Animal biology / Biologie animale

A new case of trophobiosis between ants andHeteroptera

Alain Dejeana*, Marc Gibernaub, Thierry Bourgoinc

a Laboratoire d’écologie terrestre (UMR CNRS n° 5552), université Toulouse-III, 118, route de Narbonne, 31062Toulouse cedex, Franceb Jardin botanique de Montréal, 4101, rue Sherbrooke Est, Montréal PQ H1X 2B2, Canadac Laboratoire d’entomologie (ESA CNRS n° 8043), Muséum national d’histoire naturelle, 45, rue Buffon, 75005Paris, FranceReceived 13 October 1999; accepted 6 March 2000

Communicated by Claude Combes

Abstract – Despite the large specific diversity of equatorial rainforests, Caternaultiellarugosa (Heteroptera; Plataspidae) was only noted on two euphorbiaceous trees (Brideliamicrantha and B. grandis) and was attended by two ant species, Camponotus brutus(Formicinae) and Myrmicaria opaciventris (Myrmicinae). We recorded semiochemical(attending workers palpated the dorsal abdominal glands of the nymphs) and semanticsignals (nymphs ready to excrete honeydew raised their bodies; they alternated theextrusion and withdrawal of the first honeydew droplet when the workers did notimmediately absorb the honeydew). Cat. rugosa was recorded in carton pavilions builtby the ants at the base of the tree trunks. During proliferations of the population, clustersof nymphs and adults developed outside pavilions. In the latter case, M. opaciventrisworkers did not modify their rhythm of activity in order to attend these clusters, whileCamp. brutus workers, normally nocturnal, attended them day and night. © 2000Académie des sciences/Éditions scientifiques et médicales Elsevier SAS

trophobiosis / ants / Heteroptera / Plataspidae / rhythm of activity / territoriality

Résumé – Un nouveau cas de trophobiose entre fourmis et hétérop-tères. Malgré la grande diversité spécifique de la forêt équatoriale, Caternaultiellarugosa, un hétéroptère Plataspidae, n’est associé qu’à deux Euphorbiaceae (Brideliamicrantha et B. grandis) et est exploité pour son miellat par deux espèces de fourmis,Camponotus brutus (Formicinae) et Myrmicaria opaciventris (Myrmicinae). Dans cetterelation entrent en jeu des signaux sémiochimiques (les fourmis palpent des zonesglandulaires du plataspide) et sémantiques (les larves du plataspide relèvent leur corpsavant la sécrétion de miellat ; la première goutte peut être réabsorbée à plusieursreprises si la fourmi ne réagit pas immédiatement). Ces plataspides sont généralementinstallés dans des abris en carton construits par les fourmis à la base des troncs.Toutefois, lors de proliférations, des groupes de larves et d’adultes se développent endehors des abris. Dans ce cas, les M. opaciventris ne modifient pas leur rythme d’activitépour les exploiter, alors que les C. brutus, normalement nocturnes, les exploitent jour etnuit. © 2000 Académie des sciences/Éditions scientifiques et médicales Elsevier SAS

trophobiose / fourmis / Heteroptera / Plataspidae / rythme d’activité /territorialité

* Correspondence and reprints: dejean@cict.fr

447

C.R. Acad. Sci. Paris, Sciences de la vie / Life Sciences 323 (2000) 447–454© 2000 Académie des sciences/Éditions scientifiques et médicales Elsevier SAS. Tous droits réservésS0764446900001566/FLA

Version abrégée

La trophobiose est une relation symbiotique entredes fourmis et d’autres insectes, appelés trophobiontes.Les fourmis qui obtiennent une ressource alimentaireprotègent ces derniers contre leurs ennemis. Bien quela grande majorité des cas concerne des hémiptères del’ancien sous-ordre des homoptères (Sternorrhyncha,Cicadomorpha et Fulgoromorpha) et des lépidoptèreslycénides, ce type de relation existe aussi pour deshétéroptères, mais il n’a été décrit que dans de trèsrares cas à partir d’observations de terrain limitées àl’Asie. Ces observations avaient permis de prédire qu’ils’agissait de relations élaborées, vraisemblablementplus répandues qu’on n’avait pensé initialement car 1o

dans certains cas les fourmis ne se contentaient pasd’exploiter le miellat de ces insectes de manière oppor-tuniste, mais les protégeaient en construisant des abrisde carton, et 2o plusieurs familles d’hétéroptères (sur-tout des plataspides) sont impliquées. La rareté des casétait donc le facteur limitant nos connaissances.

Au cours des enquêtes écologiques que nous avonsmenées au sud du Cameroun, portant sur plus de20 000 arbres, la canopée de plus de 350 d’entre euxayant été inspectée, nous avons observé des cas derelations trophobiotiques entre Caternaultiella rugosa,hétéroptère phytophage Plataspidae, et seulement deuxespèces de fourmis, Camponotus brutus (Formicinae)et Myrmicaria opaciventris (Myrmicinae). Malgré lagrande diversité des plantes dans cette région, cesassociations étaient localisées dans des abris de cartonsitués à la base des troncs de deux espèces d’euphor-biacées, Bridelia micrantha et B. grandis. La rareté deces associations résulte probablement de la conjugai-son entre spécificité plante–hétéroptère et spécificitéhétéroptère–fourmi.

À chaque observation sur le terrrain nous avonsrapporté à Yaoundé des nymphes et des adultes deCat. rugosa. Ils ont été installés sur deux sites où setrouvaient de nombreux arbustes de B. micrantha ainsique Camp. brutus et M. opaciventris. Les fourmis lesont acceptés, ont construit des abris de carton pour lesélever, puis les populations de plataspides se sont sibien développées qu’une partie d’entre elles a proliféréen dehors de ces abris, toujours à la base des troncs.Cette situation a permis d’étudier le comportement desfourmis et des plataspides.

Chaque ouvrière de Camp. brutus, de grande taille,peut prendre soin de cinq à dix-huit larves des pre-miers stades, balayant ces larves de l’extrémité de sesantennes. Ces dernières répondent aux contacts anten-naires en soulevant leur corps puis en excrétant leurmiellat. Les larves des derniers stades, exploitées seu-les, par deux ou par trois, font l’objet de longuespalpations antennaires sur des aires glandulaires destergites abdominaux. Au moment d’excréter leur miel-

lat ces larves basculent leur corps, puis les ouvrièrespalpent l’apex de leur abdomen pendant qu’elles absor-bent le miellat. Quand une ouvrière ne réagit pasimmédiatement au début de l’excrétion du miellat, leslarves produisent et réabsorbent la première goutte-lette à plusieurs reprises, ce signal alertant l’ouvrière.Les relations avec les adultes dont les glandes dorsalessont atrophiées semblent plus opportunistes car lesouvrières attendent qu’ils émettent leur miellat sponta-nément. Les ouvrières de M. opaciventris, de pluspetite taille, se groupent pour exploiter les larves duplataspide, quel que soit leur stade. Certaines palpentleur abdomen alors que d’autres attendent l’excrétiondu miellat. Quand celle-ci se produit, l’abondance esttelle que seules certaines gouttelettes sont récupéréesdirectement. La plupart des autres le sont après qu’ellesaient adhéré au corps des ouvrières par le jeu desforces de tension superficielle.

L’ouverture des abris a permis de noter qu’adultes etlarves du plataspide y sont élevés par les deux espècesde fourmis. La vie dans les abris de Camp. brutus a puêtre observée en perçant des fenêtres dans leur paroi eten fixant une feuille de matière plastique transparenteavec des épingles, cette dernière étant recouverte d’unfilm opaque pouvant pivoter. Après plusieurs jours,quand on fait pivoter ce dernier, les ouvrières sont trèspeu perturbées. On peut les observer nettoyer ces abrisen permanence, ce qui évite le développement demoisissures préjudiciable à la survie du trophobionte.Chez cette fourmi polymorphe les ouvrières major sontdes gardiennes qui se tiennent à l’entrée des abris, alorsque les média et quelques minor exploitent les platas-pides et ont leur gastre distendu par le miellat. Grâce àdes marquages avec de la peinture on a pu noter quecertaines media se spécialisent dans le transport dumiellat, depuis les abris jusqu’à la colonie.

À partir des groupes du plataspide qui se dévelop-pent hors des abris, lors des périodes de prolifération,nous avons étudié l’effet de leur présence sur le rythmed’activité de M. opaciventris et Camp. brutus (dix B. mi-crantha pour chaque espèce de fourmi). Nous avonsrelevé le nombre d’ouvrières exploitant des stictococ-cides (Sternorrhyncha) sur une branche de chaquearbre (lot témoin) et celui d’ouvrières exploitant lesplataspides hors des abris (lot expérimental), de jour(entre 11 h et midi) et de nuit (entre 22 et 23 h).L’expérience a été répétée deux fois, correspondant àvingt cas pour chaque espèce de fourmi. Les M. opa-civentris, toujours présentes, étaient significativementplus fréquentes de jour que de nuit, qu’elles exploitentles stictococcides ou les plataspides. Il s’agit là de leurrythme classiquement connu, qui ne subit donc pas devariation en présence des plataspides. Comme le nom-bre d’ouvrières de M. opaciventris baisse fortementdurant la nuit, on a vu des Camp. brutus venir alors

448

A. Dejean et al. / C.R. Acad. Sci. Paris, Sciences de la vie / Life Sciences 323 (2000) 447–454

exploiter certains groupes de plataspides laissés libres.Les Camp. brutus, déjà connues comme étant noctur-nes, exploitaient les stictococcides de nuit et les platas-pides nuit et jour, montrant ainsi leur intérêt pour cesderniers. Durant le jour, les stictococcides sont exploi-tés par des ouvrières de Crematogaster spp. et Camp.chrysurus. Si elles s’approchent des plataspides, cesdernières fuient face à la réaction des Camp. brutus,bien plus grandes qu’elles.

En conclusion, nous avons bien montré qu’il s’agitd’une relation mutualiste symbiotique, les fourmis rece-

vant du miellat, les plataspides étant protégés dans desabris. Cette relation a atteint un niveau élaboré, commechez certains Sternorrhyncha, Cicadomorpha ou Fulgo-romorpha car interviennent, outre la construction desabris et leur nettoyage permanent, une organisationsociale à l’intérieur de ces derniers, des signaux séman-tiques et sémiochimiques synchronisant la prise demiellat, et une spécialisation poussée de la part desCamp. brutus qui modifient leur rythme d’activité lorsde l’exploitation de ce plataspide en dehors des abris.

1. Introduction

Trophobiosis is a symbiotic relationship between antsand attended insects or trophobionts. The ants obtainhoneydew from the trophobionts and in turn protect themfrom their natural enemies, often by building protectiveshelters or pavilions [1, 2]. Although most of the knowncases concern Hemiptera of the ancient sub-orderHomoptera (i.e. Sternorrhyncha, Cicadomorpha andFulgoromorpha) or the Lepidoptera of the family Lycae-nidae, trophobiotic relationships between ants and certainHeteroptera do exist, but have rarely been reported. Anassociation between a myrmicine ant and a Plataspidaewas described in Sri Lanka at the beginning of the century[3]. More recently, cases of ants attending sap-suckingHeteroptera, including Plataspidae sheltered in pavilions,have been recorded in southern Asia [4, 5, 6].

Until now, knowledge of ant–heteropteran associationshas stemmed from field observations and not from detailedexperimental studies. After observing one case of antsattending Caternaultiella rugosa (Heteroptera: Plataspi-dae) in Cameroon, we decided to examine if the rarity ofthis kind of association is due to the relative specificity ofthe insect vis-à-vis its host plants and attending ants. Wethus introduced groups of this bug into easily accessibleareas in Yaoundé in order to observe this association indetail. We also conducted experiments on the ant’s rhythmof activity, as we hypothesized that if an ant species is welladapted to attending this bug species it could modify itsrhythm of activity in order to defend day and night clustersof the bug that develop outside pavilions.

2. Materials and methods

2.1. Research on plant and ant speciesassociated with Cat. rugosa

This study was undertaken in southern Cameroon fromDecember 1987 to July 1993. During this period, datawere recorded during 270 outings in the field by two to sixresearchers (total number of trees inspected greater than20 000). The trunks and bases of the trees were carefullyexamined. Other surveys were conducted in tree canopies

(fallen trees, trees cut by farmers, and during the ‘Opéra-tion canopée 91’ using the canopy raft and the canopysled [7]) resulting in a total of more than 350 tree crownsinspected.

Each time we encountered Cat. rugosa individuals, wenoted: a) the host plant species; b) the area of the treewhere they sucked sap (branches; trunk; root area); c)whether or not they were attended by ants; and d) the antspecies and the presence or absence of rough cartonpavilions sheltering the bugs. Plataspids and ants wereidentified on site and voucher specimens were sent to theMuseum of Natural History, London.

In order to conduct experimental studies we transportedgroups of both adults and nymphs of the bug to theUniversity campus and Mvolier Valley, Yaoundé, Cam-eroon, two sites where one of the host plants (Brideliamicrantha; Euphorbiaceae) and the two species of ants(Myrmicaria opaciventris and Camponotus brutus) associ-ated with Cat. rugosa were previously noted. In both casesthe bugs were accepted and tended by the ants.

2.2. Behavioural characteristics of the trophobionts

Observations were conducted on two groups of eightB. micrantha occupied by Camp. brutus or M. opaciven-tris. We noted which area of the bug’s body workerspalpated in order to obtain honeydew, and the reactions ofthe bug when a worker was not ready to imbibe thehoneydew that it had begun to extrude.

In order to observe life in the pavilions, we destroyed infour cases a part of the carton wall and replaced it with aplate of transparent plastic attached with pins. The trans-parent plate formed a window that we covered with ablack plastic plate that pivoted. The ants obstructed thesmall spaces remaining uncovered between the branchesand the transparent plate. They did not seem to be dis-turbed when we pivoted the black cover for observationsor in order to take photographs.

2.3. Reaction of ants to competitors

During studies on the workers’ rhythm of activity whenattending Cat. rugosa outside the pavilions, we noted thenumber of workers attending two to three clusters of bugs

449

A. Dejean et al. / C.R. Acad. Sci. Paris, Sciences de la vie / Life Sciences 323 (2000) 447–454

per tree diurnally (between 11:00 and 12:00 hours), thenat night (between 22:00 and 23:00 hours) on ten treesoccupied by Camp. brutus. A control group was obtainedby sampling workers attending stictococcids (mostly Stic-tococcus formicarius) on one branch of the same trees.This operation was repeated twice with a 2-week interval,resulting in a total of 20 cases. It permitted paired tests aswe strictly compared the same zones of each tree duringthe day and the following night. We conducted the samesampling procedure on ten other trees occupied by M. opa-civentris. Also, direct observation allowed us to note thereactions of the attending ants when competing workersapproached. Statistical comparisons were made using Stu-dent’s t-tests and Fisher’s exact tests (Statistica 5.0 soft-ware).

3. Results

3.1. Relationships between Cat. rugosa, plants and ants

Cat. rugosa, which is absent from the forest canopiesand the understories, was noted along the forest edgesonly on B. micrantha (189 small trees out of 1 450; thisspecies represented 1–4 % of the plants along the forestedges [7, 8]) and on B. grandis (27 trees out of 178). Wenever noted Cat. rugosa in the savannah, although B. fer-ruginea was frequent (9.3 % of the shrubs; N = 450 shrubs[9]). The colonies and isolated individuals were generallyinstalled in the root area, and during periods of prolifera-tion, at the base of the trunk up to a height of 1.5 m. Wenever recorded Cat. rugosa on young branches or at theextremity of the branches.

Aside from anecdotal observations of a few workers ofCamp. chrysurus who occasionally solicited the nymphsof Cat. rugosa on one Bridelia, all associations (N = 216cases) concerned two ant species, mainly Camp. brutus(82.4 % of the cases) and secondarily M. opaciventris(17.6 % of the cases). Both Camp. brutus and M. opa-civentris always sheltered Cat. rugosa colonies in pavil-ions built with small pieces of wood, bark and dry frag-ments of leaves (and earth in the case of M. opaciventris).

Cat. rugosa proliferated in the areas where it was intro-duced, with a part of its population developing outside thepavilions. This permitted us to study the attending ants’rhythms of activity.

3.2. The solicitation of honeydew

Clusters of five to 18 first-instar nymphs of Cat. rugosawere attended by workers that moved their antennaeabove them in a wide, sweeping movement. Periodically,responding to the stimulation of the antennae, a nymphraised up its body (quasi-perpendicular to the plant sur-face) and a droplet of honeydew was excreted from theextremity of the anus (figure 1). Workers attended one tothree last-instar nymphs, palpating the pleural regions ofthe abdomen (three pairs of dorsal abdominal glands)without needing to move. Once the nymphs prepared to

secrete honeydew, they raised their bodies (to about a 45°angle). In response, the ants folded their antennae backand palpated the apex of the bug’s abdomen with feverishmovements, all the while absorbing the excreted droplets.If a worker was not in the proper position when theirhoneydew was excreted, the nymphs kept their bodiesraised up, and repeated the alternate extrusion and with-drawal of the first droplet, these signals immediately attract-ing the ant. Workers placed themselves behind the adults,did not antennate them (or did so only rarely), awaiting aspontaneous secretion.

Smaller than Camp. brutus, workers of M. opaciventrisgrouped themselves to exploit the clusters of first-instarnymphs as well as grouped or isolated last-instar nymphs(figure 1). One to three workers waited for honeydewsecretion, while others moved to the nymphs’ sides oreven climbed onto their bodies in order to palpate dorsalglands. When excreted, some of the droplets were directlyingested by the workers while others that adhered to theirbodies thanks to surface tension were licked later. Duringattendance of the imagos, several workers placed them-selves behind the bugs, waiting for honeydew secretion.They also used surface tension in order to recuperate mostof the droplets.

3.3. Life in the pavilions

The pavilions were installed in the root area or at thebase of the trunk, under trunk zones where the dry bark isloose or between the trunk and suckers. When observersopened the carton walls of the pavilions for the installationof the transparent plastic windows or for experiments,aggressive workers, mandibles wide open, appeared.

In the pavilions, both ant species sheltered and attendedadults, nymphs of different instars and egg masses. Thanksto the transparent plastic windows we noted that theworkers immediately licked the honeydew fallen on thesubstrate, inhibiting the growth of mildew. In Camp. bru-tus, a polymorphic species, the majors acted as guards andremained immobile near the entrance of the pavilions andrarely had a distended gaster, while the media and the fewminors present attended the bugs and had gasters dis-tended with honeydew. We also noted, by marking antswith a spot of enamel paint, that certain media werespecialized only in transporting the honeydew betweenthe pavilions and the nest.

3.4. Ants’ rhythms of activity

M. opaciventris workers were significantly more fre-quent during the day than at night both on the brancheswhile attending stictococcids and at the base of the trunk,around the clusters of Cat. rugosa (figure 2). The coloniesof M. opaciventris, which are very numerous, did nottolerate competing ants on the Bridelia where they tendedCat. rugosa diurnally. At night, the number of attendingworkers decreased by half, but during our study bothstictococcids and clusters of Cat. rugosa remainedattended by workers. Nevertheless, during further obser-

450

A. Dejean et al. / C.R. Acad. Sci. Paris, Sciences de la vie / Life Sciences 323 (2000) 447–454

vations, we noted that certain clusters of bugs were aban-doned at night and were sometimes attended by workersof Camp. brutus.

Camp. brutus, known to be nocturnal [10], attendedstictococcids at night, with only a few incursions of soli-tary workers on the branches during the day. Nevertheless,they attended the clusters of Cat. rugosa situated outsidethe pavilions both day and night, but the number ofworkers around the clusters of the bug was significantlyhigher at night (figure 2). A localized change in the rhythmof activity around clusters of Cat. rugosa is demonstratedwhen comparing worker presence on the branches, attend-ing stictococcids, with worker presence around clusters ofCat. rugosa: noted in four and 18 cases out of 20, respec-tively (Fisher exact test: P < 0.000 1).

During the day the stictococcids were attended byworkers of Camp. chrysurus (one tree); Crematogasterstriatula (four trees); or both Camp. chrysurus and Cre-matogaster spp. (five trees). Camp. brutus workers wereaggressive toward Camp. chrysurus and Crematogasterspp. workers that approached their attended groups ofCat. rugosa. These ants, smaller in size, generally fled fromCamp. brutus. Nevertheless, Camp. chrysurus workers

repeatedly approached the bugs, triggering aggressiveattacks on the part of Camp. brutus which pursued theimpertinent intruders until they dropped onto the groundto escape. At night we never noted competing ant specieson the trees where the Camp. brutus workers attendedCat. rugosa and stictococcids.

4. Discussion

These results permit us to argue that we have recordednew cases of trophobiotic relationships between ants andheteropterans. We confirm the conclusions of Maschwitzet al. [5] who predicted that trophobiosis between antsand heteropterans may be more widespread than has beenpreviously assumed. The rareness of known cases couldbe due to the specificity of the bugs vis-à-vis their hostplants added to the specificity with ants. The relationshipbetween Cat. rugosa and B. micrantha and to a lesserdegree B. grandis, is narrowly specific when consideringthe high plant diversity in the studied area [7, 8, 11].Moreover all the species of the genus Bridelia are notequally suitable, as B. ferruginea, a very common species

Figure 1. A. Myrmicaria opaciventris workers receiving honeydew from an adult Caternaultiella rugosa.B. M. opaciventris workers attending Cat. rugosa nymphs and adults; the two workers (at top center) are sharing a droplet of honeydew that theyhold thanks to surface tension.C. Camponotus brutus workers attending Cat. rugosa nymphs; one nymph emits honeydew that attending workers prepare to collect.

451

A. Dejean et al. / C.R. Acad. Sci. Paris, Sciences de la vie / Life Sciences 323 (2000) 447–454

of the savannah, is never occupied by C. rugosa. Relation-ships with ant species are also narrowly specific, so thatthe experimental part of this study was possible only by theintroduction of the bug into easily accessible sites inYaoundé.

It is known that the pavilions built by ants provideprotection for Sternorrhyncha, Cicadomorpha andFulgoromorpha from weather and enemies, and that theimmediate cleaning by workers of the honeydew fallenonto the substrate, inhibiting the growth of mildew, is aservice considered to be important for maintaining thehygiene of the attended trophobionts [1, 2, 12–14]. Thedistribution of the tasks with major workers acting asguards, while media attend the trophobiont with certainindividuals being specialized in transporting the honey-dew, has previously been observed for Camp. brutusattending tettigometrids (Fulgoromorpha) [15].

For Camp. brutus, the solicitation of plataspid honey-dew was characterized by several elaborated behaviouraltraits if compared to those known for ants associated withSternorrhyncha, Cicadomorpha or Fulgoromorpha. Wenoted interspecific semantic communication when thenymphs raised their bodies before excreting honeydew, orwhen they repeated the alternate extrusion and with-drawal of the first honeydew droplet when an attendingworker did not immediately react to the previous signal;and interspecific semiochemical communication whenthe workers specifically palpated the dorsal abdominalglands of the attended last-instar bugs. This behaviour,similar to that of workers of the same ant species palpatingthe stigmatic glandular areas of tettigometrids [15–17],

suggests the existence of allomones, as adults ofCat. rugosa were not palpated. Indeed, the dorsal abdomi-nal glands are concealed by the scutellum in adultplataspids and their function usually ceases at the imagi-nal moult (W.R. Dolling, pers. comm.). Maschwitz et al.[5] did not mention if Meranoplus workers antennated thedorsal abdominal glands of the attended Asian plataspidnymphs, but they remarked that the workers were moreattracted to the nymphs than to the adults. In comparison,the behaviour of M. opaciventris workers seems moreopportunistic, as most of the honeydew droplets wererecuperated indirectly, after they adhered to the workers’bodies thanks to surface tension.

Interspecific territoriality in ground-dwelling ants is fre-quently noted around resources, while territories them-selves can overlap [2, 18, 19]. The same is true for certainarboreal species, particularly when they tend Sternorrhyn-cha, Cicadomorpha or Fulgoromorpha, with greateraggressiveness towards other ant species (they also attackenemies of their trophobionts), and local changes in theirrhythm of activity in order to continually defend thesefood sources against competitors [2, 9, 10, 20, 21]. In thepresent study we noted continual activity with a decreaseat night in M. opaciventris (huge ground-nesting colonies)independently of the attended insect, as has been previ-ously noted [22, 23]. On the contrary, Camp. brutus, anocturnal species [10, 24], changed its rhythm of activityin order to attend clusters of Cat. rugosa both day andnight (nevertheless, there were more ants at night thanduring the day), while the stictococcids, used as a controlgroup, were attended only at night. Camp. brutus workers

Figure 2. Comparisons between the number of workers of Camponotus brutus and Myrmicaria opaciventris attending the bugs outside pavilionsduring the day and at night.Statistical comparisons (paired t-tests; df = 19 in all cases). Camp. brutus/Cat. rugosa: t = 7.03; P < 0.001; Camp. brutus/Stictococcidae: t = 11.5;P < 0.001; M. opaciventris/Cat. rugosa: t = 3.7; P < 0.01; M. opaciventris/Stictococcidae: t = 7.97; P < 0.001.

452

A. Dejean et al. / C.R. Acad. Sci. Paris, Sciences de la vie / Life Sciences 323 (2000) 447–454

pursued and attacked competing ant species approachingthe plataspids, while they shared the attendance of thestictococcids with the same ants on the same trees.

The influence of large tettigometrids (Fulgoromorpha)has also been noted to modify the territoriality of Camp.brutus, including the rhythm of activity of colonies occu-pying fig trees [9, 15, 17]. This condition is thereforeindependent of the host-plant, whether the attended insectsare fulgoromorphs or plataspids, and the size of the tro-phobiont itself (small first-instars are attended as well aslarge last-instar nymphs). It is probably rather correlatedbetween the amount of food available and the lowestenergetic cost to the ants needed to maintain the mutual-istic relation. This implies three conditions: each tropho-biont individual produces a large quantity of honeydew,individuals are grouped (naturally or forced into shelters),and the ant–trophobiont relationship is probably regulatedby allomones that have a certain specificity.

Roughly categorizing the different hemipteran groupsinto phloem (Sternorrhyncha and Fulgoromorpha), xylem(Cicadomorpha) or parenchyma (Heteroptera) feeders isoversimplified. For example, psyllids (Sternorrhyncha) feedon phloem, xylem and mesophyll parenchyma [25, 26],and, while being phloem feeders, Fulgoromorpha alsofeed on xylem and parenchyma [27, 28]. As most cases ofant mutualisms occur with Sternorrhyncha, said to bephloem feeders, one can ask whether Heteroptera associ-ated with ants are also phloem feeders, while most of thosethat attack plants are reputed to be parenchyma feeders[29]. Indeed, it is known that certain Heteroptera speciesfrom the families Aradidae, Miridae, Coreidae and par-ticularly Plataspidae are phloem feeders [4, 5, 30]. Asseveral plataspid species were noted as phloem feeders

with some of them having trophobiotic relationships withants, one can ask if phloem feeding is, or not, general inthe family Plataspidae.

All durable hemiptera–ant associations are observedwhen the hoppers are grouped and unable to escapebecause of their subterranean habitat or sub-social behav-iour, or forced to be gregarious when kept under antpavilions [31]. This is exactly the case of the plataspidnymphs and adults that combine group living and aresheltered in ant pavilions. This allows the ants to obtain alarge amount of honeydew in the same place, a fact whichinteracts directly with their territoriality. Also, Camp. bru-tus workers seem to be favoured in these relationshipsowing to their large size, as one worker can tend severaladults or last-instar nymphs at a time, or groups of morethan ten first-instar nymphs.

Finally, we believe that allomones might play a role inthese relationships, as previously noted for tettigometrids[15–17]. This suggests that these semiochemical com-pounds are more attractive to certain ant species ratherthan others, increasing the territoriality of the most attractedant species. New studies in this direction are now needed.

Acknowledgements: We are especially indebted to DrW.R. Dolling (Brook Farm, Elstronwick, Hull, UK) forthe identification of the Plataspidae and for providinghelpful information on their biology. We are grateful toDr A. Akoa (University Yaoundé I, Cameroon) for theidentification of the plants and to Dr B. Bolton (TheNatural History Museum, London, UK) for the identifi-cation of the Formicidae. This work was supported bythe French Ministry of Co-operation (Project CAMPUS108/CD/90).

References

[1] Beattie A.J., The Evolutionary Ecology of Ant–Plant Mutualisms,Cambridge University Press, Cambridge, 1985.

[2] Hölldobler B., Wilson E.O., The Ants, The Belknap Press of Harvard,University Press, Cambridge, 1990.

[3] Green E., Note on the attractive properties of certain larvalHemiptera, Entomol. Month. Mag. 11 (1900) 1–185.

[4] Maschwitz U., Klinger R., Trophobiontische Beziehungen zwis-chen Wanzen und Ameisen, Insectes Soc. 21 (1974) 163–166.

[5] Maschwitz U., Fiala B., Dolling W.R., New trophobiotic symbiosesof ants with South East Asian bugs, J. Nat. Hist. 21 (1987) 1097–1107.

[6] Moffett M.W., Ants and plants. Friends and foes, Nat. Geogr. 195(1999) 100–113.

[7] Dejean A., McKey D., Gibernau M., Belin M., The arboreal antmosaic in a Cameroonian rainforest, Sociobiology 35 (2000) (in press).

[8] Dejean A., Amougou-Akoa, Djieto-Lordon C., Lenoir A., Compari-son of ant–plant relationships between two formations of a secondarizedequatorial zone, Sociobiology 23 (1994) 275–292.

[9] Dejean A., Bourgoin T., Gibernau T., Territoriality of an ant speciesinduced by mutualistic tettigometrid: a new case of fig protection, Eco-science 4 (1997) 446–453.

[10] Mercier J.L., Dejean A., Ritualized behavior during competitionfor food between two Formicinae, Insectes Soc. 43 (1996) 17–29.

[11] Dejean A., Gibernau M., A rainforest ant mosaic: the edge effect,Sociobiology 35 (2000) (in press).

[12] Buckley R.C., Interactions involving plants, Homoptera and ants,Annu. Rev. Ecol. Syst. 18 (1987) 111–155.

[13] Majer J.D., Ant–Plant interactions in the Darling botanical districtof Western Australia, in: Buckley R.C. (Ed.), Ant–Plant Interactions inAustralia, Junk, The Hague, 1982, pp. 45–61.

[14] Dejean A., Orivel J., Durand J.L., Ngnegueu P.R., Bourgoin T.,Gibernau M., Interference between ant species distribution in differenthabitats and the density of a maize pest, Sociobiology 35 (2000) 175–189.

[15] Dejean A., Bourgoin T., Ant defense of Euphonyarthex phyllos-toma (Homoptera: Tettigometridae) during trophobiotic associations, Bio-tropica 32 (2000) 112–119.

[16] Bourgoin T., Les glandes tégumentaires chez les Tettigometridae(Hemiptera Fulgoromorpha), Ann. Soc. Entomol. Fr. 22 (1986) 139–144.

[17] Dejean A., Bourgoin T., Relationships between ants (Hymenoptera:Formicidae) and Euphyonarthex phyllostoma (Hemiptera:Tettigometridae), Sociobiology 32 (1998) 91–100.

[18] Gordon D.M., Nest-plugging: interference competition in desertants (Pogonomyrmex barbatus and Novomessor cockerelli), Oecologia75 (1988) 114–118.

[19] Savolainen R., Vepsäläinen K., A competition hierarchy amongboreal ants: impact on resource partitioning and community structure,Oikos 51 (1988) 135–155.

[20] Cushman J.H., Addicott J.F., Intra and interspecific competition formutualists: ants as a limited and limiting resource for aphids, Oecologia79 (1989) 315–321.

[21] Buckley R.C., More aggressive ant species (Hymenoptera: Formi-cidae) provide better protection for soft scales and mealybugs (Homoptera:Coccidae; Pseudococcidae), Biotropica 23 (1991) 282–286.

453

A. Dejean et al. / C.R. Acad. Sci. Paris, Sciences de la vie / Life Sciences 323 (2000) 447–454

[22] Kenne M., Dejean A., Diet and foraging activity in Myrmicariaopaciventris (Hymenoptera: Formicidae: Myrmicinae), Sociobiology 33(1998) 171–184.

[23] Kenne M., Dejean A., Spatial distribution, size and density of nestsof Myrmicaria opaciventris Emery (Formicidae: Myrmicinae), InsectesSoc. 46 (1999) 179–185.

[24] Mercier J.L., Dejean A., Lenoir A., Limited aggressiveness amongAfrican arboreal ants sharing the same territories: the result of aco-evolutionary process, Sociobiology 32 (1998) 139–150.

[25] Ullman D.E., Mc Lean D.L., The probing behavior of the summer-form pear psylla, Entomol. Exp. Appl. 47 (1988) 115–125.

[26] Ullman D.E., Mc Lean D.L., Feeding behavior of the winter-formpear psylla, Psylla pyricola (Homoptera: Psyllidae), on reproductive andtransitory host plants, Environ. Entomol. 17 (1988) 675–678.

[27] Chang V.C.S., Feeding activity of the sugarcane leafhopper: iden-tification of electronically recorded waveforms, Ann. Entomol. Soc. Am.71 (1978) 31–36.

[28] Chang V.C.S., Ota A.K., Feeding activities of Perkinsiella leafhop-pers and Fiji didease resistance of sugarcane, J. Econ. Entomol. 71 (1978)297–300.

[29] Carver M., Gross G.F., Woodward T.E., Hemiptera. The insects ofAustralia, CSIRO, Melbourne, 1991.

[30] Backus E.A., Sensory systems and behaviours which mediatehemipteran plant-feeding: a taxonomic overwiew, J. Insect Physiol. 34(1988) 151–165.

[31] Bourgoin T., Habitat and ant-attendance in Hemiptera: a phylo-genetic test with emphasis on trophobiosis in Fulgoromorpha, in:Grancolas P. (Ed.), The origin of biodiversity in insects: phylogenetictests of evolutionary scenarios, Mém. Mus. Nat. Hist. Nat., 173, 1997,pp. 109–124.

454

A. Dejean et al. / C.R. Acad. Sci. Paris, Sciences de la vie / Life Sciences 323 (2000) 447–454