Lepidoptera, Lycaenidae - Springer Link

Oecologia (1996) 106:57--62

9 Springer-Verlag 1996

Alain Dejean 9 Guy Beugnon

Host-ant trail following by myrmecophilous larvae of Liphyrinae (Lepidoptera, Lycaenidae)

Received: 22 May 1995 / Accepted: 17 October 1995

Abstract In this study we report a case of ant-trail following by lycaenid caterpillars. Euliphyra mirifica and E. leucyana caterpillars are involved in a commensal association with the weaver ant Oecophylla longinoda. The host nests are made with leaves which over the course of time dry out or are broken open by storms, forcing the ants to migrate and build a new nest elsewhere. Euliphyra caterpillars are stimulated by recruitment behavionr which triggers the migration of their host. They then follow the host trails leading to the new nesting site. Laboratory experiments showed that these caterpillars are able to follow host trails under varied conditions: (1) flesh trails actually used by workers, (2) fresh trails in the absence of workers, (3) heterocolonial, 2-month-old trails, and (4) flesh trails washed with water (to simulate the effect of tropical rains). They can also bridge trail gaps of more than 1 cm. Under natural conditions, the trails are frequently situated along thin twigs. The forward progress of the ants in such a situation is not impeded by the presence of large Euliphyra larvae. Workers just climb over the caterpillars, even on larger trails where there is enough room to pass alongside them. This suggests that an allomone is secreted on the dorsal part of the caterpillars. When crawling along heterocolonial trails, the caterpillars are not attacked, even if about 21% of the workers from the new colony spread their mandibles when encountering them. They are then adopted and are admitted to the nest of the new host colony of O.

longinoda. Key words Interspecific trail following 9Kairomones 9 Allomones 9Lycaenidae 9Formicinae A. Dejean ( ~ ) Laboratoire d'Ethologie Exptrimentale et Comparde (URA CNRS No. 667), Universit6 Paris XIII, Av. Jean Baptiste Cldrnent, F-93430 Villetaneuse, France, Fax: (33) 1.49.40.39.75 G. Beugnon Laboratoire d'Ethologie et Psychologie Animale (URA CNRS No. 1837), Universit6 Toulouse III, 118 route de Narbonne, F-31062 Toulouse, France

Introduction The weaver ant Oecophylla Ionginoda is an arboreal species of the African tropical forest which defends chemically marked territories at both the intra- and interspecific levels (Leston 1973; H611dobler and Wilson 1978). The colonies have polydomous nests formed of living leaves bound together with larval silk. The nests can be abandoned for several reasons: leaves used for the nest can dry out, or the branches can be damaged by wind, animals or man and the nests partially or totally destroyed. Therefore emigration is frequent. Workers utilize rectal gland trails for the recruitment of nestmates which carry brood from one nest to another and protect the queen, acting as her bodyguards. The organization of this type of recruitment has been described by H611dobler and Wilson (1978, 1983, 1990). The marked territories and the long-lasting chemical trails linking the different nests and some permanent food sites are functional for several months; they are also very resistant to rainwater (Dejean and Beugnon 1991; Beugnon and Dejean 1992). The larvae of two species of lycaenid, Euliphyra mirifica and E. leucyana (Liphyrinae) live inside the nests of O. longinoda, in direct contact with workers and brood. They feed on prey collected by ants and on sugar sources provided during trophallactic exchanges between workers (Farquharson 1921; Lambom 1913; personal observation). We have never observed these caterpillars attacking host brood, as occuls in Liphyra brassolis, which live in colonies of Oecophylla smaragdina (Cottrell 1987). Hence these two Euliphyra species are commensal rather than parasitic (for definitions see Pierce and Young 1986; Bristow 1987). The butterfly, whose eclosion takes place inside the ants' nest, spreads its wings while resting on the nest itself or on neighbouring branches. It is not attacked by workers, in contrast to Anthene emolus which are particularly vulnerable to predation by O. smaragdina during eclosion (Fiedler and Maschwitz 1989). Euliphyra females lay their eggs directly on the host nest of O. longinoda. The first instar larvae are then

58

OECOLOGIA

106 (1996) 9 Spfinger-Verlag

transported into the nest by the foraging workers which discovered them (Farquharson 1921, personal observation). Thus, Euliphyra are obligate myrmecophiles, living in association with their host ant throughout their larval life. In order to discover if Euliphyra caterpillars are adapted to the particular lifestyle of their host, we decided to analyse the responses of these caterpillars when their host migrated from one nest to another. We asked the following questions: 1. Are the Euliphyra caterpillars able to perceive the information that triggers recruitment during emigration? 2. If they are able to follow the host trail, is the presence of workers necessary as it is during the emigration of the queens (to reinforce the trail and to guide the queen)? 3. After the total destruction of a nest leading to the dispersion of all the ants, are the Euliphyra caterpillars able to find another nest in activity by following the long-lasting trails previously deposited, in the same way the host workers do? 4. During emigration to a new nest, workers can walk rapidly on very thin twigs. Do slow Euliphyra caterpillars hinder or prevent the movement of ants going to the new nest or those coming back to the old nest to look for the remaining brood?

Materials and methods This work was carried out in Yaound6 (Cameroon) between 1989 and 1993 on 21 colonies of O. longinoda living in association with the last instar larvae of E. mirifica (2.5 cm long and 0.8 cm wide) and E. leucyana (2.1 cm long and 0.9 cm wide). The colonies were installed in the laboratory in wooden boxes (47 x 36 x 9 cm) which were covered with glass plates to allow the observation of both ants and caterpillars.

Reaction of caterpillars to the recruitment of the ants A rearing box containing a colony of O. longinoda was placed on a table (100 x 75 cm) (Fig. 1). Buckets of water under the legs of the table prevented the ants from escaping. Honey was provided in small dishes. A potted tree of Alchornea cordifoIia (Euphorbiaceae) was installed 3 m from the table. A string 2 m m in diameter was stretched between the table and a branch of the tree. Within a few minutes of the installation of the box, some workers climbed up the string, reached the A. cordifolia tree and explored its foliage. Within 3 0 - 6 0 rain, some workers started building a new nest in the foliage of the small tree. The recruitment of nestmates started before the building of the new nest was completed. Some workers, apparently excited, came back from the tree into the rearing box where their behaviour could be observed. The ants moved forward by jerks and antennated a great number of nestmates. The manner in which this phenomenon occurs corresponds to the marching order described by Hrlldobler and Wilson (1978, 1990). This experiment was replicated with 12 colonies sheltering 2 - 7 E. mirifica and/or 1-3 E. leucyana larvae (a total of 33 caterpillars were used). Once the ants began to move, the behaviour of the caterpillars was observed in order to be able to note at what moment they came out of the host's nest. The speeds at which the caterpillars and workers travelled along a 50-cm-long trail on the table surface near the rearing box were recorded.

Fig. 1 Diagram of the experimental room. An artificial nest (a) of Oecophylla longinoda, containing commensal caterpillars of Euliphyra mirifica and E. leucyana, was placed on a table (b). Buckets of water (c) under the legs of the table prevented the insects from escaping. A potted tree (d) of Alchornea cordifolia (Euphorbiaceae) was installed 3 m from the table. A string (e) of 2 m m diameter was fixed to the table surface with a 4 cm long nail Or) and stretched between the table and a branch of the small tree. It was then possible to raise the string to a height of 1 cm above the chemical trail (g), deposited by the workers on the table and along the string, directly linking the artificial nest to the potted tree Behaviour of the Euliphyra caterpillars and of the host workers confronted with a gap in the chemical trail The string linking the table where the artificial nest was located to the A. cordifolia tree was attached to the surface of the table with a 4-era-long nail (Fig. 1). It was then possible to raise the string 1 cm above the chemical trail on the table. We observed the behaviour of the ants and caterpillars at this rupture point and recorded the time they needed to overcome this difficulty. Tests of trail following During emigration Euliphyra caterpillars have to reach the new nest even though they are confronted with (1) a multiple choice (when they reach the string they can take any direction on the table or follow the trail on the string; Fig. 1), and (2) three binary choices (one at the place where the string is attached to a branch of the tree, and at two forks in the branches leading to the new nest). In this second situation, the probability of a caterpillar reaching the new nest by chance without error is P = 0.125 (1/8). To study whether the Euliphyra caterpillars are able to find and follow the ant trails, we used the experimental paradigm developed by Dejean and Beugnon (1991) (Fig. 2). In an observation room, two concentric circles were drawn on a wall. The centre of the circles corresponds to the edge of a board 10 cm wide. During previous experiments, O. longinoda workers laid a trail that crossed the circles, ending at their centre (where honey was deposited). For these experimental sessions we used ten societies of O. longinoda, sheltering a total of 34 caterpillars of E. mirifica and 4 of E. teucyana; all or a part of the societies were experimented on, depending on the nature of the tests.

Presence of workers on the trail A nest of O. longinoda was placed on the board at the level of the centre of the circles. Workers leaving the nest were attracted by the trail, which they follow until its terminal point where honey is placed (see Dejean and Beugnon 1991; Beugnon and Dejean 1992). For this experiment, we tested 27 E. mirifica caterpillars (each used 5 times) and 4 of E. leucyana (each used 8 times). We compared the number of cases where the caterpillars followed the trail versus those where they took another direction.

~\ k\\\\\\\\\\\\w~

OECOLOGIA 106 (1996) 9 Springer-Verlag

59

We tested four E. mirifica caterpillars originating from an alien colony of O. longinoda and conducted two sets of observations: the first 30 encounters with alien workers and then the 75 successive encounters.

Results Reaction o f Euliphyra caterpillars to the emigration recruitment o f host ants

Fig. 2 Ant-trail following tests. The solid line indicates the departure of a 1.5 cm wide chemical trail, reinforced with faecal markings, linking the artifical nest of a colony of Oecophylla longinoda to their food site. Concentric circles drawn on the wall at 50 cm and 100 cm are centred at the nest entrance. The angular position of all the lycaenid larvae crossing the circles was registered

Fresh trail, absence of workers We tested 14 caterpillars of E. mirifica (each used 4 times) and 4 of E. leucyana (each used 8 times) just after removing the ants.

Two-month-old trail The difference between this third experiment and the second one lay in the fact that we experimented with scent trails deposited by workers from an alien colony, 2 months earlier. We tested ten caterpillars of E. mirifica (each used 7 times) and four of E. leucyana (each used 8 times).

Fresh trail, washed, absence of workers The difference between this experiment and the second one lay in the fact that the wall was vigorously sponged with water. We tested seven caterpillars of E. mirifica (each used 4 times). Behaviour of workers when faced with Euliphyra caterpillars During the experimentally induced emigrations, we observed the behaviour of 500 workers faced with five E. mirifica caterpillars moving along the string located between the rearing box and the potted A. cordifolia tree. We recorded the number of workers (1) avoiding the caterpillars and coming back, (2) walking over the caterpillars, (3) immobilizing themselves on the body of the caterpillars, and (4) antennating the caterpillars. The same kinds of observations were made with 500 workers and 14 E. mirifica caterpillars following a fresh 4-cm-wide trail on the table. We distinguished those cases in which workers passed alongside the caterpillars from those where the workers climbed onto the catapillars despite having enough room to pass alongside them. If the caterpillars (0.8 cm wide) do not attract or repel ants, workers will climb on them only in one case out of five (0.8/4 cm). This results in a theoretical number of 100 cases out of the 500 observed workers. This situation permitted us to observe all of the behavioural responses of the ants when confronted with the presence of these caterpillars on their trail. This differs from testing a single response to the caterpillars by the ants as was done in a preliminary test and which resulted in 100% attraction.

The duration of the ants' emigration was in the order o f 161 +_ 24 min (total duration: 150 < g < 171 min, with a 99.00% confidence interval). The caterpillars left the nest 42 _ 25 min (g < 52 rain, with a 99.00% confidence interval) after the ants began to migrate. They were therefore able to perceive the information that triggers the emigration. N o larvae ever left the nest during other kinds o f recruitment.

Behaviour of the Euliphyra caterpillars and o f the workers when confronted with a gap in the chemical trail L o n g - l e g g e d major workers with long antennae cross a gap in the trail when most o f the minor workers cannot reach the string with their antennae and are stopped in place. After a few minutes, minor workers turn and go back to the artificial nest whereas two to five majors stop and take the string into their mandibles, making a living bridge for minors. In contrast, the Euliphyra caterpillars (6 E. leucyana and 33 E. mirifica) bridged the gap after their first attempt in 89.75% o f the cases (all 6 E. leucyana larvae and 29/33 E. mirifica) or after the second attempt. The time taken for crossing ranged between 20 s and 14.5 rain. W h e n faced with a gap in the trail, the caterpillars lifted their heads off the mantle and raised the anterior part o f their bodies. They grasped the string with their thoracic legs and then twisted themselves around the string.

Tests o f trail following

Trail following during emigration. The 33 caterpillars tested reached the new nest without any error both when they had to climb onto the string (multiple choice) and at the points where they had to make binary choices. The probability o f this last situation occurring by chance is P = 0.0038.

Presence of workers on a scent trail The Euliphyra caterpillars followed trails actually used by workers up to the end in most o f the observed cases (122/135 fbr E. mirifica and 30/32 for E. leucyana; Fig. 3). N o caterpillars walked outside the limits o f the ants' trail; hence, those that did not follow the trail

60

O E C O L O G I A 106 (1996) 9 Springer-Verlag

% 9O 80 70 60 50 40 30 2O 10

nts

motionless throughout the duration of the experimental session (11 cases for E. mirifica and 4 cases for E. leucyana) or, in one case for E. leucyana, walked outside the limits of the trail (Fig. 3).

Fresh scent trail, washed, absence of workers follow / trail

explore

stay / enter nest

% 90+ 8O

70 6O

E. leucyana

sO 4o 3o

Even under those experimental conditions which could cause trail dissipation, E. mirifica caterpillars (only one species tested; Fig. 3) were able to detect and follow the trail (21/28 cases). The caterpillars that did not follow the trail stayed motionless throughout the duration of the experimental session. Statistical comparisons between the four types of trail-following experiment outlined above did not result in significant differences (Fig. 3).

20,

1:. follow I trail

explore

stay I enter nest

Behaviour of workers when faced with EuIiphyra caterpillars

Fig. 3 Test of Oecophylla lonfiinoda-trail following by Euliphyra caterpiUars in four situations: (1) fresh trail, presence of the work- Homocolonial caterpillars ers, (2) absence of workers, (3) 2-month-old trails, and (4) absence of workers, the fresh trails were washed before the experiment. When the encounter occurred on the string during emiNumber of cases: E. mirifica equals 135, 56, 70, and 28, respec- gration, only 2% of the workers turned back after antentively, in situations 1-4; E. leucyana equals 32 in situations 1, 2 and 4. Number of caterpillars in the experiment: E. mirifica equals nating the larva (Table 1). Such behaviour can also occur 27, 14, 7 and 10, respectively, in situations 1-4; E. leucyana when two workers come face-to-face. equals 4 in situations 1,2 and 3. Statistical comparisons (Fisher exThe speed of the caterpillars (ranging from 4 to act test and Chi-square test): E. leucyana P = 0.14 (1 x 2), 11 cm/min; mean speed: 7.1 ___0.7 cm/min, n = 30), is P = 0.29 (l x 3), P = 0.56 (2 x 3), E. mirifica P = 0.17 (1 x 2), lower than for ants (mean speed: 92.6 __+ 12.2 cm/min, P = 0.29 (1 x 3), P = 0.052 (1 x 4), P = 0.93 (2 x 3), P = 0.56 n = 30). Nevertheless, their presence does not cause (2 x 4), P = 0.38 (3 x 4) much disturbance to workers either overtaking them or encountering them face-to-face (mean speed: stayed at the starting point. These data indicate that the 89.1 _ 13.2 cm/min, n = 30; the difference with the preants' trails are effectively recognized and followed by vious case being t = 1.1, NS). Only.29.8% of the workers slow down when climbing on the caterpillars and 6.8% the caterpillars. stop on their mantle for a moment. When the encounter occurs on the 4-cm-wide trail, 85.2% of the workers climb onto the back of the caterpilFresh scent trail, absence of workers lar (which are 0.8 cm wide), whereas the theoretical valThe caterpillars also followed the trail previously laid by ue should be only 20% (4/0.8 cm). The difference is the ants (46/56 of the observed cases for E. mirifica, and highly significant ()~2 = 426.2, 1 df, P < 10-9) as it is in 25/32 for E. leucyana; Fig. 3). The caterpillars that did the previous case (g 2 = 51.6. 1 df, P < 10-9) (Table 1). During emigration, some workers stop for a moment not follow the trail either remained motionless throughout the duration of the experimental session (eight cases on the caterpillars to briefly palpate the crest of their for E. mirifica and seven cases for E. leucyana) or, more mantles. The significant difference between the rate of rarely, walked outside the limits of the trail for a few such behaviour during encounters on the string versus on centimetres but then followed the trail (twice for E. mi- a large trail is probably due to the need to not block the rifica and once for E. leucyana). Thus, the presence of way on the string (6.8% and 12.2%, respectively; workers on the trail is not necessary to guide the caterpil- )~2 = 7.8, 1 df, P = 0.005). lars towards the new nesting site o f the ants.

Alien Euliphyra caterpillars Two-month-old scent trail The caterpillars of both species are able to follow the trail (59/70 for E. mirifica, and 27/32 for E. leucyana). The caterpillars that did not follow the trail either stayed

During the first 30 encounters (wide trails), 21% of the workers raised themselves up on their forelegs and spread their mandibles, but they never attacked the caterpillars.

OECOLOGIA 106 (1996) 9 Springer-Verlag Table 1 Behavior of O. long# noda workers when encountering E. mirifica caterpillars. 1stop: the workers stopped for a short moment on the mantle of the caterpillar. 2- antennating: concerns the workers which did not climb on the caterpillars. String: 2 mm in diameter.

O. longinoda workers' behaviour

Homocolonial caterpillars On string

61

Alien caterpillars On scent trail 4 cm wide 30 first contacts

75 other contacts

Climb on the caterpillars

98%

85.2%

68.3%

91%

Do not climb on the caterpillars Stop Antennating N~ of workers N ~ of caterpillars

2%

14.8%

31.7%

9%

6.8% 2% 500 5

12.2% 0% 500 14

60.8% 22.5% 120 4

30% 1.3% 300 4

The rate of workers climbing on the caterpillars was 68.3% (significantly different from a random distribution: X2 = 56.8, 1 df, P < 10-9); therefore, they are less attracted to climbing on alien caterpillars than on homocolonial ones (85.2%;)~2 = 17.5, 1 df, P < 0.0001) (Table 1). The percentage of workers climbing on the caterpillars increases drastically during the 75 ensuing encounters, reaching up to 91% (comparison with the first 30 encounters: g2 = 32.0, 1 df, P < 10-7). The workers that climbed on the caterpillars had a tendency to stop to antennate the crests of their mantles. This behaviour is more frequent for alien than for homocolonial caterpillars (60.8% versus 12.2%; Z 2 = 132.2, 1 df, P < 10-9) and decreases gradually with the number of encounters (60.8% versus 30%; )~2 = 33.0, 1 df, P < 10-8). The rate of workers which palpated the caterpillars without climbing on them was 22,5% during the 30 first encounters, after which it rapidly decreased to 1.3%. Lastly, we have the four larvae of E. mirifica entering into the alien colonies of O. tonginoda; all of them were rapidly adopted (as were three others and four of E. leucyana tested elsewhere), while six alien E. mirifica caterpillars directly introduced into nests of O. longinoda were palpated and bitten. They flattened out on the substratum and stayed immobile. One of them was killed, another was pulled back outside the nest by a worker which succeeded in biting the base of the mantle, and the remainder were eventually adopted.

Discussion Euliphyra caterpillars, which are obligate myrmecophiles (Farquharson 1921; Henning 1983), are able to perceive the information triggering the emigrations of their host and then follow their trails. Hence, these caterpillars perceive and use to their benefit several types of pheromones emitted by the host ant (see H/311dobler and Wilson 1978; for the caterpillars these products correspond to kairomones): (1) signals triggering the emigration, (2) short-lasting trails used during emigration, and (3) long-lasting trails to permanent food sources used by the host, permitting the caterpillars to find a new nest (even when belonging to anoth-

er colony or when washed by rains or very old). In addition, they are able to bridge a gap in the trail. Therefore these caterpillars are well adapted to live in O. longinoda nests, which are frequently damaged or destroyed by storms that break branches and scatter the nest contents. O. longinoda-trail following undertaken by earlier larval instars of Euliphyra was frequently observed during numerous experiments on the host ant, but in this work we used only last stage caterpillars in order to obtain homogeneous results. In some lycaenid species living in association with ants, larvae spend the first part of their life on a food plant before crawling down onto the ground. From there they can be carried to the host nest by workers that encounter them, such as in Maculinea arion and M. rebeIi (Thomas et al. 1989; Elmes et al. 1991). Maculinea teIeius caterpillars were reported to be able to find the host nest by themselves by ant-trail following (Schroth and Maschwitz 1984), but this appears to be a laboratory artefact (U. Maschwitz, personal communication) and was not confirmed in a later study with a much larger sample size (Fiedler 1990). Ant-trail following was never observed in any Maculinea species in the field (Thomas 1984; Thomas et al. 1989). Therefore, the present work reports the first case of ant-trail following by inquiline lycaenid caterpillars. Although the data remain unpublished, trail following has been observed in the larvae of the Australian obligate myrmecophile Jalmenus evagoras (Mathews 1993). It is known that myrmecophilous or termitophilous insects are able to follow the trails of their hosts (Moser 1964; Howard 1980; Cammaerts et al. 1990; Quinet and Pasteels 1995). Some guests of army ants perceive information indicating emigration (Akre and Rettenmeyer 1968) and are able to utilize the ant trails for a few hours after they have been established (Akre and Togerson 1969; Togerson and Akre 1969). In contrast, we found that even a 2-month-old ant trail still significantly channels Euliphyra caterpillars towards a new nesting site. Chemical communication plays an important role in the associations between many lycaenid species and their host ants, which are generally attracted to their guests thanks to allomones (Henning 1983). For example, Anthene emolus caterpillars are not able to follow the trails of their host, OecophyIIa smaragdina, but they are carried

62

OECOLOGIA 106 (1996) 9 Springer-Verlag

adoption and behaviour in ant-nests. J Zool Lond 223: 447-460 Eltringham H (1913) The larva of Euliphyra mirifica. Trans Entomol Soc Loud 1913:509-515 Eltringham H (1921) On the larvae and pupae of Lepidoptera, chiefly Lycaenidae, collected by CO Farquharson, WA Lamborn and the Rev Canon K St A Rogers. Trans Entomol Soc Lond 1921:473-489 Farquharson CO (1921) Five years' observations (1914-1918) on the bionomics of Southern Nigerian insects, chiefly directed to the investigation of Lycaenid life histories and to the relation of Lycaenidae, Diptera, and other insects to ants. Trans Entotool Soc Lond 1921:325-530 Fiedler K (1990) New information on the biology of Maculinea nausithous and M.teleius (Lepidoptera: Lycaenidae). Nota Lepidop 12:246-256 Fiedler K, Maschwitz U (1989) The symbiosis between the weaver ant Oecophylla smaragdina, and Anthene emolus, an obligate myrmecophilous lycaenid butterfly. J Nat Hist 23:833-846 Henning SF (1983) Chemical communication between lycaenid larvae (Lepidoptera: Lycaenidae) and ants (Hymenoptera: Formicidae). J Entomol Soc S Afr 46:341-366 H611dobler B (1979) Territories of the African weaver ant [OecophylIa longinoda (Latreille)]. A field study. Z Tierpsychol 51: 201-213 H611dobler B, Wilson EO (1978) The multiple recruitment systems of the African weaver ant Oecophylla longinoda (Latreille) (Hymenoptera: Formicidae). Behav Ecol Sociobiol 3:19-60 H011dobler B, Wilson EO (1983) The evolution of communal nestweaving in ants. Am Sci 71:490-499 HiSlldobler B, Wilson EO (1990) The Ants. Springer, Berlin Heidelberg New York Howard RW (1980) Trail-following by termitophiles. Ann Entomol Soc Am 73:36-38 Kitching RL, Luke B (1985) The myrmecophilous organs of the larvae of some British Lycaenidae (Lepidoptera): a comparative study. J Nat Hist 19:259-276 Lamborn WA (1913) On the relationship between certain West African insects, especially ants, Lycaenidae and Homoptera. Acknowledgements We are grateful to Dr. K. Fiedler for his Trans Entomol Soc Loud 1913:436-520 comments and advice concerning ant-lycaenid relationships and for sending us reference articles concerning this topic. Thanks are Leston D (1973) The ant mosaic - tropical tree crops and the limiting of pests and diseases. Pest Articles and New Summaries also due to Prof. U. Maschwitz for his comments on ant-trail fol19:311-341 lowing by lycaenid caterpillars and to M. Libert for the identification of the butterflies obtained through laboratory breeding. This Malicky H (1969) Versuch einer Analyse der/Skologischen Beziehnngen zwischen Lycaeniden (Lepidoptera) und Formiciden work was supported by a CAMPUS project (108/CD/90) of the (Hymenoptera). Tijdschr Entomol 112:213-298 French Ministry for Cooperation. Matthews JNA (1993) Aggregation and mutualism in insect herbivores. DPhil thesis, University of Oxford, Oxford Moser JC (1964) Inquiline roach responds to trail making subReferences stance of leaf-cutting ants. Science 143:1048-1049 Pierce NE, Young WR (1986) Lycaenid butterflies and ants: two species stable equilibria in mutualistic, commensal, and paraAkre RD, Rettenmayer CW (1968) Trail-following by guests of sitic interactions. Am Nat 128:216-227 army ants (Hymenoptera: Formicidae: Ecitonini). J Kansas Quinet Y, Pasteels JM (1995) Trail following and stowaway beEntomol Soc 41:165-174 haviour of the myrmecophyllous staphyllinid beetle, HomoeAkre RD, Togerson RL (1969) Behaviour of Vatesus beetles assousa acuminata, during foraging trips of its host Lasius fuliciated with army ants (Staphylinidae: Trachyporinae: Vateginosus (Hymenoptera: Formicidae). Insectes Soc 42:31-44 sini). Pan Pacific Entomol 45:269-281 Beugnon G, Dejean A (1992) Adaptative properties of the chemi- Schroth M, Maschwitz U (1984) Zur Larvalbiologie und Wirtsflndung von Maculinea teleius (Lepidoptera: Lycaenidae), eines cal trail system of the African weaver ant Oecophylla longiParasiten von Myrmica laevinodis (Hymenoptera: Formicnoda Latreille (Hymenoptera, Formicidae, Formicinae). Insecidae). Entomol Gen 9:225-230 tes Soc 39:341-346 Bristow CM (1987) On butterflies and ants: when to be a friend. Thomas J (1984) The behaviour and habitat requirements of Maculinea nausithous (the dusky large-blue butterfly) and M. teleius Trends Ecol Evol 2:29-30 (the scarce large-blue) in France. Biol Conserv 28:325-347 Cammaerts R, Detrain C, Cammaerts MC (1990) Host-trail following by the myrmecophilous beetle Edaphopaussus favieri Thomas JA, Elmes GW, Wardlaw JC, Woyciechowski M (1989) Host specificity among Maculinea butterflies in Myrmica ant (Fairmaire) (Carabidae Paussinae). Inseetes Soc 37:200-211 nests. Oecologia 79:452-457 Cottrell CB (1987) The extraordinary Liphyra butterfly, Transvaal Togerson RL, Akre RD (1969) Reproductive morphology and beMus Bull 22:5-12 haviour of a thysanuran, Trichatelura manni, associated with Dejean A, Beugnon G (1991) Persistent intercolonial trunkroutearmy ants. Ann Entomol Soc Am 62:1367-1374 marking in the African weaver ant Oecophylla longinoda LatVander Meer RK, Wojcik DP (1982) Chemical mimicry in the myreille (Hymenoptera, Formicinae). Ethology 88:89-98 rmecophilous beetle Myrmecaphodius excavaticoUis. Science Elmes GW, Thomas JA, Wardlaw JC (1991) Larvae of Maeulinea 218:806-808 rebeli, a large-blue butterfly, and their Myrmica host ants: wild

by the workers to their feeding places or back into the nests (Fiedler and Maschwitz 1989). It is not only simple attraction that Euliphyra caterpillars exert over O. longinoda workers. A more complex phenomenom is occurring. The mantle of the larva seems to be very attractive to O. longinoda workers, especially the dorsal part which contains two crests in E. teucyana and three in E. mirifica (Lamborn 1913; Eltringham 1913, 1921; Farquharson 1921; personal observation). It would be interesting to determine if the dorsal crests contain specialized, hair-derived epidermal glands, i.e. lenticles or pore cupolas, as occur in many Lycaenidae, the attractive effect of which has been demonstrated on ants (Malicky 1969; Henning 1983; Kitching and Luke 1985; Cottrell 1987). When placed on a trail of a society of O. longinoda, alien Euliphyra larvae are rapidly accepted. This is surprising insofar as the African weaver ants are fiercely aggressive especially towards alien conspecifics (H6Ildobler 1979). This underlines the protective role of the mantle and perhaps of allomones inducing appeasement. During the first contacts between ants and caterpillars, the cuticle of the caterpillars might become soaked with the ants' odours, as occurs in Martinezia dutertrei, a myrmecophylous coleopteran (Vander Meer and Wojcik 1982). This hypothesis explains the acceptance of alien caterpillars both when encountered on trails and when placed directly inside a nest. In the latter case, nevertheless, one caterpillar was killed and another transported to the outside of the nest.