Region

ühren kann. Hier untersuchen wir paarweise Interaktionen zwischen vier hoch invasiven Arten, Linepithema humile, Lasius

eglectus, Pheidole megacephala und Wasmannia auropunctata, indem wir Konfrontationen zwischen ganzen Kolonien in einemaborversuch durchführen. Jede Kolonie bestand aus 300 Arbeiterinnen und einer Königin. Die überlebenden Arbeiterinnen

n konkurrierenden Kolonien wurden täglich über einen Zeitraum von sieben Wochen gezählt. Wir haben die Überlebensrate

e Kolonie während der paarweisen Interaktionen modelliert, indem wir die Koloniedynamik mit einem nichtlinearen Modelharakterisierten. Die am wenigsten dominante Art war P. megacephala, welche immer ausgestorben ist. Die Interaktionenwischen den restlichen drei Arten zeigten eine komplexere Dynamik, die das Ergebnis der Interaktionen weniger vorhersehbarachte. Insgesamt waren W. auropunctata und L. neglectus die dominantesten Arten. Diese Studie zeigt die Bedeutung von

∗Corresponding author. Present address: Department of Ecology and Evolution, Biophore, Quartier UNIL-Sorge, University of Lausanne, 1015 Lausanne,witzerland. Tel.: +41 21 692 4218; fax: +41 21 692 4165.

E-mail address: cleo.bertelsmeier@gmail.com (C. Bertelsmeier).

ttp://dx.doi.org/10.1016/j.baae.2015.09.005439-1791/© 2015 Gesellschaft für Ökologie. Published by Elsevier GmbH. All rights reserved.

Ev©

Kn

I

mibAlbuHbnae2atSRdi2sSasPt(PCvenbuteiF&fit&i

C. Bertelsmeier et al. / Basic and Applied Ecology 17 (2016) 106–114 107

xperimenten auf dem Kolonie-Niveau, um realistischere Ergebnisse zu bekommen und das Resultat multipler Invasionenorhersagen zu können.

2015 Gesellschaft für Ökologie. Published by Elsevier GmbH. All rights reserved.

eywords: Biological invasions; Invasive ants; Colony behaviour; Interference competition; Colony survival; Linepithema humile; Lasius

it(gps

ccbtttTawiti(aHbrciawno2kcmll

ontuma

eglectus; Pheidole megacephala; Wasmannia auropunctata

ntroduction

On-going globalization and tourism facilitate speciesovements across the world and the rates of new species

ntroductions are exploding (Essl et al., 2011). As a result,iological invasions are increasingly threatening biodiversity.mong invasive species, ants (Formicidae) are a particu-

arly prominent group. Owing to their small size, they cane easily transported by accident on plants, fresh prod-cts, timber, shipment containers or personal items (Suarez,olway, & Ward, 2005). More than 200 ant species haveeen transported by humans and introduced outside theirative range (Suarez, McGlynn, & Tsuitsui, 2010), but therere potentially even more exotic ant species which havestablished outside of their native range (Miravete et al.,014). A subset of these exotic species has become invasivend has enormous impacts on native biodiversity, ecosys-em functioning and animal or human health (Holway, Lach,uarez, Tsutsui, & Case, 2002; Lach & Hooper-Bui, 2010;abitsch, 2011). In addition, many species can invade houses,estroy electrical equipment and impact agriculture, caus-ng high economic losses (Pimentel, Zuniga, & Morrison,005). A total of 19 species has been listed as highly inva-ive by the IUCN invasive species specialist group (IUCNSC Invasive Species Specialist Group, 2012) and 5 speciesre even on the IUCN “100 of the world’s worst inva-ive alien species” list (Lowe, Browne, Boudjelas, & Deoorter, 2000). Very few native species have been shown

o be able to resist the most harmful invasive ant speciese.g. Masciocchi, Farji-Brener, & Sackmann 2009; Blight,rovost, Renucci, Tirard, & Orgeas 2010; Cerdá, Angulo,aut, & Courchamp 2012). Invasive ants are frequentlyery aggressive and behaviourally dominant species (Holwayt al., 2002). Interference competition between invasive andative ant species is relatively well studied, but interactionsetween invasive ant species remain poorly known. It isnclear how two invasive ant species would interact, shouldhey be simultaneously introduced within the same area. Gen-rally, in regions where multiple invasions have occurred,nvasive ants do not co-exist in the same area (Lebrun &eener, 2007; Krushelnycky & Gillespie, 2010; Spicer Rice

Silverman, 2013). A recent modelling study has identi-ed large uninvaded areas that may be suitable for 15 of

he ‘worst’ invasive ants (Bertelsmeier, Luque, Hoffmann, Courchamp, 2015b). The suitable range of several highly

nvasive ants overlaps substantially, creating large potential

isa

nvasion “hotspots”. However, predictive species distribu-ion modelling does not take into account biotic interactionsGuisan & Thuiller, 2005). Therefore it is important to investi-ate whether a single top dominant ant species may ultimatelyrevail, displacing other aggressive, yet less competitive inva-ive species.

Classically, species interactions have been investigated inommunity studies, using pitfall traps to record the numeri-al dominance of different species and with baits to observeehavioural interactions (Cerdá, Arnan, & Retana, 2013). Inhe case of interactions between multiple invasive ant species,his is not possible because they usually do not co-occur inhe same areas yet, although they may interact in the future.o circumvent this difficulty, studies have used behaviouralssays under laboratory conditions, placing one or severalorkers of different species in a petri dish and recording their

nteractions. Several previous studies on behavioural interac-ions have carried out this type of laboratory experiments, yett remains rare in the literature on interference competitionBuczkowski & Bennett, 2008; Blight et al., 2010), especiallymong invasive ants (but see Kirschenbaum & Grace, 2008).owever, these interference experiments in petri dishes areased on single worker interactions and it is unclear whetheresults from these experiments can be extrapolated to theolony level (but see Holway and Suarez, 2004 for an exper-mental colony confrontation between the Argentine ant and

native competing species). For example, the low number oforkers used (either single workers or groups of 10–25) doesot allow species to display interference strategies dependingn a minimum number of workers (Buczkowski & Bennett,008). Further, the presence of resources or a territory isnown to influence behaviours, and thus dominance hierar-hies (Tanner & Adler, 2009). Clearly, experiments underore realistic conditions are needed – at the whole colony

evel in the presence of a queen and over an intermediate toong time span.

Here, we investigate the dominance relationships of fourf the worst invasive ant species (Linepithema humile, Lasiuseglectus, Pheidole megacephala and Wasmannia auropunc-ata, see Appendix A, Table S1 for details on these species),sing dyadic interactions at the colony-level. These speciesight colonize the same areas in the future, according to

recent modelling study (Bertelsmeier et al., 2015b). To

ncrease realism, we provided the species with sufficientpace to avoid confrontations (unlike the small petri dishes)nd we monitored the survival daily over 6 weeks.

1 d Appli

M

Cc

iioTCenopoltmmit

C

W(rnccclpbf(parat(rtwttawfin4

D

I

dBtttaba

C

c

n

wtacto

ttsii

tdfitfoisit

peftw

08 C. Bertelsmeier et al. / Basic an

aterial and methods

olony collection and maintenance of laboratoryolonies

The ants were collected between March and May 2012n New Caledonia (W. auropunctata, P. megacephala) andn Southern France (L. humile and L. neglectus) (for detailsn colony collection, please see Appendix A, Table S2).he interactions were conducted in June and July 2012.olony fragments were maintained in plastic nest contain-rs (55 × 35 × 25 cm) filled with substrate from the originalesting site (soil, wood, leaves) and contained several tubesf water. The boxes were kept at 24 ± 2 ◦C with the appro-riate soil moisture. The ants were fed daily with a variantf the Bhatkar diet (Dussutour & Simpson, 2008) and adibitum during the interactions. All colonies were kept underhese laboratory conditions for at least one month, with little

ortality (<5% in all colonies over the duration of the accli-ation) and produced brood, indicating that all colonies were

n good health before the start of the experiment and no mor-ality was observed due to the laboratory rearing conditions.

olony–colony interactions

We tested pairwise colony interactions among four species,. auropunctata, L. neglectus, L. humile and P. megacephala

i.e., 6 pairwise interactions). Each pairwise interaction waseplicated five times, except the interaction L. humile–L.eglectus, which was replicated 6 times. Each colonyonsisted of 300 workers and one queen. Colonies of P. mega-ephala, which was the only species to possess two workerastes, contained 10% soldiers and 90% minor workers (fol-owing Kirschenbaum & Grace, 2008). Each colony waslaced in a foraging arena (18 × 12 × 7 cm) with a plasterottom and a metal mesh fused to several holes on the top lidor aeration. The foraging arena was connected to a nest tube1 cm in diameter and 10 cm long), which was divided by aiece of cotton into a nesting space (adjacent to the foragingrena) and a water source. The nest tube was covered with aed filter in order not to disturb the ants in the nest but to beble to count individuals in the tube. The foraging arenas ofhe two interacting colonies were connected via a plastic tubesee Appendix A, Fig. S1). After placing the colonies in theirespective arenas, they were allowed to acclimatize and moveo the nesting tubes for 24 h. After that, the connecting tubeas unblocked by removing the obstructing piece of cotton,

hereby allowing the two adjacent colonies to interact. Duringhe first hour after connection, the number of fights in both

renas (incursions) was recorded every 10 min. Afterwards,e monitored the fights every hour over 5 h. Subsequently,ghting was only rarely observed and we recorded daily theumber of surviving individuals of interacting species, over2 days.

v

tac

ed Ecology 17 (2016) 106–114

ata analyses

ncursions

Prior to statistical analysis of the results, we examined allata distributions using the Shapiro Wilk W test for normality.ecause residuals of the data did not conform to a normal dis-

ribution, we used the nonparametric Mann–Whitney test toest for differences in mean incursions in pairs of species andhe Kruskal–Wallis rank sum test to compare mean incursionscross interactions. We then tested the correlation (Spearman)etween the difference in incursions between pairs of speciesnd the difference in mortality between pairs of species.

olony survival

We modelled the survival of each colony during pairwiseolony interactions, using a nonlinear model:

(t) = e−at+log(d−p) + p

here n(t) is the number of surviving workers at time t, d ishe asymptotic value at t0, which is the first point of the curve,

is the steepness of the slope of the function describing theolony decline and p is the asymptotic limit when t tendsowards infinity (i.e., the final colony size after 42 days ofbservation).

Because of a steep decline after the initial fights followinghe connection of the two adjacent arenas, we chose to set0 at day 1, so that d represents the number of workers thaturvived the initial confrontation during day 0, thereby giv-ng a measure of the number of workers lost during the firstnteraction.

In other words, the trend of each colony is summarizedhrough three parameters only: the importance of the initialecline (d), the speed of the subsequent decline (a) and thenal colony size (p). More competitive colonies are expected

o have a high d, a low a and/or a high p. Competitive per-ormance of colonies can be visualised through their valuesf these three parameters and the positions of each colonyn the p–d–a parameter space can be compared. In order toimplify the representation, we prepared the representationn 2D (plane d–p), with the parameter a being represented byhe size of the data point (see an example in Fig. 1B).

Based on estimations of the errors of the estimated modelarameters, confidence envelopes for each model in a param-ter space were calculated and represented in 2D (Fig. 1),ollowing Beale (1960). The parameters are estimated usinghe function nls() in R v.2.15.2, using the algorithm ‘port’hich allows constraining the values of d to 300 and thealues of p to 0.

In the following representations, we decided not to showhe confidence envelopes because they almost never overlapnd render the visual representation of the data unnecessarilyomplex.

C. Bertelsmeier et al. / Basic and Applied Ecology 17 (2016) 106–114 109

Fig. 1. (A) Example of fitted models for the survival of two colonies in replicate 6 of the interaction L. neglectus (LNEG) against L. humile(LHUM). The grey histograms represent the raw data, the colony size at time t (dark L. neglectus, light L. humile) and the two lines correspondto the models fitted for each species (purple L. neglectus, blue L. humile). Although L. neglectus has a lower initial decline, it then declinesfaster, down to extinction. In contrast, L. humile declined more the first day of the interaction, but then declined very slowly and ended upsurviving with two dozen individuals. (B) Representation of the two models in 2D parameter space d–p, with the size of the data pointsrepresenting parameter a. If the points are coloured in red, extinction has occurred. Confidence envelopes of the estimated parameters arer ing to

a subseqd

Etg

R

I

ishlsipWhmdlp

C

4aa

itpFotataatt5

epresented in light grey, with the colour of the contour correspond higher initial surviving size, ends up at zero because of a steep

ecline.

Then, we performed a classification analysis, using theuclidean distance between points in parameter space with

he hclust() function in R. This classification analysis definesroups of colonies with similar survival processes.

esults

ncursions

During colony–colony interactions, the mean number ofncursions into the arena of the opponent differed betweenpecies (Kruskal Wallis test: χ2

(3) = 41.2, p < 0.0001). L.umile made the most incursions and W. auropunctata theeast. Because of the differences in the number of incur-ions, there was a tendency for fights to take place rathern one arena than the other across the following speciesairs: P. megacephala–L. neglectus (χ2

(1) = 52.1, p < 0.0001),. auropunctata–L. neglectus (χ2

(1) = 7.1, p = 0.0078), L.umile–W. auropunctata (χ2

(1) = 6.9, p = 0.0088) and P.egacephala–W. auropunctata (χ2 = 7.0, p = 0.0082). The

(1)ifference in the number of incursions was negatively corre-ated to the difference in final survival (Pearson r2 = 0.521,

< 0.001, n = 31).

hha

the opponent species. This shows clearly that L. neglectus, despiteuent decline, while L. humile survived, thanks to a more gradual

olony survival

The fitted models showing the declining colony size over2 days of the experiment reveal that survival processes varymong species combinations but also (if less drastically)mong replicates of the same species combination (Fig. 2).

When the estimated model parameters are representedn parameter space (Fig. 3), different groups can be dis-inguished. Each replicate number of a particular speciesair appears twice: once for species A, once for species B.or example R4 of the interaction WAUR-PMEG appearsnce within the orange group ‘PMEG-waur’ representinghe colony dynamics of P. megacephala in this replicate. R4ppears again in the green group ‘WAUR-pmeg’ representinghe colony dynamics of W. auropunctata in the same inter-ction experiment. We did not present queen survival in thisnalysis because queens did not participate in the confronta-ions and were almost always the last individual in the colonyo die. In only three colonies, the queen was among the last

individuals.

All replicates of P. megacephala colonies (yellow points)

ave very high d values and p values equalling 0 (relativelyigh initial survival but extinction at the end). Similarly,lmost all colonies of L. humile (blue points) went extinct

110 C. Bertelsmeier et al. / Basic and Applied Ecology 17 (2016) 106–114

Fig. 2. Colony survival (grey bars) and fitted model functions (coloured lines) for all replicates of all pairwise species interactions. Eachr n. LNP

(iisfihthiapvvhudwvcHd

itt1a

tLtvcawtf(ot(Ps

D

maa

ow represents the replicates (R1-R6) of the same species interactioMEG = P. megacephala.

p = 0), except against P. megacephala. This indicates dom-nance of L. humile over P. megacephala only. In thenteractions between L. humile and L. neglectus, both speciesuffer high initial mortality (low d) and have a very lownal colony size (low p), but both parameters are slightlyigher in L. neglectus. Additionally, the mortality parame-ers for L. neglectus faced with P. megacephala are muchigher than those of L. humile against P. megacephala, alsondicating that overall L. neglectus has a slight competitivedvantage over L. humile. Overall, W. auropunctata (greenoints) was the top dominant species with very high sur-ival against P. megacephala and relatively high d and palues against L. humile and L. neglectus. Although a generalierarchy can be established among the species (W. aurop-nctata/L. neglectus > L. humile > P. megacephala), colonyynamics were variable among replicates. In addition, thereas a considerable inter-specific variability in terms of sur-ival processes. In particular, P. megacephala and L. humileolonies typically went extinct by the end of the experiment.owever, P. megacephala resisted much better than L. humileuring the first few days of the confrontation, (higher d).

The classification analysis reveals seven groups with a sim-lar survival process, which do not necessarily correspond to

he same species (Fig. 4). In contrast to groups 1–3, extinc-ion occurs in groups 4–7. Among the other groups, group

has a low p and a low d, i.e. it has a small initial declinend stabilizes at a low level but without going extinct. In

fsow

EG = L. neglectus, LHUM = L. humile, WAUR = W. auropunctata,

his group, three species are represented, W. auropunctata,. humile and L. neglectus. Group 2 (only W. auropunc-

ata and L. neglectus) has also a small p, but a higher dalue, i.e. a good initial resistance and stabilization of theolony size at a low level. Group 3 (only W. auropunctatand L. neglectus) has both a high d and a high p: it resistsell initially and stabilizes at a large colony size. Among

he groups where extinction occurred, there are two dif-erent types of patterns. Either, the initial decline is largelow d) and the extinction occurs more (groups 4 and 6)r less (group 5) rapidly, or the initial decline is low andhe decrease occurs more progressively, down to extinctiongroup 7). Group 7 is also notable because it comprises only. megacephala and almost all interactions concerning thispecies.

iscussion

Connecting foraging arenas of species among four of theost invasive ants, W. auropunctata, L. neglectus, L. humile

nd P. megacephala, resulted in immediate aggressive inter-ctions in all cases. After intense fighting over the first hours

ollowing colony connection, fragments of both coloniesurvived and declined progressively over the 42 days ofbservation. In these experiments, the least dominant speciesas P. megacephala, which always went extinct, although its

C. Bertelsmeier et al. / Basic and Applied Ecology 17 (2016) 106–114 111

Fig. 3. 2D representation of the three estimated model parameters that describe the colony survival after confrontation. d is the initial survivala d the so the moc

itmb(ddtfeaawuawacfoitusona

oIbacctrb

bpaotpehsimc

fter day 0, p is the final survival by the end of the experiment, anf mortality throughout the 42 days; the larger the circle, the fasterircles and the labels indicates the focal species.

nitial mortality on the first day of interaction was compara-ively low. Interactions among the three other species showed

ore complex dynamics. In all three pairwise interactions,oth opponents “won” the interaction in at least one replicatei.e. had a higher final colony size than the opponent). Asifferent replicates of the same interaction followed differentynamics, it would not be representative to simply averagehe survival data across replicates. Instead, models were fittedor each colony individually and then the estimated param-ters were represented in a parameter space, allowing thenalysis of clusters of survival dynamics. This captured moreccurately the central tendency of each species interaction,hile acknowledging for variability. Generally, W. aurop-nctata and L. neglectus were the most dominant speciesnd L. humile tended towards extinction in all interactionsith them. L. neglectus was slightly less competitive than W.uropunctata. Overall, this dominance hierarchy is generallyonsistent with the results found in classical dyadic inter-erence interactions between individuals and small groupsf workers (Bertelsmeier et al., 2015a). Behavioural exper-ments using 1:1 worker interactions have shown that thewo high-ranking species, W. auropunctata and L. neglectus,se mainly chemical defences, while the two lower-rankingpecies use mainly physical defences when paired with one

f the three other species (Bertelsmeier et al., 2015a). It isotable that the most dominant species, W. auropunctata, haslso the smallest body size.

ea(

ize of the coloured disc is proportional to parameter a (the speedrtality; the red dot indicate eventual extinction). The colour of the

Our results demonstrated variable dynamics, rendering theutcome of pairwise species interactions less predictable.nterestingly, a steep decrease in colony size was followedy a slower decrease over several weeks in all interactionsnd replicates. These dynamics observed at the whole-olony level would not be possible to predict based onlassical worker-worker interactions. To our knowledge,his is the first medium-term experiment with invasive antsecording the dynamics of colony interactions on a dailyasis.

The four species varied strongly in their exploratoryehaviour when the adjacent colonies were connected. Inarticular P. megacephala and L. humile invaded rapidly therena of the opponent species, leading to a high proportionf fights occurring on the territory of the opponent. In con-rary, W. auropunctata and L. neglectus were slower, moreassive explorers. Interestingly, the species with the high-st tendency to invade the opponent’s territory suffered theighest mortality. This suggests that the two species withtrong exploratory behaviour were less competitive throughnterference. This strong exploratory behaviour, however,

ight however confer an advantage through exploitationompetition.

In native ant communities, one factor promoting co-

xistence between species is a trade-off between the species’bility to discover and their ability to dominate resourcesLebrun & Feener, 2007; Parr & Gibb, 2012; Cerdá et al.,

112 C. Bertelsmeier et al. / Basic and Applied Ecology 17 (2016) 106–114

F nated 7p cies fori ean pa

2bctimgat

guF

ct&wetintT

ig. 4. The classification analysis of the model parameters discrimilus e for extinction) are represented next to each replicate. The spen lower case. R1 to R6 indicate the identity of the replicate. The m

013). However, it has been suggested that invasive ants canreak this discovery-dominance trade-off by excelling at dis-overing and dominating resources (Holway, 1999). This ishought to be an important mechanism leading to their highmpacts on native communities by competitively displacing

any species (Holway, 1999). Future research may investi-ate whether there is such a trade-off among invasive antsnd whether all of them generally break the trade-off relativeo native species.

Although experiments at the whole colony level offer

reater realism than behavioural observations over a few min-tes in petri dishes, they still suffer from certain limitations.or example, the degree of resource dispersion and habitat

dBw

groups with similar survival processes. The parameters (p, a and d, which the model has been fitted is written in capitals, its opponentttern of each group of time series is shown.

omplexity (Sarty, Abbott, & Lester, 2006), as well as habitatype or disturbance can modify dominance hierarchies (Parr

Gibb, 2010; Cerdá et al., 2013). In addition, the spacehere adjacent colonies could interact was constrained by the

xperimental set-up and colonies may not fight so intensely ifhey were located a few meters apart. However, previous stud-es have shown that invasive ants can even have impacts onative species at a biogeographic scale, creating a mosaic dis-ribution without co-existing locally (Gotelli & Arnett, 2000).his type of mosaic distribution can also be observed among

ominant native ant species (Ribas & Schoereder, 2002; Lereton 2003). Therefore it is likely, that even if the nestsere spaced further apart than in our laboratory experiment,

d Appli

tc

caSadncosFdmtfizrcbeei(ca(ctttaevlof

stitpshdMCraYtdwa

cw

io2Bopuc

A

2B

A

fj

R

B

B

B

B

B

B

B

C

C. Bertelsmeier et al. / Basic an

he species would not co-exist closely, within the same localommunity.

Further, different climatic suitability of a certain areaan favour one species over the other, giving a competitivedvantage to one species over another (ex. Spicer Rice &ilverman, 2013; Barbieri, Grangier, & Lester, 2015). Such

dominance shift is especially likely in species pairs thato not differ much in their capacity to dominate (e.g., L.eglectus and L. humile). Additionally, biotic interactionsan alter dominance hierarchies, for example the presencef phorid flies can compromise the competitive abilities ofome ant species but not others (Lebrun & Feener, 2007).ield experiments could be useful to determine the relativeominance of entire colonies under a given set of environ-ental conditions, using secure, artificial nests for example

o guarantee biosecurity (Luque et al., submitted). So far,eld-based observations have only been possible at contactones of two invasion fronts. For example, observations fromanges where two invasive ant species come in contact showompetitive exclusion and displacement of one invasive anty another at a local scale. Examples include the mutualxclusion of L. humile and Solenopsis invicta in the South-rn US (LeBrun et al., 2007), L. humile and P. megacephalan Hawaii (Krushelnycky & Gillespie, 2010) and BermudaLieberburg, Kranz, & Seip, 1975), L. humile and Pachy-ondyla chinensis in the US (Spicer Rice & Silverman, 2013)nd W. auropunctata and P. megacephala in New CaledoniaChazeau et al. 2000, Le Breton 2003). Studies investigatingompetitive relationships in the field have the advantage thathey take place under more realistic environmental conditionshan lab-based studies, yet, they also have the disadvantagehat they do not allow following the number of individu-ls at a daily basis over several weeks, which we did here,nabling us to precisely characterize the processes of sur-ival and decline in the competing colonies. Evidence fromab experiments and field observations complement eachther nicely and we encourage further studies at invasionronts.

Another interesting question is how different initial colonyizes would impact the outcome of colony–colony competi-ion. Is there a certain ratio at which dominance would benversed or a clear winner can be predicted in cases wherehe hierarchy was not clear-cut at equal group size? For exam-le, established supercolonies of L. humile may outcompetemaller incipient L. neglectus colonies in the field, if theyave a numerical advantage. Further, the effect of a ‘resi-ent’ colony versus an arriving colony might be investigated.odels of habitat relative suitability (Bertelsmeier, Luque, &ourchamp, 2013; Bertelsmeier et al., 2015b) based on cor-

elative predictions of suitable areas cannot take into account ‘resident’ effect, i.e. sequential arrival of invasive species.et, it seems appropriate to investigate competition between

wo colonies among the four invasive species, which all repro-uce by budding (i.e. dispersal of a newly mated queen andorkers on foot), extending and covering slowly the entire

rea (Rabitsch, 2011). In this way, small newly established

C

ed Ecology 17 (2016) 106–114 113

olonies of different species can come in contact and competeith each other.Predicting the winner of a symmetrical colony–colony

nteraction is not as straightforward as predicting the winnerf a worker-worker interaction (e.g. Buczkowski & Bennett,008; Kirschenbaum & Grace, 2008; Blight et al., 2010;ertelsmeier et al., 2015a). This study shows the importancef scaling up to the colony level in order to gain realism inredicting the outcome of multiple invasions and in betternderstanding interspecific competition as the driver of antommunity structure.

cknowledgements

This paper was supported by Région Ile-de-France (03-010/GV-DIM ASTREA), ANR (2009 PEXT 010 01) andiodivERsa Eranet grants, FFII Project.

ppendix A. Supplementary data

Supplementary data associated with this article can beound, in the online version, at http://dx.doi.org/10.1016/.baae.2015.09.005.

eferences

arbieri, R. F., Grangier, J., & Lester, P. J. (2015). Synergistic effectsof temperature, diet and colony size on the competitive abilityof two ant species. Austral Ecology, 40, 90–99.

eale, E. (1960). Confidence regions in non-linear estimation. Jour-nal of the Royal Statistical Society, 22B, 41–88.

ertelsmeier, C., Avril, A., Blight, O., Confais, A., Diez, L., Jourdan,H., et al. (2015). Different behavioural strategies among sevenhighly invasive ant species. Biological Invasions, 17, 2491–2503.

ertelsmeier, C., Luque, G. M., & Courchamp, F. (2013). Increasein quantity and quality of suitable areas for invasive species asclimate changes. Conservation Biology, 27(6), 1458–1467.

ertelsmeier, C., Luque, G. M., Hoffmann, B. D., & Courchamp,F. (2015). Worldwide ant invasions under climate change. Bio-diversity and Conservation, 24, 117–128.

light, O., Provost, E., Renucci, M., Tirard, A., & Orgeas, J. (2010).A native ant armed to limit the spread of the Argentine ant.Biological Invasions, 12(11), 3785–3793.

uczkowski, G., & Bennett, G. W. (2008). Aggressive interactionsbetween the introduced Argentine ant, Linepithema humile andthe native odorous house ant, Tapinoma sessile. Biological Inva-sions, 10(7), 1001–1011.

erdá, X., Angulo, E., Caut, S., & Courchamp, F. (2012). Antcommunity structure on a small Pacific island: Only one nativespecies living with the invaders. Biological Invasions, 14(2),

323–339.

erdá, X., Arnan, X., & Retana, J. (2013). Is competition a sig-nificant hallmark of ant (Hymenoptera: Formicidae) ecology?Myrmecological News, 18, 131–147.

1 d Appli

C

D

E

G

G

H

H

H

I

K

K

L

L

L

L

L

L

M

M

P

P

P

R

R

S

S

S

S

14 C. Bertelsmeier et al. / Basic an

hazeau, J., Potiaroa, T., Bonnet de Larbogne, L., Konghouleux,J., & Jourdan. (2000). La “fourmi électrique” Wasmanniaauropunctata (Roger) en Nouvelle-Calédonie: Expression del’invasion, moyens d’une maî trise de la nuisance en milieuagricole, praticabilité d’une préservation des milieux naturels.Convention Territoire de Nouvelle-Calédonie/IRD, 1999, 10–63.

ussutour, a., & Simpson, S. J. (2008). Description of a simplesynthetic diet for studying nutritional responses in ants. InsectesSociaux, 55(3), 329–333.

ssl, F., Dullinger, S., Rabitsch, W., Hulme, P. E., Hülber, K.,Jarosík, V., et al. (2011). Socioeconomic legacy yields an inva-sion debt. Proceedings of the National Academy of Sciences ofthe United States of America, 108(1), 203–207.

otelli, N., & Arnett, A. (2000). Biogeographic effects of red fireant invasion. Ecology Letters, 3(4), 257–261.

uisan, A., & Thuiller, W. (2005). Predicting species distribution:Offering more than simple habitat models. Ecology Letters, 8(9),993–1009.

olway, D. (1999). Competitive mechanisms underlying the dis-placement of native ants by the invasive Argentine ant. Ecology,80(1), 238–251.

olway, D. A., & Suarez, A. V. (2004). Colony-structure variationand interspecific competitive ability in the invasive Argentineant. Oecologia, 138(2), 216–222.

olway, D., Lach, L., Suarez, A., Tsutsui, V., & Case, N. D. T. J.(2002). The causes and consequences of ant invasions. AnnualReview of Ecology and Systematics, 33(1), 181–233.

UCN SSC Invasive Species Specialist Group. (2012). Global inva-sive species database. http://www.issg.org/database/welcome/

irschenbaum, R., & Grace, J. K. (2008). Agonistic responsesof the tramp ants Anoplolepis gracilipes, Pheidole mega-cephala, Linepithema humile, and Wasmannia auropunc-tata (Hymenoptera: Formicidae). Sociobiology, 51(3), 673–683.

rushelnycky, P. D., & Gillespie, R. G. (2010). Correlates of vul-nerability among arthropod species threatened by invasive ants.Biodiversity and Conservation, 19(7), 1971–1988.

ach, L., & Hooper-Bui, L. M. (2010). Consequences of ant inva-sions. In L. Lach, C. L. Parr, & K. L. Abbott (Eds.), Ant ecology(pp. 261–286). Oxford: Oxford University Press.

e Breton, J. (2003). Etude des interactions entre la fourmi Was-mannia auropunctata et la myrmécofaune: comparaison d’unesituation en zone d’introduction: la Nouvelle-Calédonie et d’unesituation en zone d’origine: la Guyane Francaise (PhD thesis).Université Paul Sabatier Toulouse 3.

ebrun, E. G., & Feener, D. H. (2007). When trade-offs inter-act: Balance of terror enforces dominance discovery trade-offin a local ant assemblage. Journal of Animal Ecology, 76(1),58–64.

T

Available online at www.s

ScienceD

ed Ecology 17 (2016) 106–114

eBrun, E. G., Tillberg, C., Suarez, V. A., Folgarait, V. P. J., Smith,C. R., & Holway, D. A. (2007). An experimental study of compe-tition between fire ants and Argentine ants in their native range.Ecology, 88(1), 63–75.

ieberburg, I., Kranz, P., & Seip, A. (1975). Bermudian ants revis-ited: The status and interaction of Pheidole Megacephala andIridomyrmex Humilis. Ecology, 56, 473–478.

owe, S., Browne, M., Boudjelas, S., & De Poorter, M. (2000). 100of the world’s worst invasive alien species – A selection from theglobal invasive species database.

asciocchi, M., Farji-Brener, A. G., & Sackmann, P. (2009). Com-petition for food between the exotic wasp Vespula germanicaand the native ant assemblage of NW Patagonia: Evidence ofbiotic resistance? Biological Invasions, 12(3), 625–631.

iravete, V., Roura-Pascual, N., Dunn, R. R., & Gomez, C.(2014). How many and which ant species are being acciden-tally moved around the world? Biology Letters, 10, 20140518.http://dx.doi.org/10.1098/rsbl.2014.0518

arr, C., & Gibb, H. (2010). Competition and the role of dominantants. In L. Lach, C. Parr, & K. Abbott (Eds.), Ant ecology (pp.77–96). Oxford: Oxford University Press.

arr, C., & Gibb, H. (2012). The discovery-dominance trade-off isthe exception rather than the rule. Journal of Animal Ecology,81, 233–241.

imentel, D., Zuniga, R., & Morrison, D. (2005). Update on theenvironmental and economic costs associated with alien-invasivespecies in the United States. Ecological Economics, 52(3),273–288.

abitsch, W. (2011). The Hitchhiker’s guide to alien ant invasions.BioControl, 56(4), 551–572.

ibas, C., & Schoereder, J. (2002). Are all ant mosaics caused bycompetition? Oecologia, 131(4), 606–611.

picer Rice, E., & Silverman, J. (2013). Propagule pressure andclimate contribute to the displacement of Linepithema humileby Pachycondyla chinensis. PLoS ONE, 8(2), e56281.

arty, M., Abbott, K. L., & Lester, P. J. (2006). Habitat complexityfacilitates coexistence in a tropical ant community. Oecologia,149(3), 465–473.

uarez, A. V., McGlynn, T. P., & Tsuitsui, N. D. (2010). Biogeo-graphic and taxonomic patterns of introduced ants. In L. Lach,C. L. Parr, & K. L. Abbott (Eds.), Ant ecology (pp. 233–244).New York, NY: Oxford University Press.

uarez, A. V., Holway, D. A., & Ward, P. S. (2005). The role ofopportunity in the unintentional introduction of nonnative ants.Proceedings of the National Academy of Sciences of the UnitedStates of America, 102(47).

anner, C. J., & Adler, F. R. (2009). To fight or not to fight: Context-dependent interspecific aggression in competing ants. AnimalBehaviour, 77(2), 297–305.

ciencedirect.com

irect