&Wo&y, 89(5), 2008, pp. 1216-1222 ? 2008 by the Ecological Society of America BLACK YEAST SYMBIONTS COMPROMISE THE EFFICIENCY OF ANTIBIOTIC DEFENSES IN FUNGUS-GROWING ANTS AINSLIE E. F. LITTLE1 AND CAMERON R. CURRIE Department of Bacteriology, University of Wisconsin, Madison, Wisconsin 53706 USA, and Smithsonian Tropical Research Institute, Apartado Box 2072, Balboa, Ancon, Republic of Panama Abstract. Multiplayer symbioses are common in nature, but our understanding of the ecological dynamics occurring in complex symbioses is limited. The tripartite mutualism between fungus-growing ants, their fungal cultivars, and antibiotic-producing bacteria exemplifies symbiotic complexity. Here we reveal how black yeasts, newly described symbionts of the ant-microbe system, compromise the efficiency of bacteria-derived antibiotic defense in fungus-growing ants. We found that symbiotic black yeasts acquire nutrients from the ants' bacterial mutualist, and suppress bacterial growth. Experimental manipulation of ant colonies and their symbionts shows that ants infected with black yeasts are significantly less effective at defending their fungus garden from Escovopsis, a prevalent and specialized pathogen. The reduction of mutualistic bacterial biomass on ants, likely caused by black yeast symbionts, apparently reduces the quantity of antibiotics available to inhibit the garden pathogen. Success of the ant-fungal mutualism is directly dependent on fungus garden health. Thus our finding that black yeasts compromise the ants' ability to deal with the garden parasite indicates that it is an integral component of the symbiosis. This is further evidence that a full understanding of symbiotic associations requires examining the direct and indirect interactions of symbionts in their ecological community context. Key words: Apterostigma; Attini; bacteria; ecological community dynamics; Escovopsis; interspecific interaction; symbiosis; yeast. INTRODUCTION Symbiotic associations shape the biology of all organisms and have been fundamental in many of the major evolutionary transitions toward greater organis- mal complexity and diversity (Ehrlich and Raven 1964, Pirozynski and Malloch 1975, Margulis and Fester 1991, Thompson 1994, Van der Heijden et al. 1998, Lutzoni et al. 2001). For example, microbial symbionts mediated the origin of the eukaryotic cell, facilitated the coloni- zation of land by plants, and aid digestion in animals. Traditionally, many symbioses have been studied in a binary framework, which allowed for more practical empirical studies and more tractable theoretical models. However, it is now evident that many host-microbe pairs are part of a more complex network of beneficial and antagonistic symbionts (Bronstein and Barbosa 2002, Moran et al. 2005). Unlike the direct effects pathogens have on their hosts, or mutualists have on one another, the majority of interactions that occur among commu- nity members are indirect, i.e., those that require an intermediate species to be transmitted (Wootton 1994). Indirect interactions can influence both the outcome and the cost-benefit ratio of symbiotic interactions (Letour- neau 1990, Currie et al. 1999a, b, Gastreich 1999). Manuscript received 16 May 2007; revised 31 October 2007; accepted 18 December 2007; final version received 15 January 2008. Corresponding Editor: D. H. Feener, Jr. 1 E-mail: alittle@wisc.edu Furthermore, selection on traits that are important to maintaining successful symbioses have the potential to be influenced indirectly by diffuse symbionts as do the coevolutionary dynamics of codependent organisms. Ants in the tribe Attini have an ancient obligate association with fungi that they cultivate for food. The nutritional benefit provided to ants by their mutualistic fungus is a prime target for exploitation by other organisms. As such, fungus-growing ants have evolved various mechanisms to defend their fungus garden, including hygienic behaviors to rid the garden of foreign microbes, and management of hazardous garden waste (Weber 1972, Currie et al. 19996, Bot et al. 2001, Currie and Stuart 2001, Hart and Ratnieks 2002, Fernandez- Marin et al. 2006). Despite these defensive mechanisms, the ant-fungal mutualism is parasitized by a lineage of specialized pathogens in the genus Escovopsis (Currie et al. 1999a, Currie 2001, Reynolds and Currie 2004). To help protect their fungal crops from the specialized pathogen, the ants engage in a second mutualism with bacteria in the genus Pseudonocardia (actinomycete), which produce antibiotics that specifically inhibit Escovopsis (Currie et al. 19996). The mutualistic bacteria typically occur on, or in, specialized cuticular structures that are connected to unique exocrine glands, which apparently produce nutrients to support bacterial growth (Currie et al. 2006). Thus, each member of the fungus-growing ant community is directly and indirectly 1216 May 2008 SYMBIONTS COMPROMISE ANTIBIOTIC DEFENSE 1217 influenced by one another in both positive and negative ways. While isolating the bacterial mutualists from species of Apterostigma, a phylogenetically basal genus of fungus-growing ants, we frequently obtained colonies of black yeasts (closely related to fungi in the genus Phialophora; Ascomycota). This monophyletic group of black yeasts are prevalent within and among ant colonies, are symbiotically associated with fungus- growers from across the phylogenetic and geographic distributions of the mutualism, and occupy a specific niche on the bodies of fungus-growing ants (Little and Currie 2007). Here we link the multiple interactions occurring in the ecological community of the fungus- growing ant-microbe symbiosis by exploring and integrating the impacts of direct and indirect interac- tions among symbionts. Specifically we test the hypoth- esis that black yeasts exploit the ant-bacterial mutualism by consuming the ants' mutualistic bacteria. Black yeast interactions with Pseudonocardia are inves- tigated through bioassays. To investigate the impact black yeasts have on ant-microbe mutualists experienc- ing stress caused by garden infection, we experimentally manipulate ant colonies. In addition, to further under- stand the results of this experiment we perform a series of bioassay challenges between microbial symbionts in vitro. We discuss mechanistic details of the complex ecological role the black yeast symbiont plays in the ant- microbe symbiosis, and suggest that the multiple indirect interactions of fungus-growing ant symbionts have helped structure the ecological community over evolu- tionary time. MATERIALS AND METHODS Study organisms Apterostigma spp. are common throughout much of Central and South America. Ants house their fungus gardens under rocks or decaying logs, between the stilts of palms, or in moist soil on the vertical planes of ravine banks (Weber 1972). They manure their fungi with decaying vegetation and insect frass. The colonies of Apterostigma pilosum used in this study were collected between 2002 and 2006 in the Republic of Panama (Canal Zone, Darien Province, Bocas del Toro Prov- ince), and were maintained in the laboratory in dual plastic chambers placed on islands in mineral oil, to prevent horizontal transfer of microbes among colonies. Colonies were watered and fed a mixture of oats, corn meal, tea leaves, and oak catkins once a week. Voucher samples of ants from each colony were collected and stored at ?80?C in ethanol, and deposited at the Smithsonian Institute. Isolation of microbes Black yeasts and actinomycetes were isolated from A. pilosum following a protocol used to isolate actinomy- cetes from fungus-growing ants (Cafaro and Currie 2005). Ants were scraped using a sterile metal probe and the material was spread on chitin agar. Plates were incubated at room temperature for four weeks. Individ- ual colony forming units (CPUs) of black yeast were then transferred to, and maintained on potato dextrose agar (PDA) (Difco, Sparks, Maryland, USA). Individ- ual CPUs of Pseudonocarida, the ants' bacterial mutualist, were transferred to yeast malt extract agar (YMEA). Black yeast growth on water and Pseudonocardia agar To determine whether the black yeast is capable of using the ant's bacterial mutualist as a nutrient source, 0.2-mm3 plugs of the fungus were placed on two types of media: (1) Pseudonocardia agar, which is water agar containing 2.0 g/L of Apterostigma-a.ssocia.ted Pseudo- nocardia tissue from pure culture, and (2) water agar (15 g agar/L H20). Cultures were incubated at room temperature for four weeks, and surface area growth was measured (mean of three diameter measurements) and compared using a two-sample t test. Infection experiment To determine whether exploitation of the ant- bacterial mutualism by black yeast disrupts the ants' ability to defend their fungus garden, we conducted experiments crossing the presence/absence of black yeasts with the presence/absence of Escovopsis using subcolonies of A. pilosum. Subcolonies of A. pilosum, composed of 50 mg of fungus garden and six worker ants, were set up in 60-mm plastic Petri plates with moist cotton along the outer edge. Each subcolony was fed a mixture of oak catkins and tea leaves daily, and subcolonies were deemed healthy once ants started incorporating substrate into the fungus garden. Each subcolony was randomly assigned to one of four treatments: (1) black yeast removed and no infection, (2) black yeast removed and infection with Escovopsis, (3) black yeast not removed and no infection, and (4) black yeast not removed and infection with Escovopsis. To remove black yeasts from ants, workers were removed from their subcolony, placed in a 4-mL glass vial, and exposed to 1.0% Amphotericin B (AmpB) three times during a 24-hour period (approximately every eight hours). AmpB is a polyene antifungal that preferentially binds to ergosterols, thus disrupting osmotic integrity of the fungal cell membrane (Terrell and Hughes 1992). This method inhibited black yeasts to the point where they were undetectable via culturing methods for at least one month, but did not measurably affect actinomycete growth on ants (n = 25 replicates). Ants were given 30 minutes to groom themselves after exposure to antibiotics, and then returned to their subcolonies. To determine if AmpB adversely affects the other symbionts in the fungus-growing ant microbe symbiosis, fungal cultivar, Escovopsis, and Pseudono- cardia isolated from Apterostigma were all incubated on PDA (fungal) and YMEA (bacterial) containing 1.0% AmpB overlays. Unlike the black yeast symbionts, 1218 AINSLIE E. F. LITTLE AND CAMERON R. CURRIE Ecology, Vol. 89, No. 5 1.0 Sg ?-8 ?; o.6 o ? c ?- o CJ ?=, en o c o_ c o .2 01-? 0.4 0.2 ? With black yeast Without black yeast No Escovopsis Escovopsis Treatment FIG. 1. Two-way interaction graph illustrating the effects of black yeast and Escovopsis infections in Apterostigma pilosum subcolonies. Ants infected with black yeasts (solid line) are significantly worse at defending their garden from Escovopsis infection than are uninfected ants (dashed line). Nonparallel lines indicate that the interaction of the two treatments is significant. The standard error for each data point is less than 0.01; thus bars are not shown. which are unable to grow on YMEA with a 1% AmpB overlay, the growth rate of fungal cultivar, Psuedono- cardia, and Escovopsis symbionts were not affected by the presence of the antibiotic (t test, P > 0.1, df = 13). Prior to being used in the experiment, ant colonies were tested to confirm an absence of detectable Escovopsis infection (Currie 2001). The strain of Escovopsis (brown morphotype; see Gerardo et al. [2006]) used for infection was isolated in Gamboa, Panama from an A. pilosum colony. The isolate was grown on PDA with 1000 iu/mL of penicillin-strepto- mycin (MP Biomedicals, Aurora, Ohio, USA). Spores were added to ddH20 (dd = double-distilled) with Tween 20 (5 X 10~5 mL/L) (Fisher Scientific, Pittsburgh, Pennsylvania, USA) to evenly disperse spores in solution. Each subcolony received 50 u.L of solution (~6000 spores per subcolony) via mist inoculation. Subcolonies that did not receive the Escovopsis infection treatment were sprayed with an equal amount of ddH20/Tween 20 solution. Changes in fungus garden biomass were measured as a fitness indicator. Garden biomass was measured 72 hours after treatment, and proportions of garden biomass (relative to original amount) were subjected to two-way ANOVA in Mini tab (Minitab Version 14 for Windows, Minitab, Inc., State College, Pennsylvania, USA). Graphical results (Fig. 1) show the two-way interaction between the AmpB and Escovopsis "infec- tion" treatments. Bioassay challenges To determine the cause of black yeast associated garden loss found in infection experiments we performed a series of co-culture bioassay challenges in which we incubated different pairs of symbionts side by side on nutrient medium in Petri dishes (Fig. 2a), and then compared their growth in the combined cultures with their growth in pure culture (Fig. 2b). Bioassay challenges between the black yeast and Pseudonocardia were conducted in Petri plate dishes containing YMEA, which allows for good growth of both symbionts, while pairings of black yeast with the fungal cultivar and Escovopsis were completed on PDA, which supports fungal growth. Black yeast hyphal material (2 mm2) was inoculated onto each plate 1.5 cm away from a point inoculation of a Pseudonocardia, Escovopsis, or fungal cultivar strain. The growth of each CFU (colony- forming unit) was measured 10, 20, and 30 days after inoculation by taking three diameter measurements of the CFUs (which grow in a circular pattern), and calculating the surface area of the CFU from the mean diameter. Surface areas of each CFU were then subjected to two-sample f tests in Minitab to identify significant changes in growth. Solid medium was used for these challenges because in liquid, the yeast and actinomycetes get intertwined, making it difficult to obtain an accurate measurement of each strain's biomass. To determine whether black yeasts suppress antibiotic production by Pseudonocardia, separate bioassay chal- lenges between the black yeast and Pseudonocardia were set up, as above, and after 10 days, Escovopsis was point-inoculated at the edge of the Petri plate. Subse- quent growth of both Pseudonocardia and Escovopsis were measured five days later, and compared to growth on control plates (i.e., no black yeast) to identify differences in the bacterial-parasite interaction. Specif- ically we quantified the difference in Pseudonocardia'^ ability to inhibit Escovopsis by measuring the zone of inhibition between the two microbes on control and black yeast infected plates. RESULTS Black yeast growth on water and Pseudonocardia agar Black yeast symbiont growth on water agar was minimal (mean surface area growth = 316.16 ? 120.9 mm2 [mean ? SD]). However, incubation on water agar containing biomass of Pseudonocardia revealed signifi- cantly greater growth of black yeast, (two-sample t test, df=65, t=?3.54, P = 0.0007; mean surface area growth 473.23 ? 224.46 mm2). This result indicates that the black yeast is capable of utilizing Pseudonocardia- derived nutrients. Infection experiment No difference in ant survivorship in the subcolonies of A. pilosum was detected in our infection experiment (t test, df=9, ( = 0.91, P = 0.3865), nor were we able to detect any changes in fungal garden biomass when black yeasts were present or absent from ants tending the garden (; test, df=9, (=1.34, f = 0.2131). Additionally, ants do not exhibit any observable behavioral aversion to black yeast tissue placed in their colonies (A. Little, May 2008 SYMBIONTS COMPROMISE ANTIBIOTIC DEFENSE 1219 70' d 60- % 50- S 40- % 30- S O 20- T 10- Bacteria : BY Bacteria : control Bacteria treatment BY : bacteria BY : control Black yeast treatment FIG. 2. Bioassay challenges used to investigate black yeast-Pseudonocardia interactions. Photos of (a) bioassay challenges between black yeasts and Pseudonocardia (arrow points to white Pseudonocardia growth), and (b) growth of Pseudonocardia in solitary culture, (c and d) Results from bioassay challenges indicating that Phialophora (black yeast) growth increases at the expense of Pseudonocardia (ants' bacterial mutualist) growth when allowed to grow side by side on yeast malt extract medium for 20 days at room temperature, (c) Bar chart showing black yeast (BY) growth in the presence of bacterial mutualists (left), and growth alone on nutrient medium (right), (d) Bar chart showing Pseudonocardia (bacterial mutualist) growth in the presence of black yeast isolated from Apterostigma (left), and growth alone on nutrient medium. Error bars represent + SD. personal observation). As in previous experiments Escovopsis infection has a significant negative impact on fungus garden biomass (t test, df=9, ( = ?36.89, P < 0.0001). Interestingly, through experimental manipula- tion of A. pilosum subcolonies we found that signifi- cantly more fungal garden biomass was lost in colonies treated with Escovopsis and tended by workers infected with black yeasts than in other treatments (ANOVA, Fli43 = 163.31, P < 0.001; Fig. 1). More specifically, when fungus-growing ant colonies are infected with the garden parasite Escovopsis, the presence of black yeasts reduces the ants' ability to suppress infection of their fungus garden. Bioassay challenges Pseudonocardia inhibition of Escovopsis in vitro was not altered by the presence of black yeast (t test, df =45, P = 0.6265, t = ?0.49). Results from paired bioassay challenges indicated that black yeasts do not significant- ly promote growth of Escovopsis (t test, df = 45, P = 0.5648, t = 0.58), nor does Escovopsis significantly alter black yeast growth (t test, df =45, t = -0.455, P = 0.6513). In addition, co-culture experiments of black yeast and the ants' cultivated fungi revealed no significant decrease in the growth of fungal cultivars or significant increase in growth of the black yeast (fungal cultivars ; test, df=45, f = 0.4814, ( = 0.71; black yeast f test, df = 45, P = 0.3472, ( = 0.95). Bioassay challenges between black yeast and Pseudonocardia (Fig. 2a, b) revealed that black yeast growth significantly increases in the presence of Pseudonocardia (t test, P < 0.0001, ; = ?4.58, df = 54; Fig. 2c). In contrast, the presence of black yeast resulted in a significant reduction in Pseudonocardia growth (t test, P < 0.001, t = 7.18, df = 13; Fig. 2d). DISCUSSION Our study exploring complex species interactions within the fungus-growing ant-microbe symbiosis re- veals that the recently discovered black yeast symbionts exploit the ant-bacterial mutualisms. Infection experi- ments and behavioral observations suggest that black yeasts do not directly impact the health of either fungus- growing ants, or their cultivated fungal mutualists. However, the black yeast is able to derive nutrients directly from Pseudonocardia, the ants' antibiotic- 1220 AINSLIE E. F. LITTLE AND CAMERON R. CURRIE Ecology, Vol. 89, No. 5 Black yeast Parasite FIG. 3. Web diagram of the direct and indirect interactions of the attine ant-microbe symbiosis. The inclusion of indirect interactions illustrates how black yeast symbionts compromise the efficiency of antibiotic defenses of fungus-growing ants. Solid lines represent direct effects, which are the result of a physiological interaction between two species. Dashed lines represent indirect effects, which require the presence of an intermediary (species) to arise. The cost (?) and benefit (+) to each organism is indicated at the tip of the arrowhead. Bold, thick black lines indicate the direct (solid line) and indirect (dashed lines) effects black yeasts have on fungus-growing ant symbionts. Black yeasts directly inhibit growth of the ants' bacterial mutualist, which results in less efficient suppression of the garden parasite (indicated by a red X). This indirectly decreases the benefit bacteria provide ants and cultivars (also indicated by red X's). Consequently, black yeasts indirectly benefit the garden parasite by compromising antibiotic defenses, which increases the cost the parasite inflicts on ants and cultivar (indicated by green arrows). producing mutualist, which increases the growth rate of the black yeast while slowing the growth of Pseudono- cardia. The direct impact black yeast has on Pseudono- cardia suggests it has the potential to indirectly impact the health of ants by disrupting their ability to inhibit the parasite Escovopsis via their antibiotic-producing bacteria. Indeed, experimental manipulation of ant colonies, crossing the presence/absence of black yeast with presence/absence of the garden parasite Escovopsis, revealed that subcolonies infected by black yeasts were significantly more impacted by Escovopsis infection (Fig. 1). Thus, our results indicate that both direct and indirect effects link multiple interactions among the fungus-growing ant-microbe symbionts and have likely been a significant factor structuring the ecological community over evolutionary time. Our finding that the presence of black yeast symbionts on ants increases the virulence of Escovopsis could be the result of several mechanisms. Ants would be less adept at defending their fungus garden from Escovopsis if the black yeasts inhibit antibiotic production by the bacterial mutualists. However, we were unable to detect any significant difference in the in vitro ability of Pseudonocardia to inhibit Escovopsis in the presence or absence of the black yeast, indicating that ant-associated black yeasts do not inhibit antibiotic production by the Pseudonocardia. Another mechanism that might increase the virulence of Escovopsis is that black yeasts may directly stimulate its growth. Similarly, the decreases in fungus garden biomass observed during the fitness experiment could result if black yeasts inhibit or kill fungal cultivars. However, we did not find evidence that May 2008 SYMBIONTS COMPROMISE ANTIBIOTIC DEFENSE 1221 black yeasts promote Escovopsis growth, or suppression or killing of fungal cultivars. Alternatively, if black yeasts are capable of significantly inhibiting the growth of the Pseudonocardia population on ants, the ants may be less able to defend their gardens from Escovopsis. Results from bioassay challenges showing that black yeast growth increases in the presence of the bacterial mutualist at a cost to Pseudonocardia growth support such a hypothesis. In addition to directly parasitizing Pseudonocardia, it is still possible that black yeasts also directly exploit the ant-bacterial mutualism by outcom- peting the bacteria for nutrients produced by the ants, as both promotion of growth in the presence of the bacteria and the ability to inhibit bacterial growth, would increase the competitive ability of the black yeast within the ant-bacterial niche. Understanding how interspecific interactions within communities drive changes in species abundance, distribution, and composition has been a common theme in ecology for over a century. However, studies clearly demonstrate that extracting pair-wise interac- tions from their community context to investigate ecological changes can be unrealistic and misleading (Sih et al. 1985, Bronstein and Barbosa 2002, Stanton 2003, this study). Most interactions among species are relatively weak and diffuse, but weak interactions can have strong effects when measured in the context of an ecological community (Paine 1992, Berlow 1999). This may be the case for the pair-wise interaction between black yeast and actinomycete symbionts. Although black yeasts significantly decrease the growth of actinomycetes in vitro, this interaction may be insignif- icant in vivo with respect to bacterial fitness since the association is so prevalent in nature. However, when this pair-wise interaction is examined within its community context the interaction becomes significant. The indirect effects black yeasts have on the fungus garden affect the fitness of the ant-cultivar-bacteria tripartite mutualism. Interestingly, the black yeast acts synergistically to benefit the fungal garden parasite by allowing it to infect the garden more severely. Synergism is not uncommon in pathogenesis. Herbivores increase the impact of many plant pathogens (Caesar 2003), UV-B increases the severity of several amphibian fungal pathogens (Kiesecker and Blaustein 1995) and viral- bacterial synergism causes excessive mortality during infiuenzae outbreaks (Jones et al. 1983). Ants infected with black yeasts likely have fewer bacteria available to produce antibiotics that inhibit Escovopsis spores from infecting their fungal cultivar, thus the black yeast indirectly synergize Escovopsis infection. An interesting point to consider is whether the black yeast and Escovopsis are actually diffuse mutualists. Currie et al. (2003) showed that actinomycetes become more preva- lent on the cuticle of ants during Escovopsis infection. The increased growth of Pseudonocardia stimulated by Escovopsis, could in turn benefit black yeasts, which derive nutrients from the bacteria. Studies monitoring the abundance of Psuedonocardia and black yeasts on the ants' cuticle during different levels of infection would shed light on this hypothesis. Understanding the antagonistic and beneficial inter- actions occurring in the ant-microbe symbiosis clearly requires an understanding of the contributions made by black yeast symbionts. Between 33% and 75% of fungus- growing ant colonies have detectable Escovopsis infec- tions, which are lethal to fungus gardens if not adequately controlled by the ants (Currie 2001). Thus black yeasts, which decrease mutualistic bacterial biomass and indirectly increase the virulence of Esco- vopsis, have the potential to substantially impact the fitness of fungus-growing ants in nature. Assessing community dynamics by examining indirect interactions may be particularly informative in communities that are strongly shaped by pathogens and/or parasites. Indirect effects that are weak or diffuse when community members are healthy may become strong effectors when community members are stressed by antagonists such as parasites or pathogens. The finding that a newly discovered symbiont has such strong ecological impacts on a well-studied symbiosis, in combination with the recent recognition that many other symbiotic commu- nities involve greater species complexity than previously recognized, highlights the need for future research to explore mutualistic and host-pathogen associations in a community context. ACKNOWLEDGMENTS This work was supported by NSF (IRCEB DEB-0110073 to C. R. Currie, DDIG DEB-0608078 to C. R. Currie and A. E. F. Little). We are grateful to STRI and ANAM of the Republic of Panama for facilitating the research and granting collect- ing permits. For valuable logistical support, we thank M. Bergsbaken, M. Cafaro, P. Foley, M. Leone, and O. Arosemana. We are thankful to K. Boomsma, E. Caldera, N. Gerardo, S. de Hoog, M. McFall-Ngai, A. Pinto, M. Poulsen, and J. Scott for invaluable comments on this study and paper. LITERATURE CITED Berlow, E. L. 1999. Strong effects of weak interactions in ecological communities. Nature 398:330-334. Bot, A. N. M., C. R. Currie, A. G. Hart, and J. Boomsma. 2001. Waste management in leaf-cutting ants. Ethology Ecology and Evolution 13:225-237. Bronstein, J., and P. Barbosa. 2002. Multi-trophic/multi-species interactions: the role of non-mutualists in shaping and mediating mutualisms. Pages 44-66 in T. Tscharntke and B. A. Hawkins, editors. Multitrophic interactions. Cam- bridge University Press, Cambridge, UK. Caesar, A. J. 2003. Synergistic interaction of soilborne plant pathogens and root-attacking insects in classical biological control of an exotic rangeland weed. Biological Control 28: 387-387. Cafaro, M. J., and C. R. Currie. 2005. Phylogenetic analysis of mutualistic filamentous bacteria associated with fungus- growing ants. Canadian Journal of Microbiology 51:441- 446. Currie, C. R. 2001. Prevalence and impact of a virulent parasite on a tripartite mutualism. Oecologia 128:99-106. Currie, C. R., A. N. M. Bot, and J. J. Boomsma. 2003. Experimental evidence of a tripartite mutualism: bacteria 1222 AINSLIE E. F. LITTLE AND CAMERON R. CURRIE Ecology, Vol. 89, No. 5 protect ant fungus gardens from specialized parasites. Oikos 101:91-102. Currie, C. R., U. G. Mueller, and D. Malloch. 1999a. The agricultural pathology of ant fungus gardens. Proceedings of the National Academy of Sciences (USA) 96:7998-8002. Currie, C. R., M. Poulsen, J. Mendenhall, J. J. Boomsma, and J. Billen. 2006. Coevolved crypts and exocrine glands support mutualistic bacteria in fungus-growing ants. Science 311: 81-83. Currie, C. R., J. A. Scott, R. C. Summerbell, and D. Malloch. 19996. Fungus-growing ants use antibiotic-producing bacte- ria to control garden parasites. Nature 398:701-704. Currie, C. R., and A. E. Stuart. 2001. Weeding and grooming of pathogens in agriculture by ants. Proceedings of the Royal Society of London B 268:1033-1039. Ehrlich, P. R., and P. H. Raven. 1964. Butterflies and plants: a study in coevolution. Evolution 18:586-608. Fernandez-Marin, H., J. Zimmerman, S. A. Rehner, and W. A. Wcislo. 2006. Active use of the metapleural glands by ants in controlling fungal infection. Proceedings of the Royal Society of London B 273:1689-1695. Gastreich, K. R. 1999. Trait-mediated indirect effects of a theridiid spider on an ant-plant mutualism. Ecology 80: 1066-1070. Gerardo, N. M., S. R. Jacobs, C. R. Currie, and U. G. Mueller. 2006. Ancient host-pathogen associations maintained by specificity of chemotaxis and antibiosis. PLoS Biology 4: 1358-1363. Hart, A. G, and F. L. W. Ratnieks. 2002. Waste management in the leaf-cutting ant Atta colombica. Behavioral Ecology 13: 224-231. Jones, W. T., J. H. Menna, and D. E. Wennerstrom. 1983. Lethal synergism induced in mice by influenza type-a virus and type-la group-B Streptococci. Infection and Immunity 41:618-623. Kiesecker, J. M., and A. R. Blaustein. 1995. Synergism between Uv-B radiation and a pathogen magnifies amphibian embryo mortality in nature. Proceedings of the National Academy of Sciences (USA) 92:11049-11052. Letourneau, D. K. 1990. Code of ant-plant mutualism broken by parasite. Science 248:215-217. Little, A. E. F., and C. R. Currie. 2007. Symbiotic complexity: discovery of a fifth symbiont in the attine ant-microbe symbiosis. Biology Letters 3:501-504. Lutzoni, F., M. Pagel, and V. Reeb. 2001. Major fungal lineages are derived from lichen symbiotic ancestors. Nature 411:937-940. Margulis, L., and R. Fester. 1991. Symbiosis as a source of evolutionary innovation. MIT Press, Cambridge, Massachu- setts, USA. Moran, N. A., P. H. Degnan, S. R. Santos, H. E. Dunbar, and H. Ochman. 2005. The players in a mutualistic symbiosis: insects, bacteria, viruses, and virulence genes. Proceedings of the National Academy of Sciences (USA) 102:16919-16926. Paine, R. T. 1992. Food-web analysis through field measure- ment of per-capita interaction strength. Nature 355:73-75. Pirozynski, K. A., and D. W. Malloch. 1975. The origin of land plants: a matter of mycotrophism. BioSystems 6:153-164. Reynolds, H. T., and C. R. Currie. 2004. Pathogenicity of Escovopsis weberi: the parasite of the attine ant-microbe symbiosis directly consumes the ant-cultivated fungus. Mycologia 96:955-959. Sih, A., P. Crowley, M. McPeek, J. Petranka, and K. Stroh- meier. 1985. Predation, competition, and prey communities? a review of field experiments. Annual Review of Ecology and Systematic^ 16:269-311. Stanton, M. L. 2003. Interacting guilds: moving beyond the pairwise perspective on mutualisms. American Naturalist 162:810-23. Terrell, C. L., and C. E. Hughes. 1992. Antifungal agents used for deep-seated mycotic infections. Mayo Clinic Proceedings 67:69-91. Thompson, J. N. 1994. The coevolutionary process. The University of Chicago Press, Chicago, Illinois, USA. Van der Heijden, M. G A., J. N. Klironomos, M. Ursic, P. Moutoglis, R. Streitwolf-Engel, T. Boiler, A. Wiemken, and I. R. Sanders. 1998. Mycorrhizal fungal diversity determines plant biodiversity, ecosystem variability and productivity. Nature 396:69-72. Weber, N. A. 1972. Gardening ants: the attines. American Philosophical Society, Philadelphia, Pennsylvania, USA. Wootton, J. T. 1994. The nature and consequences of indirect effects in ecological communities. Annual Review of Ecology and Systematics 25:443^466.