Biological Journal of the Linnean Socicly (1993), 49: 229-238. With 2 figures
Comparison of acoustical signals in Maculinea
butterfly caterpillars and their obligate host
P. J. DEVRIES* AND R. B. COCROFT
Department of ,zbology, University of Texas, Austin, Texas 78712, U.S.A.
Institute of Terrestrial Ecology, Furzebrook Research Station, Wareham,
Dorset BH20 5AS
Received 6 February 1992, accepted for publication I2 May I992
An acoustical comparison between calls of parasitic butterfly caterpillars and their host ants is
presented for the first time. Overall, caterpillar calls were found to be similar to ant calls, even
though these organisms produce them by different means. However, a comparison of Maculinca
caterpillars with those of M p i c a ants produced no evidence suggesting fine level convergence of
caterpillar calls upon those of their species specific host ants. Factors mediating the species specific
nature of the Maculincn-Myrmica system are discussed, and it is suggested that phylogenetic analysis
is needed for future work.
ADDITIONAL KEY WORDS:-Lycaenidae
- Formicidae - symbiotic association - evolution
Materials and methods
Arthropods from a diversity of phylogenetic lineages form symbiotic
associations with ants that may range from parasitism to mutualism. Some
arthropod symbionts produce semiochemical secretions that aid in maintaining
the symbioses (Holldobler, 1978; Vander Meer & Wojcik, 1982), while others
produce food secretions to ants that are also considered important in
maintaining these symbioses (Way, 1963; Maschwitz, Fiala & Dolling, 1987;
Fiedler & Maschwitz, 1988, 1989; DeVries, 1988; DeVries & Baker, 1989). Ants
*Current address for correspondence: Museum of Comparative Zoology, Harvard University, Cambridge
MA 02138, U.S.A.
0024-4066/93/070229 + 10 S08.00/0
1993 The Linnean Society of London
P. J. DEVRIES ET AL.
primarily communicate by using chemical and trophobiotic cues among
themselves (reviewed in Holldobler & Wilson, 1990), but in addition, substrate-
borne vibrations are also important in ant communications systems (Mark1 &
Holldobler, 1978; Baroni-Urbani, Buser & Schillinger, 1988). Recent work on
ant-associated butterfly caterpillars, suggests that substrate-borne calls may also
be important in the evolution of butterfly-ant symbioses (DeVries, 1990, 1991a).
Among butterflies, the habit of forming symbioses with ants is found only in
the families Riodinidae and Lycaenidae. Typically these symbioses are based on
the caterpillar’s provision of food secretions with specialized glands to ants in
exchange for protection against enemies (Pierce et ‘al., 1987; DeVries, 1988,
1991b; Thomas et al., 1989). In addition, these caterpillars may also possess other
specialized behaviours and morphological features that play a role in
maintaining symbiotic association with ants (Ross, 1966; DeVries, 1988, 1990;
Fiedler & Maschwitz, 1988), and it is likely that these characters have all been
selected for directly by ants (Cottrell, 1984; DeVries, 1988, 1991a, c).
Most symbioses between butterfly caterpillars and ants are facultative with
respect to the partners involved. For example, a recent survey indicated that a
given species of caterpillar will typically be tended by a varying number of
secretion-harvesting ant species, depending on the habitat and geographical
locality (DeVries, 1991 b). Cases of obligate, species-specific symbioses between
caterpillars and ants, however, have evolved, and the best known case involves
the European lycaenid genus Maculinea whose caterpillars parasitize colonies of
the ant genus Myrmica.
After quickly developing through three instars on an initial foodplant, the
final instar caterpillars of Maculinea fall to the ground where they are exposed to
foraging Myrmica workers (Thomas, 1977, 1992; Elmes, Thomas & Wardlaw,
1991). When a caterpillar is found by a Myrmica worker, there ensues an
adoption process, and the adopted caterpillar is carried into the ant’s nest and
placed in the brood chamber. Depending on the species, the Maculinea caterpillar
then may live as an obligate carnivore on ant larvae, or is fed directly by the ants
for the next 10 months (Thomas & Wardlaw, 1990; Thomas et al., 1991; Elmes et
al., 1991). Although in this system, any Myrmica ant species will take any species
of Maculinea caterpillar into the nest, as a rule each species of caterpillar can
survive only in the nest of a specific species of Myrmica (Thomas et al., 1989).
Thus, there should be strong selection on Maculinea caterpillars to form
symbioses with the appropriate species of Myrmica ant.
The use of sound in ant communication systems may function to enhance
alarm reaction, recruit workers, or enhance other means of communication and
behaviours (Baroni-Urbani et al., 1988; Holldobler & Wilson, 1990). As
caterpillar calls are known only among butterfly species that form symbioses
with ants, these calls are thought to function in the maintenance of ant
symbionts (DeVries, 1990, 199 1 a). From these observations arises the question of
whether caterpillar calls can function to attract specific taxa of ants.
Investigation of the obligate, species-specific nature of the Maculinea-Myrmica
system provides the first opportunity to compare caterpillar and calls, and test
whether there has been evolution within a genus of butterfly caterpillars to
produce calls that mimic those of their specific host ants.
This preliminary study examines the acoustical calls of the Maculinea-Myrmica
system in four ways. First, to determine general call convergence, the call
CATERPILLAR AND ANT ACOUSTICAL SIGNALS
characteristics of Maculinea and Myrmica, plus several distantly related
caterpillars and ants are compared statistically with respect to pulse rate, pulse
length, dominant frequency and bandwidth. Secondly, to examine if there has
been convergence between the calls of Maculinea and Myrmica, the acoustical
characteristics of four species of Maculinea caterpillar calls are compared with the
calls of four species of their obligate host Myrmica ants. Third, we discuss the
general function of caterpillar calls with respect to forming symbioses with ants,
and specifically what mechanisms are likely to mediate the obligate, parasitic
symbiosis that Maculinea caterpillars form with Myrmica ants. Finally, we suggest
that understanding the evolution and function of calls in the Maculinea-Myrmica
system depends upon developing a phylogeny for both butterflies and ants.
MATERIALS AND METHODS
Caterpillar and ant calls were detected and recorded with a Bennett-Clark
( 1984) particle-velocity microphone using the methods described in DeVries
(1991d)) and analysed subsequently with a Data 6000A Universal Wave Form
analyser with a sampling rate of 10 kHz. The caterpillars and ants used in this
study were obtained from stocks maintained at the Institute for Terrestrial
Ecology (England) , or from field-collected material originating from Westerwald
(Germany), Barcelona (Spain), California (U.S.A.) and Barro Colorado Island
The pulse rate, dominant frequency, pulse length and bandwidth at 10 dB
below peak were measured for each call of five Maculinea and four Myrmica
species. Additionally, call characteristics of Lysandra hispana and Plebulina
emigdonis caterpillars and the ant Atta cephalotes were measured for comparative
purposes. Descriptive statistics were done by taking ten measurements for each
variable from a single individual per species, and then computing the means and
standard deviations (Table 1). Pooled calls of all caterpillar and ant species were
compared with a Mann-Whitney U-test for each variable measured.
The similarity between Maculinea caterpillar and Myrmica ant calls was
estimated in two ways. First, in order to reduce the number of variables, the calls
Abbreviations: pr = mean pulse rate, df = mean dominant frequency, pl = mean pulse length,
bw = mean bandwidth. (SD) = standard deviation for call characteristics, and reported to the
1. Measurements of call characteristics for seven caterpillar species and five ant species.
( f S D )
( f S D )
M m i c a rubra
53 I .2
42 I .20
P. J. DEVRIES ET AL.
of four caterpillar species and their four obligate host ants were compared using
principal component analyses. The two components which explained most of the
variance were then plotted to determine whether a species was most similar to its
obligate host for those variables. Secondly, using the raw data the calls were
compared by tabulating how many times each variable measured for a given
caterpillar species was more similar to a wrong host ant than to the correct one.
We found that, overall, caterpillar and ant calls have both similarities and
differences with respect to the variables measured (Fig. 1, Table 1). The mean
pulse rates for caterpillar calls ranged from 3.45 to 15.55 pulses s-' and ant calls
ranged from 6.5 to 42.55 pulses s-', Mean dominant frequencies for caterpillar
calls ranged from 118.2 to 1501.9 Hz, and ant calls from 531.2 to 1249.6 Hz.
Mean pulse length ranged from 1.8 ms to 76.0 in caterpillar calls, and 18.0 to
79.3 ms in ant calls. Mean bandwidth ranged from 164.1 to 1734.5 Hz in
caterpillar calls, and 113.8 to 1303.9 Hz in ant calls.
Our measurements indicated that caterpillar call characteristics may mimic
the general calls of ants with respect to certain variables. Pooled comparisons
showed that ant calls had a significantly higher pulse rate and greater
bandwidth than did the caterpillar calls (Fig. 1, Table 2). However, the pulse
length and dominant frequencies did not differ significantly between caterpillars
and ants (Fig. 1, Table 2). No ant larva or pupa of any species produced a call.
The principal components analysis of four M u c u l i n e u - M y n n i c u species pairs
produced no compelling evidence suggesting fine-level convergence of particular
caterpillar calls upon those of their specific host ant species. Call convergence
was suggested in only one of the four host-parasite pairs; in the other three cases,
the parasitic caterpillars plot more closely to 1-3 incorrect ant species than to an
appropriate host (Fig. 2). In this set of data principal component I was
explained primarily by variation in pulse length, pulse rate and bandwidth, and
principal component I1 was explained primarily by variation in dominant
frequency (Table 3). The first two principal components explained 76.9% of the
variance in the original data; two additional components explained only a small
portion of variance and were not included. However, while caterpillar and ant
calls were not convergent at the level of species pairs, this analysis showed that
M u c u l i n e u caterpillar and M y r m i c a ant calls are, in general, broadly similar with
2. A pooled comparison of all caterpillar and ant calls with
respect to four variables. All degrees of freedom = 1
Variable Croup Rank sum
Pulse length Caterpillar (40)
Dom. frequency Caterpillar (40)
Pulse rate Caterpillar (40)
Bandwidth Caterpillar (38)
14.0 < 0.001
425.0 < 0.005
CATERPILLAR AND ANT ACOUSTICAL SIGNALS
1 r M. rebeli
- P. emigdionis
I I I
- -1 2
8 1 - M. nausithous
1 - M. releius
I I 1
1 r L. hispanus
r MY. ruginodis
r My. rubra
I1 I I
Figure 1. Waveforms of substrate borne lycaenid butterfly caterpillar and myrmicine ant calls.
Notice the general similarity of the form of most caterpillar and ant calls. Abbreviations: M =
Maculinea, L = Lysandra, P = Plebulina, My = Myrmica, Atta = Atta ccphalotes. Note-sonograms of
Myrmica rubra calls have been published by Le Roux (1977) under the name M. lacvinodis.
respect to the variables we measured, and thus making it likely that caterpillar
calls can be detected by ants.
A tabular comparison of single variables also expresses a lack of fine-level
similarity between calls of caterpillars and their appropriate host ant species
(Table 4). For each variable, the proportion of times that a caterpillar species is
+ I 1.
0 8 -0.6
.- . . . -
Principal Component I1
Figure 2. Principal component plot of the calls of four caterpillar species and their obligate host ant
species. If Maculinca calls have converged upon those of their correct host M ’ i c u
should cluster together. Note that while the distribution of caterpillar and ant calls overlap on both
axes, there is a disparity with respect to correct host-parasite associations. Upper and lower cast
forms of the same letters represent correct species pairs. Abbreviations: a = Maculinca arion, A =
M F i c a sabukti, b = Maculinca alcon, B = M m i c a ruginodis, c = Maculinca tcleius, C = Myrmica
scabrinodis, d = Maculinca nausithous, D = M ’ i c a rubra.
ant species they
variables of obligate Maculinca and Mymica species pairs
3. Principal-component loadings for the four call
Variable I I1
Proportion of variance explained
Cumulative variance explained
on single variables. Each value represents the number of
times the call of a given caterpillar was more similar to
the wrong host ant than to the correct one. A score of 0
indicates a caterpillar call was most similar to the correct
host. The correct caterpillar-ant
4. Comparison of caterpillar and ant calls based
CATERPILLAR AND ANT ACOUSTICAL SIGNALS
most similar to its appropriate host ant is about that expected by chance in a set
of four species pairs (correct caterpillar-ant association is arion-sabuleti,
and teleius-scabrinodis) . The expected proportion
is 4 in 4, whereas the observed proportions were 2 in 4 for pulse length, and 1 in
4 for pulse rate, dominant frequency and bandwidth.
Acoustical and vibrational signals can be fundamental to species recognition
in insect communication systems (Markl, 1983; Gogala, 1985; Dambach, 1989).
For example, both air and substrate borne call components are integral to
courtship and mate recognition in some tropical katydid species (Morris, 1980;
Belwood, 1990). Vibrational signals are used typically in ant communication
systems upon finding food, and when an individual ant is stressed-as
under attack or buried in the soil (Markl & Holldobler, 1978; Baroni-Urbani et
al., 1988; Holldobler & Wilson, 1990). Here the ant calls function to recruit
nestmates to food resources, help defend the colony, or to unearth a trapped
worker. Eliciting any of these responses might benefit a caterpillar that gains
symbiotic protection from ants by maintaining them nearby (DeVries, 1990,
199 1 a, b) .
The behavioural interactions between Maculinea caterpillars and Myrmica ants
are complex, both before adoption into the ant colony and during the ensuing 10
months underground. As obligate parasites of one species or genus of ant, one
might expect the calls produced by caterpillars to resemble closely the calls of
their host ant species. The pulse lengths and dominant frequencies of caterpillar
and ant calls (Fig. 1, Tables 1, 2) suggest that Maculinea calls have converged in
a general manner on those produced by members of the genus Myrmica.
However, our analysis showed little or no fine-level convergence between
caterpillar calls and those of their obligate host ant species (Figs 1, 2,
Tables 1 4 ) . Had the calls of a particular parasitic Maculinea species converged
on those of it specific Myrmica host, it should have plotted more closely to its host
species than to other ant species in whose nest the caterpillars will not survive
(i.e, Fig. 2), or shown greater similarity of characteristics than simple chance
alone (Table 4). Thus, even with strong selection on Maculinea caterpillars to
parasitize a specific host ant species (e.g. Thomas et al., 1989; Elmes et al., 1991),
there was no compelling evidence detectable in this study to suggest that
caterpillar calls are critical to attracting the correct host ant. However, as ant
recording was done by restraining individual ants, we do not know whether an
unstressed ant would produce call characteristics more similar to those of
This preliminary study provides a first step toward understanding the role
Maculinea caterpillar calls play in their symbioses with Myrmica ants. We have
established that, even though they produce calls by different means, caterpillar
calls are surprisingly similar to the stridulations of adult ants (Fig. 1, Tables 1,
2). Thus, this is the first occasion in which Maculinea calls are shown to mimic
those produced by adult Myrmica ants. Ant vibrational signals ranging from a
few hundred Hz to a few kHz can elicit behavioural changes in coworkers
(Markl, 1970; Fuchs, 1976a, b; Markl & Holldobler, 1978). As these
frequenceies are encompassed by the calls of Maculinea
P. J. DEVRIES ET AL.
myrmecophilous caterpillars (Table 1 ; DeVries, 199 1 a), caterpillar calls may
elicit responses in ants similar to the attraction shown toward vibrational signals
produced by nestmates. The exact function of Maculinea calls in mediating their
symbioses with ants remains unknown, but Maculinea calls probably function in a
manner suggested for other butterfly-ant systems (DeVries, 1988, 1990, 1991 a);
to enhance symbiotic association without influencing the species-specific nature
of the interaction.
Once an ant has found a caterpillar, several observations suggests that
chemical, rather than acoustical signals will prove fundamental to mediating the
species-specific duping of Myrmica ants by Maculinea caterpillars. First, the
chemical recognition signals among Myrmica species consist of essentially the
same chemical components, but vary in concentration between species
(Cammearts el al., 1978; Winterbottom, 1980). This similarity helps explain why
foraging workers will retrieve the larvae of any Myrmica species encountered if
they are placed above ground (G. W. Elmes, personal communication).
Secondly, individual Myrmica workers treat both their own larvae, and the
caterpillars of Muculinea in a similar fashion when removed from the nest under
laboratory conditions-they are dragged back to the brood chamber and
deposited among developing ant larvae. Ants treat caterpillars as if they were
larvae from an ant nest, even though a caterpillar may have ten times the mass
of the ant. The analyses of call characteristics, together with the observations
above strengthen the idea that each species of Muculinea caterpillar survives only
in the nest of a single Myrmica species because of particular chemical cues used by
ants to recognize and adopt their sisters, larvae, and possibly queens
(Holldobler, 1978; Thomas et ul., 1989; Elmes et al., 1991).
The Maculinea-Myrmica system has provided a useful model for understanding
the ecology and evolution of specificity in caterpillar-ant symbioses (e.g. Thomas
et al., 1989; Elmes el al., 1991). Here we used this system to consider ideas about
the evolution and function of calls in caterpillar-ant symbioses. Our preliminary
results provide little indication for convergence of parasitic caterpillar calls upon
those of obligate ant hosts. Perhaps a study with a larger sample and more
detailed analyses would show otherwise. However, we feel that our
understanding of call evolution in Maculinea caterpillars is impeded by a lack of
phylogenetic information. There are six species in the genus Muculinea (Thomas,
1992), and 40-50 species in the genus Myrmica (Bolton & Collingwood, 1975),
neither group with well-defined relationships. When resolved phylogenetic
hypotheses and sister group relationships of both Muculinea and of Myrmica
become available, the questions presented here can be addressed directly. For
example, a phylogeny based on morphological or biochemical charcters will
make it possible to track the evolution of caterpillar call characteristics on the
one hand, and those of ants on the other. This information can, in turn, then be
used realistically to search for parallel transformation in Maculineu and Myrmica
with respect to their calls by plotting call characteristics on cladograms. Given
the intimate relationship of Maculinea and Myrmica, we would not be surprised to
find that Myrmica calls resemble those of Maculinea more closely than those of the
sister group of Maculinea, whatever that group may prove to be. Finally, a
phylogeny will also provide a framework for investigation of the evolution of
semiochemical cues in this host-parasite system. Thus, our opinion is that the
resolution of these types of questions must await future phylogenetic information.
CATERPILLAR AND ANT ACOUSTICAL SIGNALS
We thank the faculty of the Institute for Terrestrial Ecology, and G. Ballmer,
K. Fiedler, K. Schurian and J. C. Wardlaw for providing caterpillars, and
M. Ryan for use of sound analysis equipment. The comments of
C. Baroni-Urbani, W. Brown Jr., J. Bull, T. Eisner, N. Greig, D. Grimaldi,
G. W. Elmes, M. Ryan and D. Stradling improved the manuscript. RBC was
supported by an NSF predoctoral fellowship. PJD thanks the MacArthur
Foundation for support, and dedicates this paper to M. Allison, I. Baker and
Barod-Urbpni C, Buscr MV, Schilligcr E. 1988. Substrate vibration during recruitment in ant social
organization. Insects Sociaux 35: 241-250.
Bolton By Collingwood CA. 1975. Handbooks for the identification of British insects. Hymenoptera:
Formicidae. The Royal Entomological Sociey of London 6: 1-34.
Belwood JJ. 1990. Anti-predator defence and ecology of neotropical forest katydids, especially the
Pseudophyllinae. In: Baily WJ, Renu DCF, eds. The Tettigoniidae: biology, systematics and evolution. Bathurst
Crawford House Press, 8-26.
Benuct-Clark HC. 1984. A particle velocity microphone for the song ofsmall insects and other measurements.
Journal of Experimental Biology 108: 459-463.
Cammacrts MC, Inwoad MR, M o w ED, Parry K, Tyler RC. 1978. Comparative study of the
pheromones emitted by workers of the ants Myrmica rubra and Myrmica scabrinodis. Joumal of Insect Physiology
Cottrcll CB, 1984. Aphytophagy in butterflies: its relationship to myrmecophily. <oologicnl Journal of the
Linnean SocieQ 79: 1-57.
Dunbach M. 1989. Vibrational Responses. In: Huber F, Moore TE, Loher V, eds. Cricket Behavior and
Neurobiology. Ithaca: Comstock Press, Cornell University Press, 178-197.
DcVricm PJ. 1988. The larval ant-organs of Thisbe irenea (Lepidoptera: Riodinidae) and their effects upon
attending ants. <oological journal of the Linnean Socicp 91 379-393.
JhVricm PJ. 1990. Enhancement of symbioses between butterfly caterpillars and ants by vibrational
communication. Science 248: 1104-1 106.
DcVrics PJ. 1991..
Call production by myrmecophilous riodinid and lycaenid butterfly caterpillars
(Lepidoptera); morphological, acoustical, functional, and evolutionary patterns. American Museum Novitates
3 0 2 5 : 1-23.
DcVricm PJ. 1991b. Mutualism between Thisbe irenea larvae and ants, and the role of ant ecology in the
evolution of myrmecophilous butterflies. Biological Joumal of the Linnean Sociely 43: 179-195.
DcVricm PJ, 1991c. Evolutionary and ecological patterns in myrmecophilous riodinid butterflies. In Cutler R,
Huxley C, eds. Intcruction between ants and plants. Oxford: Oxford University Press, 143-156.
Devrics PJ. 1991d. Detecting and recording the calls produced by butterfly caterpillars and ants. Journal of
Research on the Lcpidoptcra, 281 258-262.
DcVricm PJ, Baker I . 1989. Butterfly exploitation of an ant-plant mutualism: adding insult to herbivory.
journal of the New rork Entomological Sociey 97: 332-340.
Elmem CW, Thomas JA, Wardlaw JC. 1991. Larvae of Maculinea rcbeli, a large-blue butterfly, and their
Mynica ants: wild adoption and behavior in ants nests. Joumal of xoology, London 223: 447460.
Picdlcr K, Maschwitz U . 1988. Functional analysis of the myrmecophilous relationships between ants
(Hymenoptera: Formicidae) and lycaenids (Lepidoptera: Lycaenidae) . I I. Lycaenid larvae as trophobiotic
partners of ants-a quantitative approach. Oecologia 75: 204-206.
Ficdlcr K, Maschwitz U. 1989. Functional analysis of the myrmecophilous relationships between ants
(Hymenoptera: Formicidae) and lycaenids (Lepidoptera: Lycaenidae). I. Release of food recruitment in
ants by lycaenid larvae and pupae. Ethologv 8or 71-80.
Fuchm S. l976a. The response to vibrations of the substrate and reactions to the specific drumming in colonies
of carpenter ants (Camponotus, Formicidae, Hymenoptera). Behaviorial Ecology B Sociobiology 1: 155-1 84.
Fuclu S. 1976b. An informational analysis of the alarm communication by drumming behavior in nests of
carpenter ants (Camponotus, Formicidae, Hymenoptera). Behaviorial Ecology & Sociobiology 1: 3 15-366.
Gogala M. 1985). Vibrational communication in insect. (biophysid and b C h . v i o d ampects). In:
K.lmring K, Elmer N, eds. Acoustic and uibrational communication in insects. Berlin & Hamburg: Verlag Paul
Iuilldobler B . 1978. Ethological aspects in chemical communication in ants. Advances in the Stu4 of Behavior 8:
238 Download full-text
Holldoblcr By Wilson EO. 1990. The ants. Cambridge: Harvard University Press.
Lc Row C. 1977. Note sur I’organisation temporelle des sequences de stridulation de Myrmica laevonodis Nyl.
(Hymenopteres, Formicidae). Comptes Rendus Academic des Paris, Serie D 2851 415418.
M.rkl H . 1970. Die Verstandigung durch Stridulationssignale bei Blattschneiderameison. 111. Die
Empfindlichkeit fir Substratvibrationen. <eitschrifrjk Vergleichcnde Physiologic, 6% &37.
Markl H . 1983. Vibrational communication. In: Hiiber F, Markl H, eds. Neuroethology and behavioral physiology.
Berlin & Heidelberg: Springer Verlag, 332-353.
b k l H, HIlldobler B. 1978. Recruitment and food retrieving behavior in Novomcssor (Formicidae,
Hymenoptera). 11. Vibration signals. Behaviorial Ecology B Sociobiology 4: 183-2 16.
Maschwitz U . Fi.l. B, Dolling WR. 1987. New trophobiotic symbioses of ants with South East Asian bugs.
Journal o f Natural History 21: 1097-1 107.
Moms CK. 1980. Calling display and mating behavior of Copiphora rhinoceros Pictet (Orthoptera:
Tettigoniidae). Animal Behavior 28: 42-51.
Pierce NE, Kitchhg RL, Buckley RC, Taylor MFJ, h o w ELF. 1987. The costs and benefits of
cooperation between the Australian lycaenid butterfly, Jalmenus evagoras, and its attendant ants. Behavzorial
Ecology €9 Sociobiology 211 237-248.
Ross CN, 1966. Life history studies of a Mexican butterfly. IV. The ecology and ethology of Anatole rossi, a
myrmecophilous metalmark. Annals o f the Entomological Socieb o f America 591 985-1004.
Thomas JA, 1977. Second Report on the ecology and conservation o f the large blue butterJly. Institute of Terrestrial
Thomas JA. (1992). Relationships between butterflies and ants. In: Dennis RHL, ed. The ecolagy o f British
buftcgies. Oxford: Oxford University Press, 149-155.
Thomas JA, Wardlaw JC. 1990. The effect of queen ants on the survival of Maculinea arion larvae in Mynnica
ants nests. Oecologia 85: 87-91.
Thomas JA, Elmer CW, Wardlaw JC, Woyciechowmki M. 1989. Host specificity among Maculinea
butterflies in Myrmica ants nests. Occologia 79: 452457.
Thomas JA, Mnnguh MI, Mutin J, Elmcs CW. 1991. Basal hatching by Maculinea butterfly eggs: a
consequence of advanced myrmecophily? Biological Journal o f the Linnean SocieQ 44: 175-184.
Van Der Mccr RK, Wojdk DP. 1982. Chemical mimicry in the myrmecophilous beetle Myrmecaphodius
excauaticollis. Science 2181 806-808.
Way MJ. 1963. Mutualism between ants and honeydew producing Homoptera. Annual Review o f Entomology 8:
Winterbottom 6. 1980. The chemical basis for species and colony recognition in three species o f mynicine ant
(Hymenoptera, Formicidae). Unpublished Ph.D Thesis, University of Southampton.
P. J. DEVRIES ET AL.