Social organization in two primitive attine ants,
Cyphomyrmex rimosus and Myrmicocrypta ednaella,
with reference to thier fungus substrates
and food sources
Takahiro Murakami and Seigo Higashi
Graduate School of Environmental Earth Science
Hokkaido University, Sapporo 060, JapanCorrespondence: Takahiro Murakami
Graduate School of Environmental Earth Science,
Hokkaido University, Sapporo 060, Japan
Fax(011)747-9780
total 13 pages, 2 tables and 5 figures
ABSTRACT
In the Neotropical rainforest of Barro Colorado Island, Panama, social organization and behavior were observed in two primitive attine ant species, Myrmicocrypta ednaella and Cyphomyrmex rimosus . Both species took nutrients from mycelia on the fungus garden (i.e. mycophagy), and from plant nectar and saps which they collected outside the nest (i.e. phytophagy). They also obtained alimentary liquid by soliciting nestmates (i.e. stomodeal trophallaxis). Queens and larvae were wholly mycophagous, while older workers were much dependent on nectar, sap and alimentary liquid and younger workers were mostly mycophagous but only partly phytophagous. M. ednaella used wood chips as substrate of fungus garden and its fungus-growing behavior was similar to those hitherto observed in other primitive attine species. By contrast, the behavior of C. rimosus was unique in utilization of crop liquid as a substrate. In the rain forest, C. rimosus workers frequently foraged outside the nests to collect nectar and sap, most of which are probably regurgitated for cultivating the fungus.
KEY WORDS
Attini, Cyphomyrmex rimosus, Fungus-growing ants, Mycophagy, Myrmicocrypta ednaella
INTRODUCTION
The tribe Attini is new-world, mainly Neotropical, in distribution. All ants of this tribe cultivate fungus gardens and depend upon them for food. Therefore, many myrmecologists have been attracted to these ants, and there have been some specific hypotheses for the origin of the fungus-growing behavior (von Ihering 1894; Forel 1902; Weber 1956a, 1958 and 1972; Garling 1979). However, none of these hypotheses has not been testified yet, thus the evolution of mycophagy in attine ants remains controversial. This is partly due to the scarcity of studies on the behavior of primitive attine species, most of which occur cryptically in Neotropical rainforests. Among 12 genera of Attini, Cyphomyrmex has long been considered as the most primitive because the shape and position of the antennal scrobes apparently link Cyphomyrmex with various non-attines (e.g. Forel 1885 and 1892). Moreover, Weber (1941, 1958 and 1972), Wilson (1971) and H嗟ldobler and Wilson (1990) regarded Cyphomyrmex rimosus as the most primitive species based on its yeast-cultivating behavior. On the other hand, a largely ignored alternative hypothesis, that Myrmicocrypta is the most primitive attine genus, was first proposed by Wheeler (1910) on the basis of the queen/worker polymorphism of this genus, as well as the polymorphism within the worker caste. Kusnezov (1961 and 1963) supported this idea mostly by the characters of wing venation and male antennae. Furthermore, recently, Schultz and Meier (1995) reconsidered the phylogeny of attine genera based on the morphology of larvae and assumed that Myrmicocrypta is the most primitive genus. In the present study, therefore, we aimed to compare the behavior of Cyphomyrmex rimosus and Myrmicocrypta ednaella, with special reference to their fungus substrates and food sources.
METHODS
In October and November of 1993 and 1994, colonies of C. rimosus and M. ednaella were collected in the rainforest of Barro Colorado Island, Panama. Immediately after collection, the numbers of queens, workers, pupae, larvae, and eggs were counted in the laboratory. For some colonies, ants were fixed and preserved in 80% alcohol to examine the conditions of ovary, spermatheca and yellow bodies, which are egg remainings indicating the occurrence of oviposition. For some other colonies, ants were kept alive in an artificial nest (50 x 25 x 5 mm), top of which was covered by a glass slide in order to observe the behavior of ants. During the period of observation, food and nest materials, such as nectar, bananas, dead vegetables and insect feces were always provided in the foraging arena connected to the artificial nest.
Behavioral repertory and polyethism were studied with two M. ednaella colonies (M1 and M2) and five C. rimosus colonies (C1-C5). Workers were divided into three age groups according to their body color (yellow colored workers as young; brown-colored ones as middle-aged; and dark-colored ones as old). In each colony, the workers and queen were individually marked with colored paint, and their behavior was observed in the artificial nest for a total of 10 to 65 hours by scanning the activities of each individual 15 minutes for each. After the observation of all workers, the queen and larvae were also observed for 10 hours in an M. ednaella colony (M3) and for 15 hours in a C. rimosus colony (C6) with particular reference to their feeding.
The construction of the fungus garden was observed for a total of 30 and 40 hours in three M. ednaella colonies (M4, 5 and 6) and four C. rimosus colonies (C7-C10), respectively. In this observation, we focused on the materials which workers used for the substrates of the fungus garden.
Finally, we also collected foraging workers of C. rimosus in the rainforest in late November, 1994. The captured workers were dissected in laboratory and examined whether their crop was filled with liquid or not. Some species of flowers were also collected in the field and examined in the laboratory to see whether or not C. rimosus workers were in these flowers to collect nectar from them.
RESULTS
Myrmicocrypta ednaellaWe collected seventeen M. ednaella colonies, which nested about 10 cm below ground. In each nest, an almost spherical-shaped fungus garden was found. It was about 50 mm in diameter and composed of many wood chips which were occasionally mixed with a few pieces of insect corpses. The number of adult workers ranged from 10 to 190 with an average of 60.7 ｱ SD 42.3. Dissection revealed that the workers lacked spermathecae and had only one pair of atrophied ovarioles containing no eggs. This species was always monogynous, with a single queen which was inseminated and had 3 + 3 developed ovarioles.
Behavioral repertories
In the laboratory observation of M. ednaella, we distinguished 11 activities. Since there was no significant difference in the characteristics of polyethism between colonies M1 and M2, the data of the two colonies were combined (Table 1). The queens were always inside the nest, spending 53.7% of their time in doing nothing (= repose). It was followed by mycophagy (20.0%) in which the queens stayed on the fungus garden and fed on the mycelium. During total observation for 78 hours, oviposition by the queen was observed only once. On the other hand, workers accomplished a variety of tasks, including fungus care, grooming, nest defense and extranidal foraging. Unlike workers of other myrmicine tribes, the workers of M. ednaella rarely took care of brood, probably because the larvae of attine species are placed in the fungus garden and can eat mycelium without assistance of workers.
Polyethism
Based on the body color, workers of colony M1 were divided into six young, 12 middle-aged and five old individuals. They showed a remarkable age-polyethism (Table 1). Young workers were most frequently engaged in fungus care (percentage of acts = 58.2), while old workers performed extranidal activities such as collecting wood chips for the fungus garden (percentage of acts = 56.5). Middle-aged workers showed a broad range of activities, their characteristics being intermediate between the young workers and the old workers. These three groups were clearly distinguished from each other by discriminant analysis (Fig. 1). The similar result was obtained also in the discriminant analysis of colony M2.
The queens and young workers always groomed themselves before stepping into the fungus garden and frequently did so when feeding on mycelium. Workers guarding at the nest entrance groomed the nestmates which returned from extranidal tasks. Returning workers also did self-grooming in the nest.
Feeding
For the queens and larvae, the feeding behavior was observed 85 and 29 times, respectively, indicating that they depended exclusively on mycelia as nutrition (Fig. 2). No trophallaxis was observed between worker and larva. However, the importance of mycophagy is greatly reduced for nutrition of workers. In particular, the old workers rarely fed on mycelia. Instead, they depended much on nectar and saps of plants (54.8% of all feeding acts) and on alimentary liquid regurgitated from other old workers (38.1%).Fungus growing
Observation with colonies M4, M5 and M6 clarified the following characteristics of the fungus-growing behavior of M. ednaella. The first stage of fungus growing was cleaning of the nest floor by a worker's licking with the mouthparts. At the same time with this, outside the nests, other workers collected wood chips by crushing dead twigs with their mandibles. They retrieved and piled up the chips on the nest floor which had previously been cleaned. Then, other workers plucked hyphae from the old part of the fungus garden and planted them onto the new wood chips. These workers subsequently manured the fungi by secreting fecal liquid. As the mycelia grew out, workers often cut off long hyphae with their mandibles and replaced old substrates with newly collected wood chips. The new substrate was planted with hyphae which were subsequently manured with faecal secretion.Cyphomyrmex rimosus
A total of 40 colonies were collected from fallen twigs on the forest floor. The number of workers ranged from 7 to 334, with an average of 66.9 ｱ SD 71.1. Although most colonies were monogynous, there were two queens in two colonies each and no queen in four colonies. All the queens collected were found to be inseminated and have well-developed, 3 + 3 ovarioles. Workers were also dissected and they had no spermatheca but had a pair of atrophied ovarioles without any developed eggs. This suggests that the queens monopolize oviposition in this species. In the colonies with two queens, both queens had yellow bodies, indicating that they were equally functional. In the four queenless colonies, there was no egg, indicating that the absence of a queen was not due to mis-sampling but probably due to her natural death.
Behavioral repertories
The behavioral repertory of C. rimosus consisted of 11 activities (Table 2). Although the queens spent 66.0% of her time in repose, their repertory was more worker-like than M. ednaella queens, which rarely performed tasks other than grooming and mycophagy (cf. Table 1). The inseminated queens of C. rimosus spent 4.1 and 3.1% of their acts in fungus care and brood care, respectively. Worker behavior was also different between C. rimosus and M. ednaella; especially, C. rimosus workers frequently licked the body surfaces of larvae and often transported them to fresh areas of the fungus garden where yeast was growing vigorously.
Polyethism
Figure 3 shows the result of discriminant analysis done for colony C1, which contained 17 young, 32 middle-aged and 10 old workers. Young workers were more engaged in the care of fungi and larvae, while older workers performed most of extranidal activities (Table 2). In the discriminant analysis, three age groups were separated from each other with a probability of 82.8%. The similar result was obtained also for the other four colonies (C2-C5). Like M. ednaella, C. rimosus frequently groomed themselves or nestmates, probably to prevent the fungus garden from being contaminated.
Feeding
Food sources in C. rimosus were the yeast growing on the substrate, the nectar and saps of plants, and the alimentary liquid regurgitated from nestmates. The queen and larvae were exclusively dependent on the yeast as their nutrient (Fig. 4). In young and middle-aged workers, mycophagy accounted for 81.8% and 48.3% of their feeding, respectively (Fig. 4). Since these workers frequently engaged in trophallaxis with old workers (22.7% in young workers and 35.3% in middle-aged workers), they probably obtained nutrient also from old workers, who mostly depended on nectar and sap (79.2%). Trophallaxis among old workers was frequent (15.4%) but they have never been observed to receive the nutrient from younger workers. This suggests that the old workers do not get nutrient from the yeast even in an indirect manner via younger workers. Larval mycophagy was recorded 39 times during 15 hours observation with 6 larvae.
Fungus growing
The fungus-growing behavior of C. rimosus was quite different from those so far known in other attines. As shown in Fig. 5, an intranidal C. rimosus worker licked and cleaned a part of the nest floor and put there a piece of insect or millipede faeces that had been previously carried into the nest by an extranidal worker. Then another extranidal worker regurgitated alimentary liquid from her crop onto the surface of faeces. She occasionally mixed her fecal secretion into the alimentary liquid and then left it for one day. When the liquid had dried and became a nodule, another worker again added her two kinds of liquid to it. Out of 93 secretions observed in five colonies, 83 were the regurgitation from the mouthparts and 10 were the anal secretion. Thus it seems that the material for the substrate of the fungus garden of C. rimosus is mostly the crop liquid.
After the secretion and desiccation were alternately repeated, the brown nodule grew into a ball of about 0.8-mm diameter. At this stage the nodule was removed and transferred to the surface of insect or millipede feces on which some already yeast-covered nodule were present. The transporting worker transplanted the yeast by rubbing the new nodule against the neighboring old ones. Within 24 hours, the newly placed nodule became densely covered by the yeast. The intranidal workers maintained this fungus garden by licking and plucking the yeast. When the nodule became too old, workers gnawed off its surface and regurgitated the alimentary liquid onto such a newly exposed area and rubbed it against an yeast-covered nodules again.
In the field, we collected 59 C. rimosus workers which were foraging in the rainforest: 18 of them were found on trees, 37 on the ground and 4 in flowers fallen on the forest floor. Dissecting of these workers revealed that the crop of 20 workers was filled with a liquid. This suggests that the workers of C. rimosus frequently forage for nectar and sap of plants. Laboratory experiments showed that C. rimosus collected nectar from the flowers of four plant species: Psychotria marginata, Erythrin erythrinum, Quassia sp., and Rubiaceae sp.
DISCUSSION
Weber (1958 and 1972) and Schultz and Meier (1994 and 1995) proposed different phylogenies for attine genera. Their phylogenies are different particularly in the phyletic position of Cyphomyrmex and Myrmicocrypta. Schultz and Meier carried out a cladistic analysis on 51 attine species using 44 morphological characters of larvae, and found that Myrmicocrypta larvae have some traits common with the other myrmicine tribes, while Cyphomyrmex larvae are more specialized. In contrast, Weber adopted morphological and behavioral characters of adults, and his conclusion was that Cyphomyrmex, especially C. rimosus, is the most primitive. Hoeldobler and Wilson (1990) discussed the evolution of mycophagy based on the Weber's idea. On the other hand, Chapela et al. (1994) and Hinkle et al. (1994) emphasized Schultz and Meier's phylogeny. We believe , however, that more information is needed; in particular, DNA sequence data are essential.
If the phylogeny of Schultz and Meier is correct, the phylogenetic proximity would not be correlated with the behavioral similarity in the Attini, because the fungus-growing behavior of M. ednaella is not greatly different from those of more advanced genera, e.g. Sericomyrmex and Trachymyrmex, which construct mycelial fungus gardens with fragments of dead plants (Weber 1954, 1956b and 1972). By contrast, even when compared with other congeneric species, the behavior of C. rimosus is unique in the Attini for the following three points. First, C. rimosus does not form mycelial fungus garden but cultures a yeast on small substrates which are placed on the feces of herbivorous insects or millipedes. Second, the workers frequently take care of larvae by licking or transporting them. Finally, unlike the description of Weber (1962 and 1972) that the worker's feces is important in cultivating the mutualistic yeast, the present study demonstrated that the anal liquid has much less involvement in nutrition of the yeast than the crop liquid; or the development of the yeast was limited within the surface on the desiccated nodule of alimentary liquid. In C. rimosus the main substrates for the yeast are apparently the alimentary liquids regurgitated by the ants.
It was long believed that in the Attini all members of colony take nutrient exclusively by eating mycelia. However, Quinlan and Cherrett (1979) found that workers of Atta cephalotes obtain 95% of their carbohydrates from plant products which are collected outside the nests, and only 5% from mycophagy. The present study revealed that the workers, in particular, old individuals of the primitive attines such as M. ednaella and C. rimosus are also more dependent on plant products than on mycelium.
In the present study of M. ednaella and C. rimosus, trophallaxis from workers to larvae was not observed and mycophagy was the unique way of nutrition for larvae. Absence of trophallaxis between worker and larva is common in the member of Attini (Weber 1972; Littledyke and Cherrett 1976; Quinlan and Cherrett 1979). On the other hand, the larvae of most myrmicines are given alimentary liquid by stomodeal trophallaxis from workers (Wilson 1971; Hoeldobler and Wilson 1990). Then a question arises: how has mycophagy evolved from stomodeal trophallaxis in nutrition of attine larvae? Moreover, the fact that ant larvae are much vulnerable to coprophilous or saprophytic fungi (Wilson 1971) leads to the second question: how has the mutualistic fungus been selected from a variety of fungi, most of which seem to be harmful to ant larvae? Although various hypotheses have been proposed for the evolution of mycophagy in attines (von Ihering 1898; Forel 1902; Weber 1956a, 1958 and 1972; Garling 1974), none refers to these questions. In this regard, our finding in C. rimosus that the worker's crop liquid is prerequisite to the growth of the mutualistic fungus is much suggestive. It may be that the ancestor of mutualistic fungus first grew on the substance regurgitated by workers and the larvae fed on such fungus-infested crop liquid. In this beginning stage of mycophagy (and also presently in C. rimosus), chemical substances from worker's exocrine glands might be effective for selection of the edible fungi.
ACKNOWLEDGEMENTS
We wish to express our sincere thanks to Dr. D. Windsor, Smithsonian Tropical Research Institute, for his useful suggestions on the fieldworks in Barro Colorado Island, and to Dr. K. Ohkawara, Graduate School of Environmental Earth Science, Hokkaido University, for his assistance in the fieldworks.
REFERENCES
Chapela, I.H., S.A. Rehner, T.R. Schultz and U.G. Muellerk, 1994. Evolutionary history of the symbiosis between fungus-growing ants and their fungi. Science 266: 1691-1694
Forel, A., 1885. [1884] フudes myrm残ologiques en 1884 avec une description des organes sensoriels des antennes. Bull. Soc. Sci. Natur. 20; 316-380
Forel, A., 1892. Attini und Cryptocerini. Mit. Schweiz. Entomol. Gesell. 8; 344-349
Forel, A., 1902. Beispiele phylogenetischer Wirkungen und R歡kwirkungen bei den Instinkten und dem K嗷penbau der Ameisen als Belege f殲 die Evolutionslehre und die psychophysiologische Identit閣slehre. Psychol. Neurol. 1: 99-110
Garling, L. 1979. Origin of ant-fungus mutualism: a new hypothesis. Biotropica 11: 284-291
Hinkle, G., J.K. Wetterer, T.R. Schultz and M.L. Sogin, 1994. Phylogeny of the Attine ant fungi based on analysis of small subunit ribosomal RNA gene sequences. Science 266: 1695-1697
H嗟ldobler. B and E.O. Wilson, 1990. The Ants. The Belknap Press of Harvard University Press, Cambridge, Mass. 732 pp.
Ihering, H. von, 1898. Die Anlage neur Colonien und Pilzg較ten bei Atta sexdens. Zool. Anzeig. 21: 238-245
Kusnezov, N., 1961. Evolution der Ameisen. Symp. Genet. Biol. Italica 12; 103-121
Kusnezov, N., 1963. Zoogeografica de las hormigas en Sudamerica. Acta Zool. Lilloana 19; 25-186
Littledyke, M. and J.M. Cherrett, 1976. Direct ingestion of plant sap from cut leaves by the leaf-cutting ants Atta cephalotes (L.) and Acromyrmex octospinosus (Reich) (Formicidae, Attini). Bull. ent. Res. 66: 205-217
Quinlan, R.J. and J.M. Cherrett, 1979. The role of fungus in the diet of the leaf-cutting ant Atta cephalotes (L.). Ecol. Entomol. 4: 151-160
Schultz, T.R. and R. Meier, 1994. The phylogeny of the fungus-growing ants (Myrmicinae: Attini) and their fungi. In: A. Lenoir A, Arnold G, Lepageles M (eds) Insectes Sociaux. Univ Paris Nord, 177 pp
Schultz, T.R. and R. Meier, 1995. A phylogenetic analysis of the fungus-growing ants (Hymenoptera: Formicidae: Attini) based on morphological characters of the larvae. Systematic Entomology 20: 337-370
Weber, N.A., 1941. The biology of the fungus-growing ants. Part VII. The Barro Colorado Island, Canal Zone, species. Revista Entomol. 9; 93-130
Weber, N.A., 1954. Fungus-growing ants and their fungi. Anat. Rec. 120: 735
Weber, N.A., 1956 a. Symbiosis between fungus-growing ants and their fungus. Year book Amer. Philos. Soc. for 1955: 153-157
Weber, N.A., 1956 b. Treatment of substrate by fungus-growing ants. Anat. Rec. 125: 604-605
Weber, N.A., 1958. Evolution in fungus-growing ants. Proc. 10th Internat. Congr. Entomol. (1956) (Montreal) 2: 459-473
Weber, N.A., 1962. The fungus-growing ants. Natural History (New York) 61: 45-51
Weber, N.A., 1972. Gardening Ants: the Attines. Memory of The American Philosophical Society, vol.92. The American Philosophical Society, Independence Square Philadelphia. 146 pp.
Wheeler, W.M., 1910, Ants: their Structure, Development and Behavior. Columbia University Press, New York. 663 pp.
Wilson, E.O., 1971. The Insect Societies. Belknap Press of Harvard University Press, Cambridge, Mass. 548 pp.