Behavioral Ecology doi:10.1093/beheco/ars109 Advance Access publication 17 September 2012
Original Article
Fungus-gardening ants prefer native fungal species: do ants control their crops? Jon N. Seal, Jeffrey Gus, and Ulrich G. Mueller Section of Integrative Biology, University of Texas at Austin, Austin, TX 78712, USA
Introduction
M
utualistic symbioses are among the most ecologically and evolutionarily important phenomena (Janzen 1985; Bronstein 2001; Douglas 2010). Mutualisms are generally explained as interactions in which interacting partners are under selection to maximize reciprocal benefits while minimizing costs (Law and Lewis 1983; Bronstein 1994, 2001; Sachs et al. 2004). In imbalanced mutualism, a controlling partner (the host) has some control over host–symbiont and symbiont–symbiont conflict (Douglas 2010; Sachs et al. 2011a, 2011b). Control and stability are thought to be conferred by 2 types of mechanisms. The first is stabilization by partner choice in which the host may use either direct information on partner properties/performance or conduct “screens” when direct information is lacking about the symbionts with which it may interact (Bull and Rice 1991; Sachs et al. 2004; Archetti et al. 2011a, 2011b). Mutualisms that depend on partner choice frequently exhibit horizontal transmission of symbionts. In contrast, mutualisms that rely on partner-fidelity feedback generally depend on vertical transmission. These mutualisms tend to have partners that cooperate and perhaps even co-propagate over significant evolutionary time such that neither partner has an incentive to sabotage the other because both benefit from successful reproduction (Boucher 1985; Bull and Rice 1991; Herre et al. 1999; Sachs et al. 2004; Weyl et al. 2010). Some of the key traits exhibited by fungus-gardening (attine) ants are the monocultures of their gardens (Mueller et al. 2010), vertical transmission of their farmed fungal
Address correspondence to J. N. Seal. E-mail: trachymyrmex@ gmail.com. Received 22 September 2011; revised 18 May 2012; accepted 25 May 2012. © The Author 2012. Published by Oxford University Press on behalf of the International Society for Behavioral Ecology. All rights reserved. For permissions, please e-mail:
[email protected]
symbionts (Mueller and Gerardo 2002; Mueller et al. 2005), and profound physiological integration among the ants and the fungi (Martin 1987; de Fine Licht et al. 2010). Although the ants and the fungi represent an exosymbiosis, the integration between the partners is so extensive that it may be best described as a highly integrated superorganism (Seal and Tschinkel 2007a; Hölldobler and Wilson 2011), much in the way eukaryotic cells are comprised of the descendents of prokaryotic cells (Szathmáry and Maynard Smith 1995; Margulis and Sagan 2002; Dyall et al. 2004; Hölldobler and Wilson 2008; Sachs et al. 2011b). The symbiosis is a classic example of partner-fidelity feedback because not only are the ants and the fungi physiologically linked, but reproduction of the ants and the fungi occurs simultaneously such that selfish behavior by either member could drive the mutualism to extinction. Attine ants reproduce by sending out male and female sexuals on nuptial flights, the females carry a fragment of the natal fungus garden on the flight, which is regurgitated after mating and is used to start a new fungus garden (Quinlan and Cherrett 1978). Consequently, a single line of fungus is inherited from mother queen to daughter queens. Vertical transmission of fungal symbionts is a general rule in attine ants but not universal, because horizontal transmission (the exchange of symbionts) can occur in both basal and derived attine lineages (Mueller et al. 1998, 2011; Green et al. 2002; Mikheyev et al. 2006, 2007; Mehdiabadi et al. 2012). The current paradigm holds that the macroevolutionary patterns in ant-fungal fidelity and monocultures are maintained in a proximate sense by the fungus (the “fungal control model” [FCM]) (Bot et al. 2001; Poulsen and Boomsma 2005). Ant fungiculture is thought to have evolved from a state where ants took advantage of the metabolic activities of fungi to digest plant polysaccharides and other plant compounds that are difficult to digest (Mueller et al. 2001; Schultz and Brady 2008; de Fine Licht et al. 2010; Schiøtt et al. 2010). Like most fungi, the cultivated fungus
Downloaded from http://beheco.oxfordjournals.org/ at University of Texas at Austin on November 30, 2012
Fundamental features of mutualisms are solutions to conflict. The mutualism between fungus-gardening ants and their fungi is characterized by pronounced evolutionary ant-fungus fidelity. Ant-fungus fidelity is thought to be controlled by the fungus, which secretes somatic incompatibility compounds that suppress incompatible fungal strains that may invade gardens. Accordingly, the ants are thought to perceive incompatibilities and then reject invading cultivars. During cultivar-switch experiments, we discovered that if switched colonies of the ant Trachymyrmex septentrionalis are successful in retaining a minute fragment of their original garden, the ants will gradually co-cultivate the novel and the original fungus strains in a chimeric (polycultural) garden; this chimeric garden will revert after 1–2 weeks completely to a monoculture of the original fungal strain. Experimental examination of worker preference suggests that symbiont rejection behavior seems to be an innate response toward foreign fungi rather than mediated solely by incompatibility compounds secreted by the fungus. These observations suggest that the fungal strain cultivated by a colony imprints the attending ants, and this imprinting modifies ant behavior so that the ants prefer their original cultivar even when experimentally forced to grow a foreign cultivar for several weeks. Fungal monocultures and ant-fungus fidelities in this symbiosis therefore seem to be reinforced by factors intrinsic to both the ants and the fungi. Key words: Attini, chimera, cultivar switch, somatic compatibility, symbiosis. [Behav Ecol]
1251
Seal et al. • Do fungus ants control their crops?
MATERIALS AND METHODS Trachymyrmex septentrionalis is a common ant in the sandy soils of eastern North America, occupying a region that extends from Texas and Florida to New York and west to at least Illinois and Missouri (Rabeling et al. 2007). This species is among the most abundant and conspicuous ants in pine forests throughout the southeastern United States (Seal and Tschinkel 2006; Seal and Tschinkel, 2010). Its western range in Texas overlaps with those of T. turrifex and the leaf-cutting ant, Atta texana. Atta texana cultivates a single fungal species typical of the vast majority of leaf-cutting ants, which in telomorph (sexual) form is called Leucocoprinus gongylophorus (Mueller et al. 2010, 2011). However, because this species rarely reproduces sexually (Fisher et al. 1994; Pagnocca et al. 2001; Mikheyev et al. 2006), we use here the name of the anamorph (asexual) form, Attamyces bromatificus (Kreisel). In contrast, ants in the genus Trachymyrmex generally cultivate a more diverse assemblage of taxonomically unresolved, but closely related Leucocoprinus lineages. Trachymyrmex septentrionalis, for instance, cultivates at least 4 distinct fungal phylotypes (Mikheyev et al. 2008). Consequently, we refer here to the fungi cultivated by Trachymyrmex ants using the provisional name “Trachymyces.” Colonies were collected and maintained in the laboratory using methods identical to those in (Seal and Tschinkel 2007a, 2007b). Complete colonies were collected by excavating a 1 m3 pit approximately 30 cm from the nest entrance (Seal and Tschinkel 2008). Each colony was housed in a tray coated with Fluon© (Northern Products, Woonsocket RI) along the sides to prevent escapes. The ants grew their garden in a cylindrically-shaped, 196 cm3 depression in a box lined with dental plaster (Figure 1). Colonies were fed a mixture of substrates highly preferred by T. septentrionalis, oak catkins (staminate flowers) and feces from the eastern tent caterpillar reared on cherry, peach and pear leaves (Prunus and Pyrus spp. respectively) (Seal and Tschinkel 2007a, 2007b, 2008). Catkins and early spring leaves were obtained from the Texas live oak (Quercus fusiformis) and red oak (Quercus texana). Upon collection, fungal substrates were stored in the freezer (−20° C). Experimental design and chimera formation Chimeric fungus gardens were observed during a cultivar switch experiment conducted in March–May 2011 that involved switching T. septentrionalis colonies onto fungus from the leaf-cutting ant Atta texana, following methods described previously (Seal and Tschinkel 2007a). Fourteen T. septentrionalis colonies were collected in Florida (8–10 March, 2011) and 9 in Texas (7–28 February and 21 March 2011). In both of these populations, collections occurred just after the ants ended their winter dormancy. The colonies in Texas were collected at the University of Texas’ Stengl “Lost Pines” Biological Station (30°5′13.1″N, 97°10′25.5″W), and the colonies in Florida were collected in the Wakulla District of the Apalachicola National Forest near Tallahassee, Florida (30°22′46.3″ N, 84º20′6.5″ W), studied previously in detail (Seal and Tschinkel 2006, 2007a, 2007b, 2008). After collection, all colonies were briefly chilled (4°C) and weighed. Eight of the 14 T. septentrionalis colonies from Florida and 9 from Texas received fungus garden material from an A. texana colony, while the remaining received a T. septentrionalis cultivar from a different nest. In all cases, colonies had their native Trachymyces gardens replaced with approximately 300 mg of Attamyces garden from an Atta texana colony that had been reared from a newly mated queen collected after a mating flight in Austin, Texas in May 2007. The effect of removing a
Downloaded from http://beheco.oxfordjournals.org/ at University of Texas at Austin on November 30, 2012
secretes exo-enzymes, and the ants imbibe these enzymes when feeding on the fungus. The ants later defecate the fungal enzymes, which pass through the ant gut unmodified (Martin 1987; Schiøtt et al. 2010), on substrates that the ants collect while foraging (leaves, flowers, insect feces, etc.) (Seal and Tschinkel 2007b). An important feature of fungi is that hyphal secretions can contain somatic incompatibility compounds that stimulate rejection reactions in fungi of different genotypes (Worral 1997), where the incompatibility response is positively correlated with genetic distance between the interacting fungal strains (Hansen et al. 1993). According to the FCM, the ants perceive somatic incompatibility among genetically different strains and react accordingly by killing and removing foreign, incompatible cultivars (Bot et al. 2001). The intensity of their behaviors is correlated with genetic distances between interacting fungi, so that the more distantly related a foreign cultivar is, the greater is the likelihood of rejection by the ants (Poulsen and Boomsma 2005). Furthermore, ants can be purged of their original cultivar’s incompatibility compounds and adopt those of a foreign cultivar after a period of 10 days of repeated exposure to a novel fungus (Bot et al. 2001). Such rejection mediated via ant recognition mechanisms is thought to occur much more rapidly than rejections mediated solely via mycelial contact (minutes or hours vs. days, respectively) (Poulsen and Boomsma 2005). Apart from the studies discussed above, no additional studies have addressed the FCM in greater detail. It is unknown whether the ants have a choice in determining the genotype of fungus they are cultivating. The FCM predicts that, after adopting a foreign fungus for sufficient time (i.e., longer than 10 days), workers should behave aggressively toward their original fungus, if their behavior is a sole consequence of interacting with the fungal symbiont. Several current studies seem to contradict general applicability of the FCM by demonstrating that attine colonies can readily adopt novel cultivars without rejection (Seal and Tschinkel 2007a; Sen et al. 2010). It is thus unclear whether monocultures are in fact maintained by the fungus or by the ants using other cues to maintain their monocultures. Sen et al. (2010) found that Atta texana ants did not discriminate among 8 different fungal genotypes and often combined different genotypes of the same fungal species (Attamyces bromatificus) into intercropped gardens, some of which coexisted for months in a chimeric state. Experimental symbiont switches between the fungus-gardening ant Trachymyrmex septentrionalis and a cultivar symbiont from the leaf-cutting ant Atta texana occurred without overt rejection behavior; nor did the switches incur any clear costs in performance (fitness) (Seal and Tschinkel 2007a). The latter was surprising because the corresponding genetic distances between the fungal species of A. texana and T. septentrionalis are about the highest possible within the higher-attine fungal clades (Chapela et al. 1994; Mikheyev et al. 2010). We examine here several predictions of the FCM by repeating the cultivar-switch experiment on T. septentrionalis conducted originally by Seal and Tschinkel (2007a). We found that, if switched colonies were successful at retaining even a miniscule piece of their old garden, the ants in a colony will form a chimeric garden over the course of a month, and this chimeric garden will revert within 1–2 weeks back to the original species. These observations suggest that the workers in a colony have an innate preference for their native fungal species. Workers therefore retain their original fungal preference even when experimentally switched to a foreign fungus, which contradicts the FCM. Furthermore, behavioral preferences seem to operate independently of hypothesized fungal incompatibility compounds. Thus, the monocultures in this symbiosis seem to be reinforced by factors intrinsic to both the ants and the fungi.
1252
Behavioral tests of ant-cultivar fidelity To test the prediction of the FCM that symbiont fidelity is controlled by the fungal partner, the behavior of switched and unswitched gardening workers was observed in the presence of their own type of cultivar (Trachymyces) or a foreign Attamyces cultivar. Gardening workers from chimeric colonies were not included in these tests. To increase sample sizes required for statistical analysis (n > 10 [Sokal and Rohlf 1995]), we included 5 colonies that had been collected in April 2010 and were fungus-switched in July 2010, thus yielding a total of 11 colonies cultivating Attamyces at the time of our assays. Because the colonies switched in July 2010 produced very few offspring in the subsequent 5 months, the gardening workers in all subcolonies monitored here were at least 1 year old. Six weeks after the beginning of the experiment, 10 workers were removed from the fungus garden of 11 switched colonies and 10 unswitched (control) colonies. Unlike leaf-cutting ants where morphological castes may have different responses toward foreign fungi (Ivens et al. 2008), T. septentrionalis workers exhibit weakly polymorphic workers and size is not typically correlated with specific tasks (JNS, unpublished data). Gardening workers were therefore identified on the basis of behavior and location: these workers were not among those that were alarmed by the sudden opening of the plexiglass nest covers and instead remained calmly in the garden matrix. Thus, the individuals removed from the fungus gardens were assumed to be gardening workers and presumably engaged
in the uptake of fungal enzymes and distribution of these enzymes on substrates and other parts of the garden. The 10 workers were split into 2 groups of 5 and placed for approximately 2 h into a plastic box lined with moistened plaster. After this acclimation period, each group of 5 ants was given approximately 30 mg of fungus garden from either a colony of the leaf-cutting ant Atta texana cultivating Attamyces fungus or a colony of T. septentrionalis cultivating Trachymyces fungus. One hour later, the ants were provided with 5 mg of oak catkins. Ants were observed for 7 days, after which no further observations were conducted. Behaviors were classified into 2 groups: actions toward the fungus garden were inferred to be either positive (provisioning the fungus garden with the catkins) or negative/neutral (no obvious provisioning, which was often accompanied by ignoring or destroying the garden fragment). Frequencies of these outcomes were analyzed in 2 × 2 contingency tables using Fisher’s Exact test. Additionally, we conducted post hoc and a priori power analyses when a significant value was 0.05 < P 90% of the cultivar-switched colonies growing Attamyces preferred Trachymyces (Table 2, lower part). A complete independence of fungal compatibility compounds in the ant-fungal recognition system probably explains why, even after cultivating a novel fungal strain (Attamyces) for a month in a healthy garden containing brood, the ants allowed this garden to be taken over by their original Trachymyces strain (Figure 1). Chimera formation and the behavioral tests of cultivar preferences of gardening workers seem to rule out a sole, proximate role of fungal incompatibility-factors in constraining cultivar switches. Because all visible pieces of Trachymyces had been experimentally removed prior to switching, and because colonies had grown sizeable Attamyces gardens by the time chimeras formed (Figure 1), the Trachymyces that eventually replaced the Attamyces biomass must have been grown
Downloaded from http://beheco.oxfordjournals.org/ at University of Texas at Austin on November 30, 2012
Figure 1 Photograph of colony ANF1 with a chimeric fungus garden. The darkened portion of the fungus garden above the white line was determined by genotyping to be Trachymyces, whereas the lighter portion was Attamyces. The photo was taken on day 31 of the experiment, but by day 40 the entire garden was comprises of Trachymyces. Notice the lack of clear separation and lack of overt fungus–fungus incompatibility at the interaction zone of the 2 fungal species.
The interactions of subcolonies of gardening workers generally contradicted predictions of the FCM model. Gardeners from colonies growing native Trachymyces exhibited positive behaviors toward foreign Trachymyces fragments and negative responses toward Attamyces. This supported the FCM but is also consistent with the IPM (Table 2). However, the vast majority of workers from colonies that had been cultivating Attamyces exhibited positive interactions with Trachymyces and at best mixed interactions with Attamyces (Table 2). Although this ran counter to the FCM (because the gardeners did not exhibit a clear preference for the fungal species they were currently cultivating), the Fisher’s Exact test was only marginally not significant (P = 0.06). This result appears to be driven by gardeners from only 5 switched colonies interacting positively with Attamyces. Power analyses indicate that the post hoc power of this test was approximately 0.49, thus the probability of making a Type II Error (β) (failing to reject the null hypothesis when it is false) was 0.51. Thus, there were nearly equal chances of incorrectly rejecting or accepting the null hypothesis. A priori power analyses on these frequencies indicate that a total sample size of 18 switched colonies growing Attamyces (and then exposed gardeners from 18 colonies to each fungal species [Attamyces or Trachymyces or 36 subcolonies for the Fisher’s Exact Test]) would be required to achieve a power of at least 0.80.
1254
Behavioral Ecology
Table 1 Dates and colony size parameters for colonies of Trachymyrmex septentrionalis that were switched onto Attamyces bromatificus and that then showed satellite Trachymyces fragment or developed chimeric gardens Day of formation
Colony ID
Colony weight (mg)
Fungus garden volume (cm3)
Percent of chamber occupied by Attamyces garden
Transition time to Trachymyces
9 9 15 17 17 31
ANF 6 TX 5 ANF 8 TX 2 TX 6 ANF 1
715 1126 438 779 494 592
48 50 74 137 113 191
25 26 38 70 58 97
— 6 days — 8 days 9 days 8 days
Day of formation indicates the date after which the experiment commenced, which for the Florida and Texas colonies was 22 March and 28 March, respectively. Transition time to Trachymyces indicates the number of days for complete succession from Attamyces to Trachymyces.
Table 2 Results of preference bioassays: shown are the frequencies (number of subcolonies) where gardening workers showed positive (acceptance) or negative (rejection) reactions toward fungus as a function of the origin of the gardeners (by fungal species) and the type of fungal species offered Origin of gardeners
Fungal species offered
Positive
Negative/Neutral P-value
Trachymyces
Trachymyces
9
1
×100) (Table 1). Strict application of the FCM would predict the killing of the small Trachymyces snippets, especially in colony ANF 1, because Trachymyces was present in a far lower quantity, yet we observed that Trachymyces thrived in the garden despite the biomass dominance of Attamyces (Figure 1, Table 1). Although ants in switched colonies should be in full contact with incompatibility-factors
Downloaded from http://beheco.oxfordjournals.org/ at University of Texas at Austin on November 30, 2012
Figure 2 (A) Photograph showing inhibition between Attamyces clones (left side of plate) and Trachymyces clones (right side of plate). The darkened melanized zone indicates the contact zone between the 2 fungal species. (B) Photograph showing a lack of inhibition (lack of somatic incompatibility) between genetically identical Trachymyces clones.
produced by Attamyces, and because ant behavior should be modified accordingly (Bot et al. 2001), when fungus-switched gardeners in isolation were provided with fresh Attamyces garden-fragments, their behavior was at best ambivalent, whereas their nestmate workers promptly provisioned the Trachymyces fragments. These findings all appear to be inconsistent with the FCM. Although the low power prevents a more definitive conclusion, the data presented here (e.g., Table 2) would seem to suggest that the FCM and its emphasis on the use of fungal compatibility compounds does not entirely explain the maintenance of monocultures in this species. Some conditioning by fungally derived cues cannot be entirely ruled out, because some ants that had cultivated Attamyces did not reject Attamyces (Table 2). Imperfect conditioning is also possible because some colonies whose gardeners did not reject Attamyces did not reject Trachymyces either. If observations reflect and underlying trend (that gardeners prefer their native Trachymyces over Attamyces), and if the experiment had included the minimum sample size for adequate power (i.e., 7 additional colonies), the logical conclusion is that the result would have either statistically supported Alternative 2 (the IPM) or produced a statistically insignificant result not unlike we report here, but with the power to confidently rule out a false negative. If so, the available data would not seem to offer explicit support for the FCM. The results here do not necessarily mean that monocultures are not controlled ultimately by the fungus, rather it leaves open the possibility that the fungus may nevertheless imprint ants using other cues. For example, the ant-cultivated fungus is known to impart chemical signatures onto workers that nestmates use as recognition cues (Viana et al. 2001; Richard et al. 2004, 2007); consequently, gardening workers may use odors (and perhaps taste/gustatory cues) to distinguish a garden of their native cultivar species from gardens composed of alien cultivar species. Future work should address which cues the ants are using to differentiate between their fungal strains and how this may interact with their behavioral repertoires as individual ants grow and develop. It is uncertain how much of the succession from Attamyces to Trachymyces was due to active involvement of the ants or the fungus. The process started clearly under the control of the ants when they began cultivating a satellite garden fragment off to 1 side and gradually this fungus grew as the ants merged this fragment into the main mass of Attamyces. In colony ANF1 (Figure 1), for example, the spread of Trachymyces occurred without much change in the overall physical appearance of the fungus garden (i.e., without overt fungus–fungus incompatibility interactions typical for basidiomycete fungi). During these initial stages, it is possible, though speculative, that Trachymyces-type nuclei migrated into the
1255
Seal et al. • Do fungus ants control their crops?
significant selective pressures that these symbioses experienced as they migrated into temperate or subtropical America (Mueller et al. 2011). One of the more extreme adaptations to the temperate zone is exhibited by Trachymyrmex septentrionalis, which reduces its garden to nearly microscopic fragments (
