Commentary

Fungus-farming insects: Multiple origins and diverse evolutionary histories

Ulrich G. Mueller* and Nicole Gerardo*

Section of Integrative Biology, Patterson Laboratories, University of Texas, Austin, TX 78712

bout 40–60 million years before the subterranean combs that the con- An even richer picture emerges when com- Aadvent of human agriculture, three in- struct within the heart of nest mounds (11). paring fungiculture to two other sect lineages, termites, ants, and , Combs are supplied with feces of myriads of known -farming insects, attine ants independently evolved the ability to grow workers that forage on wood, grass, or and ambrosia beetles, which show remark- fungi for food. Like humans, the insect leaves (Fig. 1d). of consumed fungus able evolutionary parallels with fungus- farmers became dependent on cultivated are mixed with the forage in the ter- growing termites (Fig. 1 a–c). crops for food and developed task-parti- mite gut and survive the intestinal passage tioned societies cooperating in gigantic ag- (11–14). The addition of a fecal pellet to the Ant and Fungiculture. In ants, the ricultural enterprises. Agricultural life ulti- comb therefore is functionally equivalent to ability to cultivate fungi for food has arisen mately enabled all of these insect farmers to the sowing of a new fungal crop. This unique only once, dating back Ϸ50–60 million years rise to major ecological importance. Indeed, fungicultural practice enabled Aanen et al. ago (15) and giving rise to roughly 200 the fungus-growing termites of the Old to obtain genetic material of the cultivated known species of fungus-growing (attine) World, the fungus-growing ants of the New fungi directly from termite guts, circumvent- ants (4). Attine colonies typically are World, and the cosmopolitan, fungus- ing laborious nest excavation during collec- founded by a mated female who takes a growing beetles are not only dominant play- tion. Overall, 32 termite species and their fungal inoculum from her mother’s colony ers in natural ecosystems, but they are also cultivars were sampled, covering the diver- to start her new garden (5, 15) (Fig. 1e). major agriculture, , and household sity of termites throughout their range in Attine ants grow their fungi in subterranean pests (1). Not surprising, much is known Africa and Asia. For the two-part historical chambers, manuring the gardens with dead about the extermination of these pest in- reconstruction mentioned above, Aanen et vegetable debris, or in the case of the leaf- sects, but only recently have genetic tech- al. generated DNA sequence information cutter ants, with leaf fragments cut from live

niques been applied to elucidate the evolu- for, first, the Termitomyces cultivars and . Attine ants are obligately dependent COMMENTARY tionary histories of these unique nonhuman undomesticated, fungal relatives; and sec- on their fungi; their brood is raised on an agricultural systems. ond, the termite farmers and nonfarmer exclusively fungal diet. Nests of most attine In a recent issue of PNAS, Aanen et al. (2) relatives. These two evolutionary recon- species number only a few dozen workers, present such an evolutionary analysis for structions reveal a remarkably complex fab- but nest sizes can reach millions in leafcutter fungus-growing termites and their culti- ric of fungicultural evolution in termites ants. Leafcutter ants are prodigious con- vated fungal crops, complementing similar (Fig. 1a). sumers of leaves and are among the most analyses recently completed for fungus- Fungus-growing behavior originated only damaging agricultural pests in South and growing ants (3–6) and fungus-growing bee- once in termites, involving a single ancestral Central America. tles (7–9) (Fig. 1 a–c). Aanen et al.’s meth- Termitomyces lineage that diversified into Many, but not all of the species in odology followed the same two-part analysis several defined cultivar groups, each asso- the subfamilies Scolytinae and Platypodinae anthropologists have taken to unravel the ciated almost exclusively with a similarly burrow extensive gallery systems into trees histories of human agricultural societies and defined group of termite farmers. Within as adults for feeding and oviposition. Some of these beetles, collectively called ambrosia their various crops. First, they assessed the each of these termite groups, however, cul- beetles (3,400 species), grow fungi on the patterns of relatedness between cultivated tivars are exchanged frequently between walls of their galleries for food (Fig. 1f). The crops and wild, undomesticated varieties; termite lineages. This is analogous to a largely clonal ambrosia fungi are only patterns revealed which crops were derived situation where distinct human lineages known from the beetles and their galleries, from common ancestral stocks and thus the would be specialized on particular crops, for suggesting an obligate association with the number of independent domestication example corn, and corn farmers could beetle farmers (9). The fungi serve as the events. Second, they determined the rela- switch between cultivation of any kind of beetle’s primary food source and are essen- corn variety but could not easily switch to tionships between independent farmer so- tial for the completion of the beetle life cycle cieties to infer common origins of agricul- cultivation of or potato. Termite (16). Like fungus-growing termites and ants, tural practices. Even in the absence of a farmers therefore appear to have evolved ambrosia beetles protect the fungal garden fossil (or archaeological) record, a juxtapo- adaptations to certain cultivar groups (e.g., from harmful contaminants and raise their sition of these two phylogenetic histories can specific fertilizing regimes), or cultivars brood on a fungal diet (16, 17). Fungus- reveal surprising details of agricultural evolved adaptations suitable only for certain growing beetles are major forestry pests, evolution. farmers (e.g., nutrients benefiting only cer- particularly those burrowing into live trees tain termites; ref. 12), or both. and infecting them with their fungi. Termite Fungiculture. About 330 of the The juxtaposition of termite and Termi- Ͼ2,600 known termite species are obligately tomyces evolution informed Aanen et al. dependent on the cultivation of a specialized about historical details that would have been See companion article on page 14887 in issue 23 of volume 99. fungus, Termitomyces, for food (10, 11). impossible to infer from the sparse fossil *To whom correspondence may be addressed. E-mail: Termitomyces is grown on termite feces in record of the fungus-growing termites (10). [email protected] or [email protected].

www.pnas.org͞cgi͞doi͞10.1073͞pnas.242594799 PNAS ͉ November 26, 2002 ͉ vol. 99 ͉ no. 24 ͉ 15247–15249 Downloaded by guest on September 30, 2021 Whereas ant and termite fungiculture originated only once in each group (Fig. 1 a and b), fungus-growing by ambrosia beetles has arisen at least seven times (Fig. 1c). Multiple origins of fungiculture in beetles is perhaps not surprising, given the sheer di- versity of beetle species and given the im- portance of feeding specializations in beetle diversification (19). Multiple origins, how- ever, do not preclude beetle-fungus coevo- lution within each independently derived system. Indeed, in each of the independently evolved farmer beetle lineages, entire groups of species are specialized on partic- ular groups of cultivars, paralleling not only the specializations already discussed above for termites, but also a series of special- izations known for different ant groups each associated with its own cultivar group (3, 5, 6). In general, switching of farmers to novel cultivars is possible, but limited almost exclusively to cultivars from within the same cultivar group as the farmers’ typical cultivar. Evolutionary reversal back to a non- fungus-farming lifestyle has apparently not occurred in any of the nine known, inde- pendently evolved farmer lineages (one ter- mite, one ant, and seven beetle lineages). This supports the view, formulated first for humans (ref. 20 and references therein), that the transition to agricultural existence is a drastic and possibly irreversible change that greatly constrains subsequent evolution. Fig. 1. Evolutionary histories of fungiculture in termites, ants, and beetles. (a–c) Comparison of the patterns of evolutionary diversification in the insect farmers (left cladograms) and their cultivated fungi Cultivar Exchange and Transmission Between (right cladograms). In the left cladograms, farmer lineages are shown in black; nonfarmer relatives are shown in gray. In the right cladograms, fungal cultivar lineages are shown in black; noncultivated feral Generations. Aanen et al.’s analyses indicate fungi are shown in gray. The number of independent origins of fungicultural behavior appears as that fungal cultivars are exchanged fre- independent farmer lineages in the left cladograms. The number of independently domesticated fungal quently between termite species. Fungal ex- lineages appears as separate cultivar lineages in the right cladograms. Cladograms synthesize pertinent change had been suspected for a long time information from published evolutionary reconstructions, but the number of lineages per group has been because most fungus-growing termites im- reduced because of space constraints. Similarly reduced is the number of nonfarmer lineages and port cultivar spores from external sources noncultivated fungi interpositioned between, respectively, the farmer lineages and the cultivated fungi. during nest initiation (13, 14). Acquisition of The history of successive diversification therefore is preserved in the cladograms, but evolutionary fungal starter material from external distances between lineages are reduced in many cases. Information is taken from ref. 2 for the termites, from refs. 3 and 5 for the ants, and from refs. 7–9 for the beetles. (d) Garden of the fungus-growing termite sources represents a primitive (ancestral) Macrotermes bellicosus (photo by K. Machielsen). The fungus is grown on fecal pellets that are stacked condition that arose at the origin of termite into walls of the fungus garden (comb). (e) Incipient garden of the fungus-growing ant Atta cephalotes fungiculture, and it is still practiced by the (photo by U.G.M.). The queen is shown resting on the incipient garden that is tended by her first cohort great majority of fungus-growing termite of workers. (f) Gallery of the ambrosia beetle Trypodendron lineatum (photo by S. Ku¨hnholz). The lineages. The exceptions are two derived ambrosia fungus appears as the black lining of the gallery. Developing beetle brood is shown in niches off lineages that arose late in termite evolution, the gallery. Fungal growth is monitored constantly by adult beetles (not shown) patrolling the galleries. the genus Microtermes, in which new queens carry asexual spores in their guts as starter inoculum for their new nests, and the single Fungicultural Origins. Termite, ant, and bee- ation may have been one in which the fungi species Macrotermes bellicosus, in which the tle farmers appear to have made the tran- used insect hosts for dispersal of spores new king is the sole carrier of spores (13, 14). (similar to flowering plants using bees as sition to fungiculture via different evolu- acquisition at the nest founding pollen vectors), and fungus-feeding and fun- tionary avenues. In the termites, fungi stage generally occurs during the rainy sea- probably were an important food source giculture arose secondarily out of such an ancestral vectoring system. This most likely son, a time when Termitomyces fungi pro- before true cultivation, and fungiculture duce fruiting structures () (11). arose when the termites secondarily devel- was the case in the beetles because many nonfungus feeding relatives of the ambrosia During fruiting, Termitomyces sends out my- oped an ability to manipulate fungal growth beetles are important vectors of fungal celial growths toward the nest surface and in their nests. Many nonfarming termite spores and because the ambrosia fungi are forms mushrooms for spore production. species are attracted to feed on fungus- derived from free-living fungi that depend These spores are produced sexually (meiot- infested wood (11, 12), and termite fungi- on arthropods for dispersal (18). It is unclear ically), and they are thus different from the culture therefore may represent an elabo- whether fungiculture in attine ants arose asexual spores produced in combs. Sexual ration of such ancestral feeding habits. In from ancestral mycophagy or from a system spores may be brought back into the nest by contrast, the ancestral insect-fungus associ- of fungal vectoring by ants (15). the termites for inoculation of new combs,

15248 ͉ www.pnas.org͞cgi͞doi͞10.1073͞pnas.242594799 Mueller and Gerardo Downloaded by guest on September 30, 2021 or they may disperse to recombine with Garden Parasites, Pesticides, and Antimicro- be that long-term evolutionary success as other fungi. It is these recombinants that are bial Defenses. All three insect farmers are farmers depends on effective strategies for thought to be picked up at the nest-founding proficient gardeners. They control cultivar agricultural disease management. For ex- stage by termite farmers (13, 14), thus ex- growth and remove weedy fungal competi- ample, the exceedingly successful termite plaining the observed high levels of lateral tors to maintain healthy gardens. If the and ant agriculturists rely on the propaga- cultivar transfer. As mentioned before, the farmers are removed, the gardens quickly tion of cultivar clones in huge monocultures, exceptions to this rule are the vertically deteriorate as fungal growth either congests and the ants are known to suppress crop inherited fungi cultivated by the genus the galleries (16) or the gardens are devas- diseases with antibiotics (24, 27, 28). Man- Microtermes and by Macrotermes bellicosus. tated by contaminants (11, 13, 23, 24). agement strategies thought to minimize the Interestingly, these termite lineages are also Farming insects also prevent mites and nem- evolution of specialized crop diseases and exceptional in that fungal fruiting structures atodes, common invaders of fungus- the emergence of antibiotic resistance in have never been observed on their nests (13, gardens, from feeding on the fungus and human agriculture, strategies such as inter- 14), suggesting that evolution of inheritance from contaminating gardens with alien cropping, sexually recombining crops, and of fungal clones from parent to offspring spores. Yet despite the farmer’s weeding low-pesticide farming, are possibly imple- nest led to the abandonment of sexual fruit- efforts, the stable conditions of gardens, ing as an integral part of the termite-fungus optimized for fungal growth, occasionally mented by some termites (sexual crops) and symbiosis. attract specialized fungal garden parasites. some beetles (sexual crops, intercropping), Ant and beetle fungiculture is most sim- This has been documented in detail for ant but clearly not by all insect farmers, partic- ilar to that of Microtermes and Macrotermes gardens, where the fungal parasite Escovop- ularly the ants. The ant farmers’ disease bellicosus. Each generation, reproductive sis can infect gardens to the detriment of resilience may lie in the combination of adults actively disperse their own fungus crop productivity and colony growth (23– several adaptations, for example the reli- when they found new colonies. In ambrosia 26). A similarly parasitic association has ance on a live source of pesticides (e.g., beetles, the beetles either ingest fungal been hypothesized for Xylaria fungi in ter- antibiotic-producing Streptomyces bacteria) spores before dispersal from their natal nest, mite combs (11, 13), because Xylaria over- that can coevolve with the pathogens; or or more commonly, gather spores in spe- grows the comb when the termites abandon more importantly, the intense monitoring of cialized storage structures (mycangia) for it (11, 13), resembling Escovopsis outbreaks gardens to eradicate immediately resistant fungal transport. Whereas predispersal ac- in ant gardens. Specialist fungal parasites of pathogen mutants before an epidemiologi- quisition of fungi is relatively unspecific, beetle fungiculture have yet to be identified. cal outbreak. Perhaps, apart from providing only certain fungal species appear to survive To combat Escovopsis infections of gar- us with a glimpse at possible evolutionary in the mycangia, selectively eliminating un- dens, attine ants use antimicrobial ‘‘pesti- outcomes of human agriculture millions of wanted fungi (17). As Aanen et al. argue, cides’’ that they derive from bacteria grown years from now, the insect farmers can teach similar selection must also occur at the nest on specialized regions of their own bodies us more imminently about epidemiological founding stage in the termites: either the (24, 27, 28). The bacteria belong to the principles and disease management of agri- first workers actively select a cultivar for the genus Streptomyces, a well-known genus of cultural pathogens. After all, they have been new combs (e.g., by filtering out appropriate soil bacteria that has been used by pharma-

farmers for millions of years, and maybe COMMENTARY spores from ingested forage in their guts), or ceutical industry for the discovery of novel they have figured out a trick or two that lead the specific growth conditions provided by antibiotics (e.g., streptomycin). Interest- to their remarkable agricultural success. the termites select indirectly by favoring ingly, Streptomyces bacteria and close rela- certain cultivars only (13). No such selection tives have also been isolated from termite We thank D. Aanen, S. Ku¨hnholz, and K. occurs at the nest-founding stage in the ants, guts (10, 11). It is unclear whether the gut Machielsen for the insect images in Fig. 1 and D. but queens may select between different Streptomyces function as sources of ‘‘pesti- Aanen, B. Farrell, and S. Ku¨hnholz for com- cultivars in their natal nest before their cides’’ in termite fungiculture, and whether ments. N.G. is funded by a Graduate Fellowship mating flight, provided that multiple culti- bacteria known to be associated with am- from the University of Texas at Austin. The var genotypes coexist within single ant col- brosia beetles (17) could serve similar anti- Mueller laboratory is funded by National Science onies. At present, however, there exists no microbial functions in beetle fungiculture. Foundation Faculty Early Career Development support for such within-nest cultivar diver- If there is an important message to be Program and Integrated Research Challenges in sity in attine gardens (21, 22). learned from the insect agriculturists, it may Environmental Biology grants.

1. Wilson, E. O. (1971) The Insect Societies (Belknap, (Kluwer Academic, Dordrecht, The ). 18. Malloch, D. & Blackwell, M. (1993) in Ceratocystis Cambridge, MA). 11. Batra, L. R. & Batra, S. W. T. (1979) in Insect- and Ophiostoma: Taxonomy, and Pathoge- 2. Aanen, D. K., Eggleton, P., Rouland-Lefe`vre, C., Fungus Symbiosis, ed. Batra, L. R. (Allanheld and nicity, eds. Wingfield, M. J., Seifert, K. A. & Webber, Guldberg-Frøslev, T., Rosendahl, S. & Boomsma, J. J. Osmun, Montclair, NJ), pp. 117–163. J. (Am. Phytopathol. Soc., St. Paul), pp. 195–206. (2002) Proc. Natl. Acad. Sci. USA 99, 14887–14892. 12. Rouland-Lefe`vre, C. (2000) in Termites: Evolu- 19. Farrell, B. D. (1998) Science 281, 555–559. 3. Chapela, I. H., Rehner, S. A., Schultz, T. R. & tion, Sociality, Symbioses, Ecology, eds. Abe, T., 20. Diamond, J. (1997) Guns, Germs, and Steel: The Mueller, U. G. (1994) Science 266, 1691–1694. Bignell, D. E. & Higashi, M. (Kluwer Academic, Fates of Human Societies (Norton, New York). 4. Schultz, T. R. & Meier, R. (1995) Syst. Entomol. Dordrecht, The Netherlands), pp. 289–306. 21. Bot, A., Rehner, S. A. & Boomsma, J. J. (2001) 20, 337–370. 13. Wood, T. G. & Thomas, R. J. (1989) in Insect- Evolution (Lawrence, Kans.) 55, 1980–1991. Fungus Interactions, eds. Wilding, N., Collins, 5. Mueller, U. G., Rehner, S. A. & Schultz, T. R. 22. Mueller, U. G. (2002) Am. Nat. 160 (Suppl.), N. M., Hammond, P. M. & Webber, J. (Academic, (1998) Science 281, 2034–2038. S67–S98. New York), pp. 69–92. 6. Green, A. M., Mueller, U. G. & Adams, R. M. M. 23. Currie, C. R., Mueller, U. G. & Malloch, D. (1999) 14. Darlington, J. E. C. P. (1994) in Nourishment and (2002) Mol. Ecol. 11, 191–195. Proc. Natl. Acad. Sci. USA 96, 7998–8002. Evolution in Insect Societies, eds. Hunt, J. H. 88, 55, 7. Cassar, S. & Blackwell, M. (1996) Mycologia & Nalepa, C. A. (Westview, Boulder, CO), 24. Currie, C. R. (2001) Annu. Rev. Microbiol. 596–601. pp. 105–130. 357–380. 8. Jones, K. G. & Blackwell, M. (1998) Mycol. Res. 15. Mueller, U. G., Schultz, T. R., Currie, C. R., 25. Currie, C. R. (2001) Oecologia 128, 99–106. 102, 661–665. Adams, R. M. M. & Malloch, D. (2001) Q. Rev. 26. Currie, C. R., Bot, A. & Boomsma, J. J. (2003) 9. Farrell, B. D., Sequeira, A. S., O’Meara, B. C., Biol. 76, 169–197. Oikos, in press. Normark, B. B., Chung, J. H. & Jordal, B. H. 16. Batra, L. R. (1966) Science 153, 193–195. 27. Currie, C. R., Scott, J. A., Summerbell, R. & (2001) Evolution (Lawrence, Kans.) 55, 2011–2027. 17. Beaver, R. A. (1989) in Insect–Fungus Interactions, Malloch, D. (1999) Nature 398, 701–704. 10. Abe, T., Bignell, D. E. & Higashi, M. (2000) eds. Wilding, N., Collins, N. M., Hammond, P. M. & 28. Poulsen, M., Bot, A., Currie, C. R. & Boomsma, Termites: Evolution, Sociality, Symbioses, Ecology Webber, J. F. (Academic, New York), pp. 121–143. J. J. (2002) Insectes Sociaux 49, 15–19.

Mueller and Gerardo PNAS ͉ November 26, 2002 ͉ vol. 99 ͉ no. 24 ͉ 15249 Downloaded by guest on September 30, 2021