Fungal Adaptations to Mutualistic Life with Ants

faculty of science university of copenhagen centre for social evolution department of biology

Fungal adaptations to mutualistic life with ants

This thesis dissertation has been submitted in accordance with the requirements for the degree of PhD, at the PhD School of The Faculty of Science, University of Copenhagen, Denmark to be defended publicly before a panel of examiners.

by Pepijn W. Kooij December 2013

Academic advisors Prof. Jacobus J. Boomsma Dr. Morten Schiøtt Cover Front: Atta sexdens workers carrying leaves down a tree. Picture: P. W. Kooij, Location: Santa Cruz, Gamboa, Panama

Back: incipient colony of Atta colombica with queen on top. Picture: P. W. Kooij, Location: Gamboa, Panama Preface

This thesis is the result of a three-year PhD project carried out at the Cen- tre for Social Evolution (CSE), Department of Biology, University of Copenhagen in Denmark under the supervision of Professor Jacobus J. Boomsma and Dr. Morten Schiøtt. During the project I spent five weeks at the Laboratory of Genetics, Wagenin- gen University and Research Center, The Netherlands, hosted by Dr. Duur K. Aanen. I additionally carried out a total of seven weeks of fieldwork at the Smithsonian Tropical Research Institute in Panama. I was funded by the Danish National Research Founda- tion and the Department of Biology.

The thesis is comprised of a general introduction on the current understanding of life-time mutualistic interactions and how this can be applied to my model system where the fungal symbiont adapted to a life-time commitment with ants. I also present the objectives of the thesis research, and conclude with a summary of the main results in a broader conceptual framework. This is followed by five chapters of original emperical work, one of which is in press in a peer-reviewed journal, one has been submitted to a peer-reviewed journal and three that are in preparation for submission.

Pepijn W. Kooij Copenhagen, December, 2013

Contents

Summaries...... 7

General introduction...... 13

Mutualism, commitment and co-adaptation...... 14

The fungus-growing ants as model system...... 15

Open questions and objectives of this thesis...... 17 The influence of domestication on fungal genetics...... 17 Incompatibility between fungal crops of different ant genera...... 18 Decomposition enzymes in the fungus-growing ant symbiosis...... 19 Special Acromyrmex echinatior fungal symbiont enzymes...... 20

Summary and future perspectives...... 21

References...... 27

Chapter 1:...... 33 Advanced farming ants rear polyploidy crop fungi Chapter 2:...... 61 Somatic incompatibility of fungal crops in sympatric Atta and Acromyrmex leaf-cutting ants Chapter 3:...... 85 Differences in forage-acquisition and fungal enzyme activity contribute to niche segregation in Panamanian leaf-cutting ants Chapter 4:...... 109 Leucoagaricus gongylophorus uses leaf-cutting ants to vector proteolytic enzymes towards new plant substrate Chapter 5:...... 133 Cellulose degradation patterns in Acromyrmex echinatior colonies

Acknowledgements...... 153 6 English, Dansk, Español, Nederlands Summaries 7 Summary

Fungus-growing ants (Attini) feed off a fungus they cultivate in a mutualistic symbiosis in underground chambers by providing it substrate they collect outside the colony. The tribe of Attine ants ranges from small colonies of the paleo- and basal Attine species with a few hundred workers that forage on crude substrates such as insect frass and dry plant material, to large colonies of the leaf-cutting ants with several thousands to several million workers that provide live plant material to their fungus gardens. Leaf-cutting ants are the dominant herbivores of the Neo-tropics, and have a major contribution to cycling of nitrogen and phosphorus in their direct environment and are, furthermore, considered pest species as they have a large impact on human agriculture. These factors make leaf-cutting ants an ideal study subject to better understand the mechanisms that make this mutualistic symbiosis so successful. To understand the evolutionary development of domestication of the fungus over the phylogeny of the Attine ants, I compared the average number of nuclei per cell for the fungal symbionts, for each of the different groups of fungus-growing ants. I found that the fungal symbionts of the paleo- and basal Attine ants, which have a relative low level of domestication, have two nuclei per cell, the standard for Basidiomy- cete fungi, but that the average number increased to 7-17 nuclei per cell in the highly domesticated fungi of the higher Attine ants and leaf-cutting ants. Furthermore, I was able to estimate that approximately half of these nuclei were represented by different genomes, giving the fungus a ploidy level of 5n-6n. In mutualistic symbioses it is important the partners stay true to each other. In fungus-growing ants, new founding queens bring a piece of fungus to build up their new colony. However, in rare occa- sions fungal symbionts might come into contact with symbionts from other colonies. I showed that in both leaf-cutting ant genera incompatibility reactions between fungal strains can avoid intermixing of different strains, and that these reactions strengthen when genetic distance is increased. This pattern, however, becomes distorted when fungal symbionts are contested across ant genera. The most important mechanism in the succession of this mutualism of leaf-cutting ants is the controlled degradation of plant material. I show that in the area of Gamboa, Panama, the two leaf-cutting ant genera forage for rather different plant material, with Atta species specializing on tree-leaves and Acromyr- mex focusing more on flower material and herbal plant material. This difference is reflected in the overall enzyme activity patterns in the fungus gardens, with Atta specializing more on specific enzyme groups and Acromyrmex having an overall high enzyme activity. Finally, I show that the fungal symbiont of the leaf-cutting ant Acromyrmex echinatior produces large amounts of biodegrading enzymes in special structures called gongylidia. The ants eat these structures, but enzymes pass the ant gut without being digested, and are excreted by the ants in their fecal fluid which they mix with freshly foraged plant material placed on the top of the fungus garden. The enzymes are still active and have therefore an important role in the biodegradation of the plant material. With this I show that the fungus evolved some incredible adaptations to a mutualistic life with the ants.

8 Resumé

Svampedyrkende myrer lever af en svamp de dyrker i en mutualistisk symbiose i underjordiske kamre ved at forsyne den med substrat som de henter udefra. Underfamilien af svampedyrkende myrer (At- tini) spænder over paleo- og basale arter med få hundrede arbejdere der fouragerer på tørret plantemateriale og insekt ekskrementer, til store kolonier af bladskærermyrer med flere tusinde til millioner af arbejdere der indsamler friskt plantemateriale til deres svampehaver. Bladskærermyrer er den dominante herbivor i neotroperne hvor de bidrager væsentligt til nitrogen og fosfor kredsløbet i deres nærmiljø og bliver ydermere betragtet som et skadedyr idet de har stor effekt på landbruget. Disse faktorer gør det ideelt at benytte bladskærermyrer til bedre at forstå mekanismerne der har gjort denne mutualistiske symbiose så succesfuld. For at forstå udviklingen af domesticeringen af svampen over hele den Attine fylogeni, sam- menlignede jeg det gennemsnitlige antal kerner per celle i svampesymbionten hos forskellige grupper af svampedyrkende myrer. Jeg fandt at svampesymbionten hos myrerne fra paleo- og basale arter, som har relativ lavt niveau af domesticering, havde to kerner per celle som er standard for Basidimyceter, hvorimod gennemsnittet steg til 7-17 kerner per celle hos de meget domesticerede svampesymbionter fra de højere udviklede svampedyrkende myrer og bladskærermyrerne. Ydermere kunne jeg estimere at ca. halvdelen af disse kerner havde forskellige genomer hvilket gav svampen et ploidi niveau på 5-6n. I en mutualistisk symbiose er det vigtigt at parterne er tro mod hinanden. Hos svampedyrkende myrer medbringer de nye dronninger et stykke svamp til at grundlægge nye kolonier, men i sjældne tilfælde kan denne svampesymbiont komme i kontakt med symbionten fra en anden koloni. Jeg påviste at i begge bladskærermyrerslægter sørger inkompatibilitets reaktioner imellem varianter af svampe for, at symbionter ikke bliver blandet og at dette styrkes når den genetiske variation forøges. Dette mønster bliver dog for- vrænget når svampesymbionter fra forskellige slægter af myrer testes imod hinanden. Den vigtigste mekanisme i successionen af denne mutualisme hos bladskærermyrer er den kontrol- lerede nedbrydning af plantemateriale. Jeg påviser at i området omkring Gamboa, Panama, fouragerer de to bladskærermyrerslægter på forskelligt plantemateriale, Atta er specialiseret på træblade og Acromyrmex mere på blomster og urter. Denne forskel reflekteres i enzymaktiviteten i svampehaverne, hvor Atta er specialiseret indenfor specifikke grupper af enzymer ogAcromyrmex overordnet har en høj enzymaktivitet. Endelig viser jeg, at svampesymbionten hos bladskærermyren Acromyrmex echinatior producer- er store mængder bionedbrydende enzymer i specielle strukturer kaldet gongylidia. Myrerne spiser disse strukturer, men enzymerne passerer ufordøjet igennem tarmsystemet og ender i myrernes fækalievæske som de blander med friskt indsamlet plantemateriale og placerer øverst i svampehaven. Enzymerne er sta- dig aktive og spiller derfor en vigtig rolle i den biologiske nedbrydning af plante materialet. Dette viser at svampen har udviklet nogle utrolige tilpasninger til en mutualistisk eksistens med myrerne.

9 Resumen

Las hormigas atinas se alimentan de un hongo que cultivan en simbiosis mutualista en cámaras subterráneas al que proveen con sustrato colectado fuera de la colonia. La tribu Attini presenta colonias que oscilan en tamaño desde las pequeñas colonias de las especies basales, con unos pocos cientos de trabajadores que forrajean substratos crudos como heces de insectos y material vegetal seco, hasta las grandes colonias de las hormigas cortadoras de hojas con varios miles a millones de trabajadores que forrajean material vegetal vivo. Las hormigas cortadoras de hojas son los herbívo- ros dominantes del Neotrópico, y contribuyen extensamente a los ciclos de nitrógeno y fósforo en su entorno directo. Por otra parte, estas hormigas son consideradas una especie peste debido a su fuerte impacto en la agricultura humana. Estos factores hacen de las hormigas cortadoras de hojas un grupo de estudio ideal para intentar comprender mejor los mecanismos que hacen tan exitosa esta simbiosis mutualista. Para entender la sucesión de la domesticación a través de la filogenia de las hormigas Attini, se comparó el número promedio de núcleos por célula para los simbiontes fúngicos de cada uno de los diferentes grupos de hormigas cortadoras de hojas. En este estudio encontramos que el hongo simbionte de las hormigas atinas basales, las cuales presentan un nivel relativamente bajo de domesticación, tienen dos núcleos por célula; lo estándar para hongos basidio- micetos. Sin embargo, este número incrementa a un rango de 7 a 17 núcleos por célula en los hongos altamente domes- ticados de los grupos más recientes de hormigas atine y cortadoras de hojas. Asimismo se estimó que aproximadamente mitad de estos núcleos son representados por diferentes genomas, dándole al hongo un nivel de ploidía de 5n-6n. En simbiosis mutualistas, es importante que las partes sean fieles entre sí. En hormigas cultivadoras de hongos, nuevas reinas fundadoras traen un pedazo de hongo para construir su nueva colonia. No obstante, en raras oca- siones, los simbiontes fúngicos puede entrar en contacto con simbiontes de tras colonias. Aquí demostramos que en ambos géneros de hormigas cortadoras de hojas, reacciones de incompatibilidad entre cepas de hongos pueden prevenir combinaciones de diferentes cepas, y que estas reacciones se fortalecen con el incremento de la distancia genética. Sin embargo, este patrón se distorsiona cuando los simbiontes fúngicos se oponen a través de los géneros de hormigas. El mecanismo más importante en la sucesión de este mutualismo de las hormigas cortadoras de hojas es la de- gradación controlada del material vegetal. Se encontró que en el área de Gamboa, Panamá, los dos géneros de hormigas cortadoras de hojas forrajean material vegetal bastante diferente; las especies Atta se especializan en hojas de árboles y las Acromyrmex en material de flores y hierbas. Esta diferencia se refleja en los patrones generales de la actividad enzimática en los cultivos de hongos, donde Atta se enfoca más en grupos enzimáticos especializados y Acromyrmex posee una alta actividad enzimática generalizada. Finalmente, en este estudio demostramos que el simbionte fúngico de las hormigas cortadoras de hojas Ac- romyrmex echinatior produce grandes cantidades de enzimas biodegradadoras en estructuras especiales denominadas gongilidios. Las hormigas comen estas estructuras y las enzimas pasan a través del sistema digestivo de la hormiga, aunque sin ser digeridas. Luego, estas son excretadas en el líquido fecal, el cual las hormigas mezclan con el material vegetal fresco que ha sido recientemente colectado y ubicado en la parte superior del jardín de hongo. Las enzimas permanecen activas y por lo tanto juegan un papel importante en la biodegradación del material vegetal. Con esto dem- uestro que el hongo evolucionó adaptaciones extraordinarias para una vida mutualista con las hormigas.

10 Samenvatting

Mieren van de geslachtengroep Attini eten een schimmel die ze in ondergrondse kamers kweken in een mutualisme. Deze schimmel kweken ze door groeisubstraat te geven dat ze buiten het nest verzamelen. De Attine-mieren zijn divers, met aan de ene kant kleine kolonies van een paar honderd werkers - bij de basale soorten - die ruw materiaal verzamelen zoals insecten-faeces en dood plantenmateriaal, en aan de andere kant grote kolonies - bij de bladsnijmieren - met enkele duizenden en soms miljoenen werkers die levend plantenmateriaal verzamelen. Deze bladsnijmieren zijn de belangrijkste herbivoren in de Neotropen en hebben een grote rol in het recyclen van stikstof en fosfor. Daarnaast worden ze gezien als ongedierte vanwege hun impact op de landbouw. Dit alles tezamen maakt bladsnijmieren ideaal model om te onderzoeken wat de mechanismen zijn die de mutualisme tussen mieren en schimmel zo succesvol maakt. Om te begrijpen hoe de domesticatie van de schimmel evolueerde over de fylogenie van de Attine-mieren, heb ik van de schimmel het gemiddelde aantal celkernen per cel vergeleken voor de verschillende groepen van deze mieren. Uit de resultaten kan ik concluderen dat de minder gedomesticeerde schimmels, die gecultiveerd worden door de basale groep van Attine-mieren, altijd twee celkernen per cel hebben, de standaard voor Basidiomyceten. Maar bij de meer gedomesticeerde schimmels van de hogere Attine-mieren en bladsnijmieren is dit aantal toegenomen tot een gemiddelde van 7-17 celkernen per cel. Daarnaast ben ik in staat geweest om aan te tonen dat ongeveer de helft van deze celkernen een verschillende genoom hebben, wat betekent dat de schimmel een ploïdie-niveau heeft van 5n-6n. Voor een mutualisme is het belangrijk dat beide partners trouw aan elkaar blijven. Bij de Attine-mieren neemt de nieuwe koningin een stukje schimmel mee om de nieuwe kolonie te beginnen. Het kan echter gebeuren dat de schimmel van de ene kolonie in contact komt met de schimmel van een andere. Ik toon hier aan dat bij de schimmels van twee soorten, van twee verschillende genera, van bladsnijmieren incompatibiliteitsreacties kunnen voorkomen dat verschillende lijnen van schimmels met elkaar mengen, en dat deze reacties toenemen naarmate de genetische afstand tussen deze lijnen toeneemt. Dit patroon verdwijnt echter wanneer schimmellijnen van mieren van verschillende genera met elkaar in contact komen. De belangrijkste reden waarom bladsnijmieren zo succesvol zijn is het gecontroleerd afbreken van plantenmateriaal. Ik laat hier zien dat in de omgeving van Gamboa, Panama, de twee bladsnijmierengenera verschillend plantenmateriaal verzamelen. De Atta-soorten zijn gespecialiseerd in boombladeren en de Acromyrmex-soorten richten zich op bloem- en kruidachtig plantenmateriaal. Dit verschil is ook te zien bij de enzymactiviteiten in de schimmeltuinen, waarbij tuinen van Atta-soorten zich specialiseren op specifieke enzymen, maar de tuinen van Acromyrmex-soorten in het algemeen een hoge enzymactiviteit hebben. Tot slot laat ik zien dat de schimmel van bladsnijmierensoort Acromyrmex echinatior grote hoeveelheden enzymen, gespecialiseerd in het afbreken van plantenmateriaal, produceert in speciale organen die gongylidia worden genoemd. De mieren eten deze organen, maar de enzymen daarin worden niet verteerd door de mier en komen in de faeces van de mier terecht. De mieren mengen hun faeces met vers plantenmateriaal en plaatsen dit bovenop de schim- meltuin. De enzymen in de faeces zijn daar nog steeds actief en hebben daardoor een belangrijke rol in het afbreken van het plantenmateriaal. Hiermee laat ik zien dat de schimmel belangrijke adaptaties aan een mutualistisch leven met mieren heeft geëvolueerd.

11 12 General introduction 13 Fungus-growing ants of the tribe Attini live in a tight relationship with crop fungi. This relationship is so tight that the one partner cannot live without the other, making this symbiosis an obligate mutualism (Weber 1966; Howe 1984). For more than a century, substantial knowledge has accumulated about this intriguing symbiosis. However, most of the research has been focused on the natural history of the ants and the system in general (Wheeler 1907; Weber 1966; Mueller et al. 2005), or on the adaptations of the ants to maintain their crops. Some of these studies have investigated differences in worker polymorphism (e.g. Weber 1966; Mehdiabadi and Schultz 2010), the domestication of actinobacteria at the origin of fungus farming 50 MYA to protect the fungus against parasites (e.g. Currie et al. 1999; Currie et al. 2003; Poulsen and Currie 2009), the development of antibiotic-producing metapleural glands (see e.g. Do Nascimento et al. 1996; Hughes et al. 2008; Yek et al. 2012), followed by studies on biological control by the ants versus these evolutionary derived chemical controls (see e.g. Fernandez-Marin et al. 2007; Fernandez-Marin et al. 2009).

However, far less is known about the fungal adaptations to a mutualistic life with ants, and the research to date has tended to focus on the role of the fungus from the ants perspective (e.g. Martin 1970; Martin 1974; Quinlan and Cherrett 1979; Bass and Cherrett 1995). In recent years, research- ers have shown an increased interest in how the fungal symbiont adapted to make this mutualism one of the dominant herbivores in the neo-tropics (e.g. Schiøtt et al. 2008; Schiøtt et al. 2010; Semenova et al. 2011; De Fine Licht et al. 2013; Aylward et al. 2013). Therefore, in this thesis I attempt to further explore fungal adaptations by investigating fungal morphology, genetics and biochemistry. The first section will give an overview of the research presented in this thesis within a broad perspective of mutualistic life-style, commitment to mutualistic partners and the need for co-adaptations to sustain and elaborate mutualisms that have become obligate. Next, I will introduce my model system of fungus-growing ants and what is already known about the roles of the fungal symbiont in the mutualism. Finally, I’ll conclude with an outline of how I investigated fungal adaptations to mutualistic life with ants.

Mutualism, commitment and co-adaptation

“Despite their different fundamental organization, ant colonies and mycelia of fungi exhibit striking similarities in their social organization.” (Rayner and Franks 2003)

In order for a stable mutualistic symbiosis to be maintained it is necessary to have a high degree of partner commitment and functional complementarity (Janzen 1985; Keeler 1985). This can

14 mostly be observed in endosymbioses where the levels of interdependency are so high that the symbiont becomes part of the cells and tissues of the host. However, also in ectosymbioses like the fungus-growing ant mutualism, where the fungus lives outside the ants, these high levels of partner commitment and functional complementarity can be seen, because the fungus acts as an ectosymbiont for the individual ants, but as an endosymbiont for the colony (Poulsen and Booms- ma 2005). This can lead to a reduced genetic diversity and sometimes even asexuality of the endosymbiont compared to its host (Law and Lewis 1983; Law 1985), something that was later shown to be the result of co-speciation, which is present in tight mutualisms (Herre et al. 1999). Another condition to maintain a stable mutualism, is that the host should only benefit relatives of the other partner, which will then stabilize the mutualism, but can also lead to reduced population-wide genetic variability in the partner that becomes domesticated (Frank 1994; Doebeli and Knowlton 1998; Sachs et al. 2004; Foster and Wenseleers 2006).

Hosts and symbionts are potentially in conflict over the direction of symbiont transmission (horizontal vs vertical), and the magnitude of symbiont dispersal (Frank 1996a; Frank 1996b; Douglas 1998; Douglas 2008). Hosts may therefore reduce the production of sexual symbiont investments, because this may incur cost to the host and because the host risks being confronted with intro- ductions of new symbiont strains that may reduce host fitness (Frank 1996a; Frank 1996b; Leigh 2010). Symbiont internal conflicts (antagonism or incompatibility) might be beneficial for hosts to maintain control of symbiont acquisition (Frank 1996a; Frank 1996b; Bot et al. 2001b). Due to these factors the symbionts can get so integrated with their hosts that their genomes reduce in such manner that a life without the host is merely impossible (McCutcheon and Moran 2012). At this point the host-symbiont conflicts seem to be resolved, so that high levels of cooperation and low levels of conflict will lead the symbiosis towards organismality (Queller and Strassmann 2009) which in turn is maintained by life-time commitments of both partners (Boomsma 2013).

The fungus-growing ants as model system

“Among the multitudinous activities of insects, none are more marvelous than the fungus-growing and fungus-eating habits of Attiine ants.” (Wheeler 1907)

The tribe of fungus-growing ants, the Attini, which contains ca. 14 genera and >230 species, evolved approximately 50 MYA from a hunter-gatherer ancestor towards obligate farming of a fungal crop (Schultz and Brady 2008; Mehdiabadi and Schultz 2010) which they maintain by providing growth substrate and by removing of any contaminations (Möller 1893; Wheeler 1907;

15 Weber 1966; Weber 1972; Mueller et al. 2005). The ants cultivate their crop in monoculture and by default transmit their crop vertically with incipient queens to new colonies (Möller 1893; Wheeler 1907; Weber 1955; Weber 1972; Mueller et al. 2005; Poulsen and Boomsma 2005; Mueller et al. 2010). However, horizontal transmission seems to be common between shallow phylogenetic branches (Mikheyev et al. 2006; Mikheyev et al. 2007; Poulsen et al. 2009), but more constrained between older lineages (Mueller et al. 1998; Johnson and Vilgalys 1998; Vo et al. 2009)

The ants can be divided in different groups of farming, with the lower agriculture attines at the base of the phylogenetic tree, mainly growing fungi from the Leucocoprinaea tribe, including the specialized yeast agriculture clade, which evolved to grow the same fungal species in yeast form, and the coral fungus agriculture in the genus Apterostigma, which cultivate a fungal crop from the Pterula family (Mueller et al. 1998; Munkacsi et al. 2004; Schultz and Brady 2008; Mehdiabadi and Schultz 2010). This group of lower attine ants provide their fungal crops with crude substrates such as insect frass, nectar, seeds and dead and dry plant material (De Fine Licht and Boomsma 2010). The ants and fungi in these clades do not show signs of tight coevolution, with still free-living fungal relatives in both the Leucocoprinaea clades (Chapela et al. 1994; Mueller et al. 1998; Mueller 2002; Vo et al. 2009) and the Pterula clades (Chapela et al. 1994; Mueller 2002; Munkacsi et al. 2004; Villesen et al. 2004; Dentinger et al. 2009), and signs of horizontal transfer between ant species (Mueller 2002; Green et al. 2002; Mikheyev et al. 2010; Kellner et al. 2013). But when focused on terminal clades of the ant phylogeny, such as the Cyphomyrmex wheeleri group, a strict coevolution between the ants and fungi can be seen (Mehdiabadi et al. 2012).

Several changes in the symbiosis occurred with the introduction of the higher attine ants of the genera Trachymyrmex and Sericomyrmex 20 MYA(Schultz and Brady 2008; Mehdiabadi and Schultz 2010). The colonies have an increased size of approximately ten fold (Fernandez-Marin et al. 2004; Schultz and Brady 2008; Baer et al. 2009; Fernandez-Marin et al. 2009; Mehdiabadi and Schultz 2010), the growth substrate for these genera now includes fresh leaves, flowers and fruits (De Fine Licht and Boomsma 2010) accompanied by differences in fungus garden enzyme activities (De Fine Licht et al. 2010). Furthermore, lack of free living close relatives shows increased level of domestication (Mikheyev et al. 2006; Mikheyev et al. 2010), and specialized hyphal structures, gongylidia, which act solely as food for the ants found their origin in these fungal strains (Möller 1893; Weber 1955; Quinlan and Cherrett 1978; Quinlan and Cherrett 1979; Bass and Cherrett 1995).

The final step was the evolution of the leaf-cutting ant genera Atta and Acromyrmex 10 MYA

16 (Schultz and Brady 2008; Mehdiabadi and Schultz 2010), which are considered major herbivores in the neo-tropics and are major contributors of nitrogen and phosphorus recycling, but are also often seen as pest species (Cherrett and Peregrine 1976; Fowler et al. 1989). The switch to primarily live plant material with the addition of flowers and fruits (De Fine Licht and Boomsma 2010) caused another shift in overall fungus garden enzyme activity towards plant cell wall degradation (De Fine Licht et al. 2010) and allowed the colonies to grow to sizes of several thousands of workers up to several millions (Fernandez-Marin et al. 2004; Schultz and Brady 2008; Baer et al. 2009; Fernandez-Marin et al. 2009; Mehdiabadi and Schultz 2010). Along with this increase in colony size, came an increase in worker polymorphism (Weber 1966; Mehdiabadi and Schultz 2010) and queen mating frequency (Villesen et al. 2002; Baer et al. 2009; Mehdiabadi and Schultz 2010) to allow the colonies to become this large. This together shows that the ecological footprint increased with every single step in the evolution of the fungus growing ants, which was supported by the arrival of the gongylidia 20 MYA and a recent horizontal sweep of the cultivars in the leaf-cutting ants 2-4 MYA replacing all fungal symbionts in the ant clade with a new one (Mikheyev et al. 2010).

Open questions and objectives of this thesis The influence of domestication on fungal genetics

“The study of the Attini and other fungus-growing insects has only just begun, and further ad- vance in this fascinating subject will be more difficult for the mycologist than for the entomolo- gist.” (Wheeler 1910)

Domestication of crop fungi can bring complications for the fungal symbiont such as degenera- tion of genetic diversity due to asexuality caused by vertical transmission (Möller 1893; Wheeler 1907; Weber 1955; Weber 1972; Mueller et al. 2005; Poulsen and Boomsma 2005; Mueller et al. 2010), or mushroom castration by the ants (Fisher et al. 1994a; Fisher et al. 1994b; Mueller 2002), personal observations M Schiøtt, HH De Fine Licht and PW Kooij). However, there are signs of horizontal transfer in the fungal symbionts of the higher attine ants (Mikheyev et al. 2006; Mikhe- yev et al. 2007; Poulsen et al. 2009), and recent studies showed signs of polyploidy in these fungi (Scott et al. 2009; De Fine Licht et al. 2013), but no further research has been done specifically on the latter.

For comparison, in the human-domesticated Agaricus bisporus an increase in the number of haploid nuclei (up to 26)(Saksena et al. 1976) was shown compared to its free-living sister species

17 Agaricus bitorquis (Raper 1976; Hintz et al. 1988). A similar increase in number of nuclei can be seen in the termite-domesticated fungus Termitomyces with up to 12 nuclei per cell (De Fine Licht et al. 2005). The fungus-growing ants have different levels of domestication of fungal crops, from the paleo- and basal agriculture with free living relatives of the fungal symbionts (Mueller et al. 1998; Mueller 2002; Munkacsi et al. 2004; Schultz and Brady 2008), to more complex domesticated agriculture by the higher attine ants and aggressive herbivore agriculture by the leaf-cutting ants with no free living relatives of the symbionts (Mikheyev et al. 2006; Schultz and Brady 2008; Mikheyev et al. 2010). Because of these different levels of domestication these ants are a perfect subject to investigate the influence of domestication on increasing number of nuclei in the fungal cells and the level of ploidy of the crop fungi. Thus, the questions remain, how is the average number of nuclei correlated with increased domestication? And, does the level of ploidy correlate with increased domestication?

Incompatibility between fungal crops of different ant genera

“A leaf-cutter ant colony is a mutualist to on large plant (its fungus), parasite to many others, and commensal to yet others” (Janzen 1985)

Each individual colony of leaf-cutting ants maintains a single strain of the fungal symbiont Leu- coagaricus gongylophorus in monoculture (Acromyrmex, Poulsen and Boomsma 2005; Atta, Mueller et al. 2010). The maintenance of the monoculture is done though behavioral adaptations of the ants, e.g. weeding of foreign strains, (Bot et al. 2001b; Ivens et al. 2009) or by somatic incompatibility reactions by the fungal strain itself (Poulsen and Boomsma 2005). In Basidomycete fungi, such as the fungal symbiont of the ants, such stepwise incompatibility reactions regulate the allorecognition, by rejecting a foreign strain using programmed cell death and cell pigmentation at the meeting point of the two different fungal strains (Rayner et al. 1984; May 1988; Rayner 1991; Worrall 1997). The exact genetic mechanisms involved in these reactions are still unknown, but multiple loci are involved, and the reactions are correlated with genetic distance (Worrall 1997).

The ants often kill alien strains that are genetically different from their native strain, but this antagonism disappears after force feeding the ants with a foreign strains (Bot et al. 2001b; Ivens et al. 2009). Furthermore, when the ants are almost entirely depleted from their native fungus, and provided with a new foreign fungal strain, the original native strain will grow back and take over the whole colony again within two weeks (Seal et al. 2012). It has been suggested that this is caused by imprinting of the ants by the native fungal strain (Seal et al. 2012), which could be

18 done by using chemical recognition profiles (Richard et al. 2007). However, it was shown that the fungal symbionts of Acromyrmex show somatic incompatibility on agar plates, which gradually increases with genetic distance (Poulsen and Boomsma 2005). The same study showed that the fecal fluid of the ants showed increasing levels of incompatibility on foreign fungi with increasing genetic distance, but that these reactions disappeared when the ants were force fed with the foreign fungus, suggesting the incompatibility reactions are of fungal origin. The somatic incompatibility in this system could therefore be explained by the lifetime commitment of the symbiont to live as a monoculture with the ants, and to avoid conflicts within the symbionts (Boomsma 2013). However, the questions, remaining are: Does the incompatibility in Atta fungal crops show similar incompatibility mechanisms as Acromyrmex fungal crops? And, how are the incompatibility mechanisms between fungal symbionts across the two ant genera?

Decomposition enzymes in the fungus-growing ant symbiosis

“I believe that they are, in reality, mushroom growers and eaters.” (Belt 1874)

The leaf-cutting ant mutualism is based in the degradation of fresh plant material, transported by the ants to the colony, by the fungal symbiont, which in turn provides the ants with nutrients (We- ber 1966; Bass and Cherrett 1995; De Fine Licht and Boomsma 2010). However, the degradation of fresh plant material brings many complications due to the complex structure of the plant cell walls, such as the plants defense mechanism in the shape of phenolic compounds (Coley et al. 1985; De Fine Licht et al. 2013), strong cellulose fibers that require a complex mix of decomposition enzymes (Béguin 1990; Martin et al. 1991), and pectins and hemicelluloses that bind the cellulose fibers together into a strong matrix (Selvendran 1984). Furthermore, a fungivore lifestyle for the ants (Davidson 2004) as well as a herbivore lifestyle in general is likely to be nitrogen limited (Mattson 1980; Behmer 2009) which might explain why the ants normally collect highly nutritious plant material (Mundim et al. 2009) and the presence of nitrogen fixing bacteria in the fungus gardens (Pinto-Tomás et al. 2009).

However, the fungus-growing ants with their fungal symbiont seem to have found ways to overcome these complications and become one of the dominating herbivore systems in the Neotropics. This is reflected in the patterns of forage material acquisition that are found throughout the phylogeny of the Attine tribe, with more crude materials at the basal lineages and mainly fresh plant material in the higher attine clades (De Fine Licht and Boomsma 2010). This pattern corresponds with the differences that can be found in enzyme activities of the fungus gardens in each of these

19 groups (De Fine Licht et al. 2010) as well as more specific changes in protease activities for the fungal symbionts (Semenova et al. 2011). Even though these patterns have been found, the leaf-cutting ants acquire forage material from a wide range of plant species which correlate with the availability during the different seasons (Wirth et al. 2003).

It was shown in colonies of Atta cephalotes that the fungus gardens are able to deal with this due to a fast and high rate of plasticity in enzyme activity towards different substrates (Kooij et al. 2010). But because these were laboratory colonies rather than colonies in the field, the question remains, how are acquired plant substrate and fungus garden enzyme activities correlated in field colonies? Furthermore, in the locality of Gamboa, Panama, six species of leaf-cutting ants co-occur, three from each of the two genera. Within each type of habitat one species of each genus can be found: Atta cephalotes and Acromyrmex volcanus as canopy foragers, Atta colombica and Acromyrmex octospinosus as forest edge foragers, and Atta sexdens and Acromyrmex echinatior as open sunlit area foragers. This raises the question; do the leaf-cutting ant genera Atta and Acromyrmex that live in the same area forage for the same material? And, do the fungus gardens in both genera express the same enzyme activity?

Special Acromyrmex echinatior fungal symbiont enzymes

“As to leaf-cutting ants, I have always held the same view as which is proposed by Mr. Belt, viz. that they feed upon the fungus growing on the leaves, they carry into their nests, though I had not yet examined their stomachs. Now I find that the contents of the stomach are colourless, showing under the microscope some minute globules, probably the spores of the fungus. I could find no trace of vegetable tissue which might have been derived from the leaves they gather; and this, I think, confirms Mr. Belt’s hypothesis.” (Müller 1874)

Aside from the general enzyme patterns of the fungus gardens described above, it has been shown that the fungal symbiont alone too has made incredible adaptations to live their mutualistic life with the ants. It was shown that both xylanase and cellulase activities are particularly high in the bottom layers of fungus gardens of Acromyrmex echinatior and Atta cephalotes (Schiøtt et al. 2008; Moller et al. 2011; Aylward et al. 2013; Grell et al. 2013). To show that these activities came from the fungal symbiont ant not from other organisms, a xylanase gene was isolated from the fungus and shown to be functional when expressed in yeast (Schiøtt et al. 2008). Furthermore, fungal genes for cellulases were expressed mostly in the in the bottom layer of the fungus garden (Aylward et al. 2013; Grell et al. 2013).

20 Recently, a total of 33 proteins were identified and sequenced in the fecal fluid of the leaf-cutting ant Acromyrmex echinatior (Schiøtt et al. 2010). The ants mix this fecal fluid as manure with fresh leaf material is chewed to pulp and some tufts of mycelium taken from elsewhere in the fungus garden and put this mixture on top of the fungus garden to ensure new fungal growth (Weber 1966). Of these 33 proteins, seven proved to be pectinases, important in the breakdown of the plant cell wall, of which at least five showed to be active in the fecal fluid of the ants. Further- more, genes for six of these pectinases were upregulated in the gongylidia compared to normal mycelium (Schiøtt et al. 2010). Later, similar patterns were shown for one laccase, important for the breakdown of phenolic compounds, the chemical defense mechanism of plants, present in the fecal fluid, which was both active in the fecal fluid and the gene for this enzyme was upregulated in the gongylidia compared to normal mycelium (De Fine Licht et al. 2013).

Because herbivore systems are nitrogen limited, it is to be expected that proteases are present in the fecal fluid as well, as part of the plant degrading enzyme package. Thus, the raised questions are: Are fungal proteases also found in the fecal fluid? If so, are these proteases active in the fecal fluid? And, are the genes encoding these enzymes upregulated in the gongylidia when compared to normal mycelium? Also, because cellulose is the main component in plant material (Manzoni et al. 2010; Bell et al. 2014), are cellulases present in the fecal fluid and what is the role of these cellulases? And, what is the role of cellulases in the different layers of the fungus gardens?

Summary and future perspectives The influence of domestication on fungal genetics

“Indeed, crop domestication in the context of coevolving and codomesticated microbial consortia may explain the 50-million year old agricultural success of insect farmers.” (Mueller et al. 2005)

I show that the level of domestication in the fungus-growing ants influences the average number of nuclei per fungal cell, with the fungal symbionts of the paleo- and basal agriculture always having two nuclei, the standard for Basidiomycete fungi. However, when domestication intensi- fies in the higher attine ants, the fungal symbionts have between 7-17 nuclei per cell on average. Furthermore, using microsatellite markers, we were able to estimate that these higher attine fungal symbionts have an increased level of ploidy of approximately 5n-6n, meaning that about half the nuclei present in these strains have different genomes. This sudden increase coincides with the arrival of gongylidia as well as the obligate commitment of being a domesticated crop without free-living relatives (Mikheyev et al. 2006; Schultz and Brady 2008; Mikheyev et al. 2010).

21 The increase in number of nuclei and ploidy level seems to be a clever solution to resolve complications of asexuality due to loss of possible gene flow with free-living relatives and the increased lack of viable fruiting bodies, which are, though rarely, generally only seen at colonies of the paleo- and basal attine ants (Mueller et al. 1998; Pagnocca et al. 2001; Mueller 2002). However, it would be interesting to obtain a better estimate of the ploidy level in the fungal symbiont, as well as the nuclear ratio of the different nuclear types, i.e. different genomes, by sequencing the nuclei individually. The challenge for this is to find the right techniques to make this possible. One way could be to create protoplasts of the fungal strains, which is a technique that removes the cell walls of the fungal cells to leave each nucleus separate in a suspension (Sonnenberg et al. 1988). This suspension can then be plated out on a growth medium and subsequently single pro- toplast colonies can be isolated and sequenced. Another techniques is the newly developed laser capture microdissection of single nuclei, which then can be isolated a sequenced using single cell sequencing methods (Guo et al. 2012).

In the fungus-growing wasps Sirex, the fungus is asexually and vertically transmitted as in the fungus-growing ants, but recent results have shown that the fungus maintained the possibility for sexuality (Van der Nest et al. 2013). The diversity in recognition genes in the fungal symbiont of the wasps is relatively high compared to other genes, which indicates that the fungal symbiont has not lost its ability for sexual reproduction. However, the genetic variation in these genes is significantly lower than similar genes in free-living sister species. It would be interesting to see whether the same is true for the fungal symbionts of the ants, because even though it is extremely rare fruiting bodies in colonies of the higher attines can be observed occasionally, though mainly in lab colonies (Fisher et al. 1994a; Fisher et al. 1994b; Pagnocca et al. 2001; Mueller 2002), P.W. Kooij, M. Schiøtt, H.H. de Fine Licht, personal observations).

Finally, polyploidy generally has an influence in genetic aspects such as gene redundancy (Comai 2005), higher gene expression diversity (Groth and Christ 1992) and other stress related advantages (Lidzbarsky et al. 2009). It would therefore obvious to investigate what the phenotypic consequences are of the increase in ploidy level found in this system. One might expect to find increased levels of expression as well as diversity of biodegrading enzymes when comparing transcriptomes of the higher attine ant fungal symbionts with those of the paleo- and basal attine ants, due to a shift in growth substrate from more crude substrates, e.g. insect frass and dry plant material, in the latter group to more complex substrates, i.e. fresh plant material (De Fine Licht and Boomsma 2010).

22 Incompatibility between fungal crops of different ant genera

“Nevertheless, on our functional grounds, the interests of ants and fungi in a colony seem best viewed as organismally merged.” (Queller and Strassmann 2009)

Following up on the incompatibility study on Acromyrmex fungal symbionts by Poulsen & Booms- ma (2005), I present here that the Atta fungal symbionts show similar incompatibility reactions. Even though the strength of the incompatibility reactions is increasing with increasing genetic distance for Atta fungal symbionts, they tend to be lower in general than in Acromyrmex fungal symbionts. Furthermore, this signal is distorted once the fungal symbionts are in confrontations with strains across the ant genus. To achieve a better insight in the differences found between the Atta symbionts and Acromyrmex symbionts, phylogenies based on microsatellite markers and on AFLP were compared. This not only revealed that the symbionts, which are supposed to be the same species (Mikheyev et al. 2010), separate in each of the ant genera, but also revealed that the phylogeny for Acromyrmex symbionts is much more strict and less variable than that of the Atta symbionts.

There must be interesting differences between Atta and Acromyrmex symbionts, which may be due to conditional gene expression as these are essentially the same cultivars. This could be related to the fact that Atta ant species have claustral colony founding, where the queen never leaves the colony and closes of the entrance until the first workers open it to go out foraging 80-100 days later (Weber 1966), securing life time commitment. In contrast, the Acromyrmex queens go out to forage and are therefore most likely to exchange cultivars in this founding phase of the colony (Poulsen et al. 2009). Another possible explanation might lie in the possibility that the Atta symbionts might use imprinting of the ants (Seal et al. 2012), e.g. with chemical cues (Richard et al. 2007), rather than the incompatibility reactions shown in Acromyrmex symbionts (Bot et al. 2001b; Poulsen and Boomsma 2005). A final explanation might lie in the extra AFLP bands that were found in the Atta symbionts compared to those of Acromyrmex, which indicate there might be a presence of RNA/DNA from other organisms such as mycovirusses (Pearson et al. 2009), bacteria (Suen et al. 2010) or prions (Wickner et al. 2007). However, no DNA was amplified with bacteria specific 16S primers, which would exclude the possibility of any bacteria left to explain the extra AFLP bands.

23 In order to get a better understanding of how this incompatibility system works it would be interesting the measure the overall gene expressions in the interaction zones of both compatible as well as incompatible reactions, by sequencing the full transcriptome in these interaction zones. This has proven to be insightful in the fungus-growing wasp mutualism, and showed clear upregula- tions of multiple genes involved in recognition, stress response as well as programmed cell death (Van der Nest et al. 2011). Furthermore, similar analyses might give an insight in the difference in incompatibility reactions between the Atta and Acromyrmex symbionts at the gene level.

Decomposition enzymes in the fungus-growing ant symbiosis

“Virtually every facet of the ants’ behavior and life cycle has been shaped by their association with the fungus they culture.” (Martin 1970)

Foraging behavior in May 2011 by the six sympatric species of leaf-cutting ants in Gamboa, Pan- ama, was different between the two genera, however, it was similar between the species within these genera. The Atta species focus their foraging primarily on tree-leaves, whereas the Acromyr- mex species focused more on collecting flower parts and herbal leaves. This separation in foraging behavior might help explain why the six Panamanian leaf-cutting ant species can be found in strict groups of two species per habitat in this locality (Atta cephalotes and Acromyrmex volcanus as canopy foragers, Atta colombica and Acromyrmex octospinosus as forest edge foragers, and Atta sexdens and Acromyrmex echinatior as open sunlit area foragers), because competition for forage material is higher between species rather than between genera. The question that raises now is whether similar separation between genera can be found in other smaller communities.

Furthermore, the fungus garden enzyme activities in the colonies of Atta and Acromyrmex correlated with the forage material collection by the ants, showing that even though these ant species grow the same species of crop fungus, the expression of enzymes is being adjusted according to the substrate provided. A more striking finding is that the overall enzyme activity in Acromyrmex fungus gardens is higher than that in Atta fungus gardens. Aside from that, Atta fungus gardens seem to have a more specialized activity than Acromyrmex. These difference could be a possible explanation why it seems that Acromyrmex colonies produce less waste material than Atta colonies (Bot et al. 2001a; Hart and Ratnieks 2002), and it would therefore be interesting to investigate whether the amount of generated waste material is the same when the genera are provided with the same forage material.

24 Special Acromyrmex echinatior fungal symbiont enzymes

“…it is not their biochemical machinery which the ants are contributing to this mutualistic association, but rather their capacity to serve as vehicles of transport.” (Boyd and Martin 1975b)

In the first proteomic analysis of the fecal fluid of the leaf-cutting ant Acromyrmex echinatior six proteases and one cellulase were identified and sequenced. The protease fall into three groups of biodegrading enzymes, metalloendoproteases, serine proteases and aspartic proteases, of which the first two groups showed activity in the fecal fluid of the ants. Furthermore, the genes coding for five of these enzymes (peptidyl-Lys metallopeptidase I and II, subtilisin, grifolisin and saccharopepsin) showed to be upregulated in the gongylidia compared to normal mycelium. This shows that the fungal symbiont made a special adaptation to use the ants as vehicles to transport enzymes from the middle part of the fungus garden where there is enough mycelium to produce the gongylidia as well as enzymes (De Fine Licht et al. 2013) to the top of the fungus garden to degrade new substrate. The new techniques used confirm results from earlier studies, though with much more detail (Boyd and Martin 1975a; Boyd and Martin 1975b). Furthermore, it shows that the tightly regulated intake of proteins by insects in this nitrogen limited environment (Behmer 2009), has been taken over by the fungal symbiont, something that is reflected in the fact that the fungus growing ants are the only ants with extremely high protease activities in the rectum rather than in their midgut (Martin 1970).

The single cellulase that was identified and sequenced was identified as an endocellulase, -Lg Cel12, from the Glycoside Hydrolase family 12 (GH12), and has previously not been described (Aylward et al. 2013; Grell et al. 2013). Endocellulases are the first in a series of three cellulases that complete the degradation of cellulose (Béguin 1990; Martin et al. 1991), the main component in plant cell walls (Manzoni et al. 2010; Bell et al. 2014). Even though activity was present for all types of cellulases in the fecal fluid of the ants, this particular enzyme showed to be upregulated in the gongylidia compared to normal mycelium. And other than previously studied genes (Aylward et al. 2013; Grell et al. 2013), this gene was mostly expressed in the middle layer of the fungus garden rather than the bottom layer. This shows that the LgCel12 has an important function in the fecal fluid of the ants, even though the highest cellulase activity can be found in the bottom layer of the fungus garden and the debris material (Moller et al. 2011; Aylward et al. 2013; Grell et al. 2003; Chapter 5 of this thesis).

25 Together with previously studied pectinases (Schiøtt et al. 2010) and laccases (De Fine Licht et al. 2013), now proteases and a cellulase proved to have an important role in the initial stages of plant cell wall degradation in this mutualistic symbiosis, which all together show interesting parallels to the enzyme package used by phytopathogens to attack plants (St Leger et al. 1997). Furthermore, because this system is most likely nitrogen limited due to the herbivore way of living (Mattson 1980; Behmer 2009), it is of interest that the combination of proteases and cellulases is in balance. The proteases are necessary to utilize as much nitrogen as possible, but the cellulases also play an important role in keeping the C:N ratio in balance. When this balance gets disrupted the efficiency of utilizing both carbon and nitrogen decreases (Manzoni et al. 2010; Hartman and Richardson 2013; Sinsabaugh et al. 2013), which in turn would have a detrimental effect on the development of the colony. A reason to keep the cellulase activities particularly high in the bottom layer of the fungus as suggested by Grell et al. (2013) is to maintain the water balance in this section, so to keep the fungus healthy and prevent it from obtaining any diseases before it is discarded by the ants.

These results however, only test the presence and activity of biodegrading enzymes in the leaf-cutting ant Acromyrmex, but earlier studies show increased protease levels in the fecal fluid in ants from the whole phylogeny of fungus-growing ants when compared to other ant species (Martin 1970; Martin and Martin 1971; Martin 1974). An obvious next step would thus be to investigate how these fungal enzymes are regulated in the fecal fluid of the leaf-cutting ant genusAtta , which have even larger colonies (Schultz and Brady 2008; Mehdiabadi and Schultz 2010), in the higher attine genera Trachymyrmex and Sericomyrmex, which feed their fungal crop cruder substrates but where gongylidia are present (De Fine Licht and Boomsma 2010), and in lower attines, which do not have gongylidia, but where the ants too produce fecal fluid (Martin and Martin 1971).

26 References Aylward FO, Burnum-Johnson KE, Tringe SG, et al. (2013) Leucoagaricus gongylophorus produces diverse enzymes for the degradation of recalcitrant plant polymers in leaf-cutter ant fungus gardens. Appl Environ Microb 79:3770–3778. doi: 10.1128/AEM.03833-12 Baer BC, Dijkstra MB, Mueller UG, et al. (2009) Sperm length evolution in the fungus-growing ants. Behav Ecol 20:38–45. doi: 10.1093/beheco/arn112 Bass M, Cherrett JM (1995) Fungal hyphae as a source of nutrients for the leaf-cutting ant Atta sexdens. Physiol Entomol 20:1–6. Behmer ST (2009) Insect herbivore nutrient regulation. Annu Rev Entomol 54:165–187. doi: 10.1146/annurev.ento.54.110807.090537 Bell C, Carrillo Y, Boot CM, et al. (2014) Rhizosphere stoichiometry: are C : N : P ratios of plants, soils, and enzymes conserved at the plant species-level? New Phytol 201:505-517. doi: 10.1111/nph.12531 Belt T (1874) The naturalist in Nicaragua. J. M. Dent & Sons, Ltd, London, UK, pp 290 Béguin P (1990) Molecular biology of cellulose degradation. Annu Rev Microbiol 44:219–248. doi: 10.1146/annurev.mi.44.100190.001251 Boomsma JJ (2013) Beyond promiscuity: mate-choice commitments in social breeding. Philos T Roy Soc B 368:20120050. doi: 10.1126/science.1208473 Bot A, Currie CR, Hart AG, Boomsma JJ (2001a) Waste management in leaf-cutting ants. Ethol Ecol Evol 13:225–237. Bot A, Rehner SA, Boomsma JJ (2001b) Partial incompatibility between ants and symbiotic fungi in two sympatric species of Acromyrmex leaf-cutting ants. Evolution 55:1980–1991. Boyd ND, Martin MM (1975a) Faecal proteinases of the fungus-growing ant, Atta texana: properties, significance and possible origin. Insect Biochem 5:619–635. Boyd ND, Martin MM (1975b) Faecal proteinases of the fungus-growing ant, Atta texana: their fungal origin and ecological significanca. J Insect Physiol 21:1815–1820. Chapela IH, Rehner SA, Schultz TR, Mueller UG (1994) Evolutionary history of the symbiosis between fungus- growing ants and their fungi. Science 266:1691–1694. Cherrett JM, Peregrine D (1976) A review of the status of leaf-cutting ants and their control. Ann Appl Biol 84:124–128. Coley PD, Bryant JP, Chapin FS (1985) Resource availability and plant antiherbivore defense. Science 230:895–899. doi: 10.1126/science.230.4728.895 Comai L (2005) The advantages and disadvantages of being polyploid. Nat Rev Genet 6:836–846. doi: 10.1038/nrg1711 Currie CR, Bot A, Boomsma JJ (2003) Experimental evidence of a tripartite mutualism: bacteria protect ant fungus gardens from specialized parasites. Oikos 101:91–102. Currie CR, Scott JA, Summerbell RC, Malloch D (1999) Fungus-growing ants use antibiotic-producing bacteria to control garden parasites. Nature 398:701–704. Davidson DW (2004) Ecological stoichiometry of ants in a New World rain forest. Oecologia 142:221–231. doi: 10.1007/s00442-004-1722-0 Dentinger BTM, Lodge DJ, Munkacsi AB, et al. (2009) Phylogenetic placement of an unusual coral mushroom challenges the classic hypothesis of strict coevolution in the Apterostigma pilosum group ant-fungus mutualism. Evolution 63:2172–2178. doi: 10.1111/j.1558-5646.2009.00697.x Do Nascimento RR, Schoeters E, Morgan ED, et al. (1996) Chemistry of metapleural gland secretions of three attine ants, Atta sexdens rubropilosa, Atta cephalotes, and Acromyrmex octospinosus (Hymenoptera: Formicidae). J Chem Ecol 22:987–1000. doi: 10.1007/BF02029949

27 Doebeli M, Knowlton N (1998) The evolution of interspecific mutualisms. P Natl Acad Sci USA 95:8676–8680. Douglas AE (1998) Host benefit and the evolution of specialization in symbiosis. Heredity 81:599–603. Douglas AE (2008) Conflict, cheats and the persistence of symbioses. New Phytol 177:849–858. doi: 10.1111/j.1469-8137.2007.02326.x Fernandez-Marin H, Zimmerman JK, Nash DR, et al. (2009) Reduced biological control and enhanced chemical pest management in the evolution of fungus farming in ants. P R Soc B 276:2263–2269. doi: 10.1098/rspb.2002.2044 Fernandez-Marin H, Zimmerman JK, Wcislo WT (2007) Fungus garden platforms improve hygiene during nest establishment in Acromyrmex ants (Hymenoptera, Formicidae, Attini). Insect Soc 54:64–69. doi: 10.1007/s00040-007-0907-z Fernandez-Marin H, Zimmerman JK, Wcislo WT (2004) Ecological traits and evolutionary sequence of nest establishment in fungus-growing ants (Hymenoptera, Formicidae, Attini). Biol J Linn Soc 81:39–48. De Fine Licht HH, Andersen A, Aanen DK (2005) Termitomyces sp. associated with the termite Macrotermes natalensis has a heterothallic mating system and multinucleate cells. Mycol Res 109:314–318. doi: 10.1017/S0953756204001844 De Fine Licht HH, Boomsma JJ (2010) Forage collection, substrate preparation, and diet composition in fungus- growing ants. Ecol Entomol 35:259–269. doi: 10.1111/j.1365-2311.2010.01193.x De Fine Licht HH, Schiøtt M, Mueller UG, Boomsma JJ (2010) Evolutionary transitions in enzyme activity of ant fungus gardens. Evolution 64:2055–2069. doi: 10.1111/j.1558-5646.2010.00948.x De Fine Licht HH, Schiøtt M, Rogowska-Wrzesinska A, et al. (2013) Laccase detoxification mediates the nutritional alliance between leaf-cutting ants and fungus-garden symbionts. P Natl Acad Sci USA 110:583–587. doi: 10.1073/pnas.1212709110 Fisher PJ, Stradling DJ, Pegler DN (1994a) Leaf cutting ants, their fungus gardens and the formation of basidiomata of Leucoagaricus gongylophorus. Mycologist 8:128–131. doi: 10.1016/S0269-915X(09)80159-6 Fisher PJ, Stradling DJ, Pegler DN (1994b) Leucoagaricus basidiomata from a live nest of the leaf-cutting ant Atta cephalotes. Mycol Res 98:884–888. doi: 10.1016/S0953-7562(09)80259-1 Foster KR, Wenseleers T (2006) A general model for the evolution of mutualisms. J Evolution Biol 19:1283–1293. doi: 10.1111/j.1420-9101.2005.01073.x Fowler HG, Pagani MI, Da Silva OA, et al. (1989) A pest is a pest is a pest? The dilemma of Neotropical leaf-cutting ants: keystone taxa of natural ecosystems. Environ Manage 13:671–675. Frank SA (1994) Genetics of mutualism: the evolution of altruism between species. J Theor Biol 170:393–400. doi: 10.1006/jtbi.1994.1200 Frank SA (1996a) Host control of symbiont transmission: The separation of symbionts into germ and soma. Am Nat 148:1113–1124. Frank SA (1996b) Host-symbiont conflict over the mixing of symbiotic lineages. P Roy Soc Lond B Bio 263:339–344. Green AM, Mueller UG, Adams RMM (2002) Extensive exchange of fungal cultivars between sympatric species of fungus-growing ants. Mol Ecol 11:191–195. doi: 10.1046/j.1365-294X.2002.01433.x Grell MN, Nielsen KL, Linde T, et al. (2013) The fungal symbiont of leaf-cutter ants expresses the full enzyme complement to degrade cellulose and other plant polymers. BMC Genomics 14:928. doi: 10.1186/1471-2164-14-928

28 Groth JV, Christ BJ (1992) Diversity in plant pathogenic fungi as influenced by host characteristics. In: Fritz RS, Simms EL (eds) Plant Resistance to Herbivores and Pathogens: Ecology, Evolution, and Genetics. The University of Chicago Press, Chicago, USA, pp 172–192 Guo Y, Yang Y, Zhou J, et al. (2012) Microdissection of spatially identified single nuclei in a solid tumor for single cell whole genome sequencing. Biotechniques 0:1–3. doi: 10.2144/000113860 Hart AG, Ratnieks FLW (2002) Waste management in the leaf-cutting ant Atta colombica. Behav Ecol 13:224–231. Hartman WH, Richardson CJ (2013) Differential nutrient limitation of soil microbial biomass and metabolic quotients (qCO2): Is there a biological stoichiometry of soil microbes? PLoS ONE 8:e57127. doi: 10.1371/journal.pone.0057127.s024 Herre EA, Knowlton N, Mueller UG, Rehner SA (1999) The evolution of mutualisms: exploring the paths between conflict and cooperation. Trends Ecol Evol 14:49–53. Hintz W, Anderson JB, Horgen PA (1988) Nuclear migration and mitochondrial inheritance in the mushroom Agaricus bitorquis. Genetics 119:35–41. Howe HF (1984) Constraints on the evolution of mutualisms. Am Nat 123:764–777. Hughes WOH, Pagliarini R, Madsen HB, et al. (2008) Antimicrobial defense shows an abrupt evolutionary transition in the fungus-growing ants. Evolution 62:1252–1257. doi: 10.1111/j.1558-5646.2008.00347.x Ivens ABF, Nash DR, Poulsen M, Boomsma JJ (2009) Caste-specific symbiont policing by workers of Acromyrmex fungus-growing ants. Behav Ecol 20:378-384. doi:10.1093/beheco/arn150 Janzen DH (1985) The natural history of mutualisms. In: Bucher DH (ed) The biology of mutualism. Oxford University Press, New York, USA, pp 40–99 Johnson J, Vilgalys R (1998) Phylogenetic systematics of Lepiota sensu lato based on nuclear large subunit rDNA evidence. Mycologia 90:971–979. Keeler KH (1985) Cost:benefit models of mutualism. In: Bucher DH (ed) The biology of mutualism. Oxford University Press, New York, USA, pp 100–127 Kellner K, Fernandez-Marin H, Ishak HD, et al. (2013) Co-evolutionary patterns and diversification of ant-fungus associations in the asexual fungus-farming ant Mycocepurus smithii in Panama. J Evolution Biol 26:1353– 1362. doi: 10.1111/jeb.12140 Kooij PW, Schiøtt M, Boomsma JJ, De Fine Licht HH (2010) Rapid shifts in Atta cephalotes fungus-garden enzyme activity after a change in fungal substrate (Attini, Formicidae). Insect Soc 58:145–151. doi: 10.1007/s00040-010-0127-9 Law R (1985) Evolution in a mutualistic environment. In: Bucher DH (ed) The biology of mutualism. Oxford University Press, New York, pp 145–170 Law R, Lewis DH (1983) Biotic environments and the maintenance of sex - some evidence from mutualistic symbioses. Biol J Linn Soc 20:249–276. Leigh EG (2010) The evolution of mutualism. J Evolution Biol 23:2507–2528. doi: 10.1111/j.1420-9101.2010.02114.x Lidzbarsky GA, Shkolnik T, Nevo E (2009) Adaptive response to DNA-damaging agents in natural Saccharomyces cerevisiae populations from “Evolution Canyon,” Mt. Carmel, Israel. PLoS ONE 4:e5914. doi: 10.1371/journal.pone.0005914 Manzoni S, Trofymow JA, Jackson RB, Porporato A (2010) Stoichiometric controls on carbon, nitrogen, and phosphorus dynamics in decomposing litter. Ecol Monogr 80:89–106. Martin MM (1970) The biochemical basis of the fungus-attine ant symbiosis. Science 169:16–20. Martin MM (1974) Biochemical ecology of the Attine ants. Accounts Chem Res 7:1–5.

29 Martin MM, Jones CG, Bernays EA (1991) The evolution of cellulose digestion in insects [and discussion]. Philos T Roy Soc B 333:281–288. doi: 10.1098/rstb.1991.0078 Martin MM, Martin JS (1971) The presence of protease activity in the rectal fluid of primitive attine ants. J Insect Physiol 17:1897–1906. Mattson WJ Jr (1980) Herbivory in relation to plant nitrogen content. Annu Rev Ecol Syst 11:119–161. May G (1988) Somatic incompatibility and individualism in the coprophilous Basidiomycete, Coprinus cinereus. T Brit Mycol Soc 91:443–451. doi: 10.1016/S0007-1536(88)80121-9 McCutcheon JP, Moran NA (2012) Extreme genome reduction in symbiotic bacteria. Nat Rev Microbiol 10:13–26. doi: 10.1038/nrmicro2670 Mehdiabadi NJ, Mueller UG, Brady SG, et al. (2012) Symbiont fidelity and the origin of species in fungus-growing ants. Nat Commun 3:840. doi: 10.1038/ncomms1844 Mehdiabadi NJ, Schultz TR (2010) Natural history and phylogeny of the fungus-farming ants (Hymenoptera: Formicidae: Myrmicinae: Attini). Myrmecol 13:37–55. Mikheyev AS, Mueller UG, Abbot P (2006) Cryptic sex and many-to-one coevolution in the fungus-growing ant symbiosis. P Natl Acad Sci USA 103:10702–10706. doi: 10.1073/pnas.0601441103 Mikheyev AS, Mueller UG, Abbot P (2010) Comparative dating of attine ant and lepiotaceous cultivar phylogenies reveals coevolutionary synchrony and discord. Am Nat 175:E126–33. doi: 10.1086/652472 Mikheyev AS, Mueller UG, Boomsma JJ (2007) Population genetic signatures of diffuse co-evolution between leaf- cutting ants and their cultivar fungi. Mol Ecol 16:209–216. doi: 10.1111/j.1365-294X.2006.03134.x Moller IE, De Fine Licht HH, Harholt J, et al. (2011) The dynamics of plant cell-wall polysaccharide decomposition in leaf-cutting ant fungus gardens. PLoS ONE 6:e17506. doi: 10.1371/journal.pone.0017506 Möller A (1893) Die Pilzgärten einiger südamerikanischer Ameisen. Verlag von Gustav Fischer, Jena, Germany, pp 127. Mueller UG (2002) Ant versus fungus versus mutualism: Ant-cultivar conflict and the deconstruction of the attine ant-fungus symbiosis. Am Nat 160:S67–S98. Mueller UG, Gerardo NM, Aanen DK, et al. (2005) The evolution of agriculture in insects. Annu Rev Ecol Evol S 36:563–595. doi: 10.1146/annurev.ecolsys.36.102003.152626 Mueller UG, Rehner SA, Schultz TR (1998) The evolution of agriculture in ants. Science 281:2034–2038. Mueller UG, Scott JJ, Ishak HD, et al. (2010) Monoculture of leafcutter ant gardens. PLoS ONE 5:e12668. doi: 10.1371/journal.pone.0012668 Müller F (1874) The habits of various insects. Nature 10:102–103. Mundim FM, Costa AN, De Vasconcelos HL (2009) Leaf nutrient content and host plant selection by leaf-cutter ants, Atta laevigata, in a Neotropical savanna. Entomol Exp Appl 130:47–54. doi: 10.1111/j.1570-7458.2008.00789.x Munkacsi AB, Pan J, Villesen P, et al. (2004) Convergent coevolution in the domestication of coral mushrooms by fungus-growing ants. P Roy Soc Lond B Bio 271:1777–1782. doi: 10.1098/rspb.2004.2759 Pagnocca FC, Bacci M Jr, Fungaro MH, et al. (2001) RAPD analysis of the sexual state and sterile mycelium of the fungus cultivated by the leaf-cutting ant Acromyrmex hispidus fallax. Mycol Res 105:173–176. Pearson MN, Beever RE, Boine B, Arthur K (2009) Mycoviruses of filamentous fungi and their relevance to plant pathology. Mol Plant Pathol 10:115–128. doi: 10.1111/j.1364-3703.2008.00503.x Pinto-Tomás AA, Anderson MA, Suen G, et al. (2009) Symbiotic nitrogen fixation in the fungus gardens of leaf-cutter ants. Science 326:1120–1123. doi: 10.1126/science.1173036 Poulsen M, Boomsma JJ (2005) Mutualistic fungi control crop diversity in fungus-growing ants. Science 307:741–744. doi: 10.1126/science.1106688

30 Poulsen M, Currie CR (2009) Symbiont interactions in a tripartite mutualism: exploring the presence and impact of antagonism between two fungus-growing ant mutualists. PLoS ONE 5:e8748–e8748. doi: 10.1371/journal.pone.0008748 Poulsen M, Fernandez-Marin H, Currie CR, Boomsma JJ (2009) Ephemeral windows of opportunity for horizontal transmission of fungal symbionts in leaf-cutting ants. Evolution 63:2235–2247. doi: 10.1111/j.1558-5646.2009.00704.x Queller DC, Strassmann JE (2009) Beyond society: the evolution of organismality. Philos T Roy Soc B 364:3143–3155. doi: 10.1098/rstb.2009.0095 Quinlan R, Cherrett JM (1979) The role of fungus in the diet of the leaf-cutting ant Atta cephalotes (L.). Ecol Entomol 4:151–160. Quinlan R, Cherrett JM (1978) Aspects of the symbiosis of the leaf-cutting ant Acromyrmex octospinosus (Reich) and its food fungus. Ecol Entomol 3:221–230. Raper CA (1976) Sexuality and life-cycle of the edible, wild Agaricus bitorquis. J Gen Microbiol 95:54–66. Rayner ADM (1991) The challenge of the individualistic mycelium. Mycologia 83:48–71. Rayner ADM, Coates D, Ainsworth AM, et al. (1984) The biological consequences of the individualistic mycelium. In: Jennings DH, Rayner ADM (eds) The ecology and physiology of the fungal mycelium. Cambridge University Press, Cambridge, UK pp 509–540 Rayner ADM, Franks NR (2003) Evolutionary and ecologica parallels between ants and fungi. Trends Ecol Evol 2:127– 133. Richard F-J, Poulsen M, Hefetz A, et al. (2007) The origin of the chemical profiles of fungal symbionts and their significance for nestmate recognition in Acromyrmex leaf-cutting ants. Behav Ecol Sociobiol 61:1637–1649. doi: 10.1007/s00265-007-0395-1 Sachs JL, Mueller UG, Wilcox TP, Bull JJ (2004) The evolution of cooperation. Q Rev Biol 79:135–160. doi: 10.1086/383541 Saksena KN, Marino R, Haller MN, Lemke PA (1976) Study on development of Agaricus bisporus by fluorescent microscopy and scanning electron microscopy. J Bacteriol 126:417–428. Schiøtt M, De Fine Licht HH, Lange L, Boomsma JJ (2008) Towards a molecular understanding of symbiont function: Identification of a fungal gene for the degradation of xylan in the fungus gardens of leaf-cutting ants. BMC Microbiol 8:40. doi: 10.1186/1471-2180-8-40 Schiøtt M, Rogowska-Wrzesinska A, Roepstorff P, Boomsma JJ (2010) Leaf-cutting ant fungi produce cell wall degrading pectinase complexes reminiscent of phytopathogenic fungi. BMC Biol 8:156. doi: 10.1186/1741- 7007-8-156 Schultz TR, Brady SG (2008) Major evolutionary transitions in ant agriculture. P Natl Acad Sci USA 105:5435–5440. doi: 10.1073/pnas.0711024105 Scott JJ, Weskin MK, Cooper M, Mueller UG (2009) Polymorphic microsatellite markers for the symbiotic fungi cultivated by leaf cutter ants (Attini, Formicidae). Mol Ecol Resour 9:1391–1394. doi: 10.1111/j.1755-0998.2009.02684.x Seal JN, Gus J, Mueller UG (2012) Fungus-gardening ants prefer native fungal species: do ants control their crops? Behav Ecol 23:1250–1256. doi: 10.1093/beheco/ars109 Selvendran RR (1984) The plant cell wall as a source of dietary fiber: chemistry and structure. Am J Clin Nutr 39:320– 337.

31 Semenova TA, Hughes DP, Boomsma JJ, Schiøtt M (2011) Evolutionary patterns of proteinase activity in attine ant fungus gardens. BMC Microbiol 11:15. doi: 10.1186/1471-2180-11-15 Sinsabaugh RL, Manzoni S, Moorhead DL, Richter A (2013) Carbon use efficiency of microbial communities: stoichiometry, methodology and modelling. Ecol Lett 16:930–939. doi: 10.1111/ele.12113 Sonnenberg AS, Wessels JG, Van Griensven LJ (1988) An efficient protoplasting/regeneration system for Agaricus bisporus and Agaricus bitorquis. Curr Microbiol 17:285–291. St Leger RJ, Joshi L, Roberts DW (1997) Adaptation of proteases and carbohydrases of saprophytic, phytopathogenic and entomopathogenic fungi to the requirements of their ecological niches. Microbiol-Sgm 143:1983–1992. Suen G, Scott JJ, Aylward FO, et al. (2010) An insect herbivore microbiome with high plant biomass-degrading capacity. PLoS Genet. doi: 10.1371/journal.pgen.1001129 Van der Nest MA, Steenkamp ET, Slippers B, et al. (2011) Gene expression associated with vegetative incompatibility in Amylostereum areolatum. Fungal Genet Biol 48:1034–1043. doi: 10.1016/j.fgb.2011.08.001 Van der Nest MA, Steenkamp ET, Wilken MP, et al. (2013) Mutualism and asexual reproduction influence recognition genes in a fungal symbiont. Fungal Biol 117:439–450. doi: 10.1016/j.funbio.2013.05.001 Villesen P, Mueller UG, Schultz TR, et al. (2004) Evolution of ant-cultivar specialization and cultivar switching in Apterostigma fungus-growing ants. Evolution 58:2252–2265. doi: 10.1111/j.0014-3820.2004.tb01601.x Villesen P, Murakami T, Schultz TR, Boomsma JJ (2002) Identifying the transition between single and multiple mating of queens in fungus-growing ants. P R Soc B 269:1541–1548. doi: 10.1098/rspb.2002.2044 Vo TL, Mueller UG, Mikheyev AS (2009) Free-living fungal symbionts (Lepiotaceae) of fungus-growing ants (Attini: Formicidae). Mycologia 101:206–210. doi: 10.3852/07-055 Weber NA (1966) Fungus-growing ants. Science 153:587–604. Weber NA (1972) The fungus-culturing behavior of ants. Am Zool 12:577–587. Weber NA (1955) Pure cultures of fungi produced by ants. Science 121:109–109. Wheeler WM (1907) The fungus-growing ants of North America. B Am Mus Nat Hist 23:669–U130. Wheeler WM (1910) Ants: their structure, development, and behavior. Columbia University Press, New York, USA pp 663. Wickner RB, Edskes HK, Shewmaker F, Nakayashiki T (2007) Prions of fungi: inherited structures and biological roles. Nat Rev Microbiol 5:611–618. doi: 10.1038/nrmicro1708 Wirth R, Herz H, Ryel RJ, et al. (2003) Herbivory of leaf-cutting ants: a case study on Atta colombica in the tropical rainforest of Panama. Springer-Verlag, Berlin, Germany pp 230. Worrall JJ (1997) Somatic incompatibility in basidiomycetes. Mycologia 89:24–36. Yek SH, Nash DR, Jensen AB, Boomsma JJ (2012) Regulation and specificity of antifungal metapleural gland secretion in leaf-cutting ants. P Roy Soc Lond B Bio 279:4215–4222. doi: 10.1098/rspb.2012.1458

32 Advanced farming ants rear polyploid crop fungi

Pepijn W Kooij, Duur K Aanen, Morten Schiøtt, Jacobus J Boomsma

(Manuscript) Chapter 1 33 Advanced farming ants rear polyploid crop fungi

Pepijn W. Kooij1, Duur K. Aanen2, Morten Schiøtt1 and Jacobus J. Boomsma1

1Centre for Social Evolution, Department of Biology, University of Copenhagen, Universi- tetsparken 15, DK-2100 Copenhagen, Denmark

2Laboratory of Genetics, Wageningen University, PO Box 309, 6700 AH, Wageningen, The Neth- erlands

Phone: +45 35321239 Fax: +45 35321250

34 Abstract Polyploidy is common in plants, presumably because advantages of increased functional heterozygosity often surpass costs of destabilized mitosis or epigenetic instability, but rare in fungi, where ‘diploid’ cells retain the two parental haploid nuclei rather than merging them into a zygote. Such dikaryotic mycelia are actively maintained in Basidiomycota where clamp connections ensure honest propagation of nuclei during cell division, but the same mechanism may constrain the evolution of multinucleate cells. Previous research has indicated that the domesticated basidiomycete crop-fungus of leaf-cutting ants, which lacks clamp connections, is a functional polyploid, but without pursuing this further. Here, we used microscopy to estimate the mean number of nuclei per somatic cell for 42 fungal symbionts reared by 14 species (eight genera) of fungus-growing ants in Panama, and mapped these numbers on a fungal symbiont phylogeny generated by ITS and LSU sequencing. This showed that all higher attine ants that rear specialized fungi without free-living close relatives had 7-17 nuclei per cell, whereas non-specialized fungal crops of the basal paleo- and lower attines were dikaryons. Analysis with ten microsatellite markers revealed that ca. 40% of the additional nuclei represented novel haplotypes, yielding an estimate of average ploidy in higher attine crop fungi of 5-6. The fungal symbionts reared by Atta and Acromyrmex leaf-cutting ants were distinct from the clade reared by other higher attine ants, but we found no differences in nuclei number between these lineages. We hypothesize that functional polyploidy in these asexual and vertically transmitted crop symbionts evolved because benefits of increased heterozygosity outweighed costs that could be met by the farming ants. Polyploid crops may thus represent a form of symbiont chimerism for the sake of enhanced genetic diversity, similar to multiple queen-mating that evolved in the leaf-cutting ants to generate higher heterozygosity among colony workers. The transition to crop-polyploidy coincided with at least an order of magnitude increase in colony size of farming ants.

Keywords Attini, mutualism, co-evolution, multi-nucleate, Leucoagaricus, Basidiomycota, fungus-growing ants

35 Introduction Polyploidy, the possession of more than two sets of haploid chromosomes, is widespread in flow- ering plants (Masterson 1994; Otto and Whitton 2000), with 15% of speciation events having likely been facilitated by polyploidization, a figure that is even twice as high (31%) in ferns (Wood et al. 2009). Ploidy levels can be as high as 10x in Fragaria (Hummer et al. 2009) and 12x in Rhodondendron (Jones et al. 2007), but have almost always remained even. Polyploidy may be advantageous because it increases functional heterozygosity, which helps masking the phenotypic effects of deleterious recessives, but this may come at the cost of more likely chromosome loss during mitosis, spindle irregularities and other imbalances during meiosis, and more epigenetic instability (Comai 2005). In contrast, polyploidy is rare in metazoans (Wertheim et al. 2013), suggesting that the benefits of polyploidy outweigh the costs more consistently in plants, which retained modular growth in spite of sharing zygote formation and embryonic development with animals (De Mendoza et al. 2013).

Polyploidy is also rare in fungi, but this may be primarily related to fungal cells not being haploid or diploid in the normal sense, but having one (monokaryon) or two (dikaryon) haploid nuclei per cell, with just a few exceptions such as the unicellular Saccharomyces cerevisiae and the multicellular Armillaria mellea (honey fungus) which occasionally have merged diploid nuclei (Rizzo and May 1994; Carvalho et al. 1995). Monokaryon hyphae emerging from germinating spores produced after meiosis are thus analogous to gametophytes in vascular plants, so the sexual process remains uncompleted until two monokaryons merge to form a dikaryon. When such contact is initiated and proves to be compatible, the nuclei of the parental monokaryons distribute themselves evenly throughout the hyphal system so all cells end up having both types of nuclei that will henceforth be passed jointly when cells divide. This makes dikaryons functionally analogous to diploid vascular plant sporophytes (Raper 1955; Ellingboe and Raper 1962).

As the original haploid nuclei of sexual spores do not become life-time committed in a zygote, their association needs to be actively reinforced with every new cell division of a dikaryon to ensure that nuclei retain equal probabilities of taking part in sexual reproduction wherever mycelia will initiate the formation of sexual organs. The Basidiomycota that occur mostly as dikaryons have resolved this challenge by evolving clamp connections, specialized bridges between the new cell, which has received a copy of one nucleus, and the old cell that is about to become separated by a new cell wall (Buss 1987). This septum forms simultaneously in the clamp connection where it separates the two copies of the second nucleus, so one of these is predictably returned to the old cell (Moore et al. 2011). Once a basidiomycete dikaryon has formed, incompatibility mechanisms

36 will normally prevent that additional nuclei from another monokaryon or dikaryon can be accepted unless they are genetically identical or very closely related (Adaskaveg and Gilbertson 1987; Wilson 1991; Marçais et al. 2000; Poulsen and Boomsma 2005).

Multinucleate hyphal cells have only rarely been observed in Basidiomycota. In some cases such as Agaricus bisporus there may be up to 25 nuclei per cell, but these are all clonal copies of the two original nuclei (Saksena et al. 1976) and thus an example of autopolyploidy. However, rare examples of allopolyploidy are documented for Heterobasidion species that have three (James et al. 2009) or four types of nuclei in a single mycelium (Johannesson and Stenlid 2004). These studies used neutral microsatellites markers to estimate the number of haplotypes in excess of the standard two and achieved high resolution, but there is only indirect evidence that the usual explanation of increased heterozygosity might apply. The particularly demanding necrotrophic life history of Heterobasidion phytopathogens, relative to saprotrophic fungi, might offer unusual fitness rewards to the higher gene expression diversity that allopolyploidy would likely allow (Groth and Christ 1992), but stress-related advantages of this kind have only been documented for polyploid yeasts (Lidzbarsky et al. 2009).

In a recent genome sequencing project of the basidiomycete fungal symbiont of the leaf-cutting ant Acromyrmex echinatior, we found strong indications for functional polyploidy (De Fine Licht et al. 2013). This was consistent with earlier findings (Scott et al. 2009) and effectively precluded that a draft genome could be assembled. This fungus, Leucoagaricus gongylophorous, is a specialized domesticated crop organism that the ants rear as their almost exclusive food source in special underground gardens (Mueller et al. 2005). The evolutionary history of fungus farming goes back ca. 50 million years when the ancestor of all extant attine ants gave up its hunter-gatherer life style to start farming vertically transmitted leucocoprinaceous fungi for food (Schultz and Brady 2008). However, it took 30 million years until a single lineage of these crop fungi was irreversibly domesticated, and evolved specialized hyphal tips (gongylidia) to feed the ants (Quinlan and Cherrett 1978; Quinlan and Cherrett 1979; Bass and Cherrett 1995) while losing genetic exchange with free-living relatives (Mueller et al. 2005; Schultz and Brady 2008; De Fine Licht et al. 2010). This gave rise to four genera of so called ‘higher’ attine ants that all rear cultivars belonging to this clade of gongylidia-bearing crop fungi, of which the most advanced Atta and Acromyrmex leaf-cutting ants arose only 10 million years ago and rear a single, genetically polymorphic species of crop fungus (Mikheyev et al. 2010).

Given the rarity of basidiomycete polyploidy, and the fact that the human domesticated mushroom

37 Agaricus bisporus has multinucleate cells while its free-living close relative Agaricus bitorquis has normal dikaryons (Raper 1976; Hintz et al. 1988), we decided to test the hypothesis that domestication might have induced functional polyploidy, to compensate for the absence or extreme rarity of sexual reproduction (Pagnocca et al. 2001; Mueller 2002; Mikheyev et al. 2006) and possibly also to meet novel requirements of enzyme production (Schiøtt et al. 2010; De Fine Licht et al. 2013). We therefore set out to estimate the average number of nuclei per cell across the fungal symbionts of 14 species of attine ants from Panama and used internal transcribed spacer (ITS) and large subunit rRNA (LSU) ribosomal RNA sequencing and microsatellite genotyping to map nuclei-numbers on the symbiont tree and to estimate functional ploidy.

Materials and methods Biological material Fungal symbionts of attine ants were isolated on Potato Dextrose Agar (PDA) from 42 lab-colonies collected in Panama from 2004 to 2012. Samples included representatives of the different stages of ant fungus farming (see Table S1, for details on species and laboratory nest numbers), such that we had 25 samples of higher attine ants rearing domesticated gongylidia- bearing fungi, and 17 samples of lower- and paleoattine ants that rear fungi without obvious adaptations to being crops (Mueller et al. 1998; Schultz and Brady 2008). Plates were maintained at ca. 25°C, and DNA was obtained with 5% Chelex extractions when sufficient biomass had been produced.

Nuclei counting Small pieces of mycelium from each colony were stained with DAPI on a microscope slide, after which we counted the number of nuclei per cell in five randomly chosen cells under a fluores- cence microscope. For analysis the data were grouped in five categories of fungus-farming that are generally recognized as characterizing the diversification and progression of attine fungus farming: coral fungus agriculture, yeast agriculture, lower agriculture, domesticated agriculture, and leaf-cutting agriculture (Schultz and Brady 2008; Mehdiabadi and Schultz 2010). Differences in number of nuclei per cell were subsequently analyzed with a General Linear Model in SAS with colonies as random factor nested within species, species nested within genera, and genera nested within the five categories of fungus farming.

Internal transcribed spacer (ITS) and large subunit rRNA (LSU) sequencing To make sure that the plated fungi were the same strains as the ones maintained by the fungus growing ants and to verify whether the two known clades of lower agriculture crops (Clade 1 and Clade 2)(Mueller et al. 1998; Schultz and Brady 2008; Mehdiabadi and Schultz 2010) were

38 both represented, we amplified and sequenced two conserved regions: Internal transcribed spacer (ITS4: 5’-TCC TCC GCT TAT TGA TAT GC-3’; ITS5: 5’-GGA AGT AAA AGT CGT AAC AAG G-3’) and the nuclear large subunit rRNA (LR0R: 5’-ACC CGC TGA ACT TAA GC-3’; LR5: 5’- TCC TGA GGG AAA CTT CG-3’), using PCR with 8µL dNTP’s, 2µL gold buffer, 2µL 10mM MgCl2, 0.1µL BSA, 1µL forward and reverse primer each, 0.2µL gold Taq polymerase, 4.7µL

ddH2O and 1µL DNA. The PCR program had 10 min denaturing at 94°C followed by 35 cycles of 1 min denaturing at 94°C, 30 sec annealing at 54°C and 1 min extension at 72°C, and finally a 10 min extension at 72°C. All PCR products were sequenced at BGI Europe in Copenhagen and sequences were deposited in GenBank (see Table S1 for accession numbers). In cases of bad quality sequences due to multiple divergent copies with small length mutations, PCR products were cloned in pCR4-TOPO before sequencing them again using the TOPO TA cloning method (Invitrogen, Carlsbad, CA, USA). A phylogenetic tree similar to Mueller (1998) was constructed using Maximum Likelihood calculations with 500 bootstrap replicates, using MEGA5.1 software (Tamura et al. 2011).

Microsatellite screening As we were finding multinucleate cells throughout the higher attine ant symbionts, we decided to screen these fungi for allelic variation at ten microsatellite loci: A128, A1030, A1132, A1151, B12, B447, C101, C126, C647 and D115 taken from Scott (2009), using PCR with 5µL VWR Red Taq DNA polymerase Master Mix (VWR International, Haasrode, Belgium), 0.25µL forward and

reverse primer each, 1.5µL ddH2O and 1µL DNA, and a program of 5 min denaturing at 95°C, followed by 14 cycles of 30 sec denaturing at 95°C, 30 sec annealing at 68-58°C with a touchdown of -0.5°C per cycle, and 30 sec extension at 72°C followed by 20 cycles of 30 sec denaturing at 95°C, 30 sec annealing at 58°C and 30 sec extension at 72°C, and finally a 15 min extension at 72°C. 1µL of each PCR product was then mixed with 8.75µL formaldehyde and 0.25µL LIZ500 and analyzed on an ABI3130xl (Applied Biosystems, Nærum, Denmark) sequencer. Chromato- grams were analyzed with Genemapper 4.0 (Applied Biosystems, Nærum, Denmark) to obtain

specific allele scorings, after which a phylogenetic tree was constructed using pairwise Fst values, followed by Neighbor Joining analysis with 500 bootstrap replicates using Populations 1.2.32 software (Langella 2001).

Estimating ploidy The microsatellite screenings often showed more than two allelic peaks, indicating that excess nuclei were not all copies of dikaryon nuclei, but at least partly represented allopolyploidy. If we would have had infinite variation at a single microsatellite marker, the number of alleles observed

39 C E D F Atta sexdens Atta Sericomyrmex amabilis Sericomyrmex Cyphomyrmex costatus Cyphomyrmex Apterostigma dentigerum Apterostigma 0 B 5 uclei per cell of n r 10 Average numbe 15 20 Trachymyrmex cf. zeteki Trachymyrmex Trachymyrmex cf. zeteki Trachymyrmex Trachymyrmex cf. zeteki Trachymyrmex Sericomyrmex amabilis Mycocepurus smithii Cyphomyrmex longiscapus Myrmicocrypta ednella Myrmicocrypta ednella Cyphomyrmex longiscapus Apterostigma dentigerum Apterostigma dentigerum Cyphomyrmex rimosus Cyphomyrmex rimosus Cyphomyrmex rimosus Trachymyrmex zeteki Trachymyrmex Sericomyrmex amabilis Trachymyrmex zeteki Trachymyrmex Trachymyrmex cornetzi Trachymyrmex Cyphomyrmex costatus a Cyphomyrmex costatus t t a co r Atta colombica Atta colombica Atta cephalotes Atta cephalotes Atta colombica Atta sexdens Atta sexdens Mycocepurus smithii Myrmicocrypta ednella Cyphomyrmex costatus Apterostigma dentigerum Sericomyrmex amabilis Cyphomyrmex longiscapus Trachymyrmex cornetzi Trachymyrmex Trachymyrmex zeteki Trachymyrmex Acromyrmex octospinosus Acromyrmex echinatior Acromyrmex echinatior Atta cephalotes Acromyrmex octospinosus Acromyrmex octospinosus Acromyrmex echinatior 01 4 120523-01 Tz e 120520-02Cri m 120521-04Cri m 120520-03Cri m 120529-05 Tz e S S S S S Ace-19-B B Ace2 3 Tzet02 8 MS120523-05 T M JOA120525-03Sa m Cylo00 5 JOA120605-02Cyl o Agaricus bisporus Ao51 2 Ae45 1 Ac-2009-4 6 Ac-2011- 3 As-13-B B M M M JOA120605-08Cyl o AI 150511-04C yc o Cyco00 6 100610-02Cyc o Ao51 0 Ao61 6 Ac-2006-2 7 As-2011- 4 Ace-16-B B Ae36 0 Ae37 2 Tcor00 2 M RMMA100611-11BigRe d RMMA100629-17BigRe d Sa m AI110511-05Sa m RMMA100629-16BigRe d MS120523-08Myc S MycS01 5 MS120529-12Med n MyrE002 MS120529-11Med n MS120519-03Aden MS120524-03Aden Aden003 6 5 8 4 0 9 4 5 5 3 2 2 10 0 10 0 10 0 10 0 3 9 4 4 8 3 9 4 7 2 1 10 0 7 2 6 3 8 1 9 2 9 9 3 9 0 5 1 0 5 2 10 0 2 4 6 1 10 0 8 8 8 10 0 10 0 8 8 Outgroup fungus agriculture Coral agriculture Lower agriculture Yeast agriculture Domesticated Leaf-cutting agriculture 10 0 1 6 A

40 would have given us exact estimates of ploidy. However, the limited cumulative detection power of our 10 moderately variable marker loci necessitated that we estimated the detection error of not observing genetically different haplotypes (because alleles are equal by chance) before estimating the average rate at which haplotype numbers increase with the number of nuclei per cell. To do this, we first plotted the number of different alleles observed against the mean number of nuclei for each of the ten markers and calculated the slopes of regression lines forced through the origin. As our dikaryon data for lower attine fungi showed that cells were sometimes counted after having initiated division, we use that quantified error rate of each nucleus being counted as 1.03 to correct scores along the x-axis (Fig. S2). Because ‘the origin’ was the dikaryon expectation of two nuclei giving an expected allele number score of 1+h, where h is the expected heterozygosity at that particular locus, we subtracted these values from the respective x and y scores before running the analyses. Samples with null alleles were excluded, since they provide incomplete information about heterozygosity. As expected the positive slopes of these regression lines increased with locus-specific heterozygosity for the eight most variable marker loci, so we next plotted these slopes against the locus-specific expected heterozygosity (based on observed allele frequencies). As the slopes entered could only vary between zero and one, we fitted a logistic regression through these data points, while weighting them according to their standard error, to obtain an overall estimation of the number of haplotypes we should have detected if we would have had a marker locus with infinite allelic variation.

Figure 1 Ancestry, diversification and mean number of nuclei per cell in garden symbionts of 14 Panamanian fungus-growing ants. A) Phylogenetic tree of attine ant fungal symbionts based on ITS and LSU sequences and Maximum Likelihood analysis with 500 bootstrap replicates and Agaricus bisporus (ITS: JX684008.1, LSU: AY635775.1) as outgroup (black branch). Different levels of agriculture are presented in different colors: coral fungus agriculture gray, lower agriculture yellow, yeast agriculture red, domesticated agriculture blue, and leaf-cutting agriculture green. Numbers at nodes are percentages consensus support and colony numbers with species names are given next to branches. Because Apterostig- ma dentigerum grows a completely different (secondarily acquired) coral fungus its tree was generated in a separate analysis with the same human domesticated A. bisporus as outgroup. B) Histogram of the mean (± SE) number of nuclei per cell of the fungal symbionts, showing a clear separations between the three basal levels in the phylogeny (2.40 ± 0.29 nuclei per cell for coral fungus farming, 2.07 ± 0.05 for lower attine fungus farming, and 2.00 ± 0.00 for yeast farming) and the two advanced levels of fungus farming that both rear clades of gongylidia-bearing fungi (10.80 ± 0.43 for Trachymyrmex and Sericomyrmex symbionts and 12.46 ± 0.41 for Atta and Acromyrmex symbionts). Representative pictures of stained nuclei are given towards the right: C) Atta sexdens (14 nuclei per cell), D) Sericomyrmex amabilis (8 nuclei per cell), E) Cyphomyrmex costatus (2 nuclei per cell), and F) Apterostigma dentigerum (2 nuclei per cell). Arrows indicate the septae that separate the cells. 41 Results The ITS and LSU sequences of the strains obtained from the different colonies were compared to the NCBI GenBank database using blastn, which showed that all sequences had their highest similarities with previously sequenced fungal symbionts from each of the different attine ant species. Consistent with these earlier results, the phylogenetic tree showed distinct clades among the fungal symbionts comparable to those published previously (Mueller et al. 1998; Munkacsi et al. 2004; Vo et al. 2009; Mikheyev et al. 2010), with all leaf-cutting agriculture strains clustering together, all other gongylidia-bearing domesticated strains clustering together, and all lower- and paleoattine ant strains appearing as basal branches when using the closely related human domesticated Agaricus bisporus as outgroup (Figure 1A). A separate analysis with the same outgroup provided a complementary tree for the three pterulaceous fungal symbionts of Apterostigma dentigerum. Although microsatellite markers are known to evolve much faster than ribosomal sequences, these markers also recovered the split between the gongylidia-bearing fungi reared by Atta and Acromyrmex on one hand, and those reared by Trachymyrmex and Sericomyrmex on the other (Figure S1). This indicates that there is no gene flow between these fungal clades at our study site in Panama.

Practically no variation was found in the number of nuclei per cell across the lower agriculture lineages. With only two exceptions we always observed a constant number of two, with only two slightly higher counts (Cyphomyrmex longiscapus [Cylo005] and Apterostigma dentigerum [Aden003]) that were likely due to hyphal tips being observed while cells were dividing so that nuclei were already duplicated but a new septum had not yet been formed. Mapping the number of nuclei per cell on the branches of the phylogenetic tree produced a clear pattern, with all lower- and paleo-attine strains, coral fungus strains and yeast strains having two nuclei per cell, and all higher attine ant strains having high numbers (7-17) of nuclei per cell (Figure 1B). This sharp transition thus coincided with the origin of the fungal gongylidia, the uniquely modified hyphal tips whose only known purpose is to be eaten by the ants, an irreversible adaptation to obligate symbiotic life. Statistical analyses showed that the number of nuclei per cell differed between higher (domesticated) fungus-farming on one hand and lower fungus-farming on the other (colors in Figure 1A: F4,27 = 74.46, p < 0.0001), but not between the ant genera within each of these farming types (F4,27 = 1.64, p = 0.1941) or between species within ant genera (F6,27 = 1.71, p = 0.1560) (see also Figure S2).

Microsatellite allele numbers per locus varied between 2 and 10 (Table S2) and all loci produced higher allele number scores when the number of nuclei per cell was higher, although the slopes for

42 0.8 0.8 Figure 2 Regression coefficients (± SE) for the locus-specif-

0.6 0.6 ic increase of alleles scored with increasing number of nuclei plotted against the expected locus-specific heterozygosities. The fitted logistic regression line is 0.4 0.4 based on the eight loci with the highest allelic diver- ●

Ploidy coefficient -11.077+10.553x -11.077+10.553x ● sity: p(x) = e / (1 + e ) (QAIC = 0.2 0.2

● ● ● 13.150). The two excluded loci, shown in light grey, ●

● ● ● had standard errors that overlapped with the 95% c.l.

● 0.0 0.0 of the fitted regression line, corroborating that their contribution just lacked information rather than devi- 0.5 0.6 0.7 0.8 0.9 1.0 ating from the overall trend in a significant manner. Heterozygosity the two least variable loci (B447 with 3 alleles and C126 with a total of 4 but maximally 2 alleles per fungal clone) were very shallow and thus considered to be hardly informative (Figure S3). The regression line predicting the rate at which haplotypes would increase with the number of nuclei per cell for an ideal marker locus with an infinite allele number (h = 1) produced an estimate of 0.37 ± 0.21 (95% CL 0.09 – 0.77)(Figure 2). As this regression slope estimated the rate of ploidy increase beyond the dikaryon level of two, this indicates that ca. two out of every five nuclei in excess of the dikaryon starting point represented genetically distinct haplotypes. This implies that the average ploidy of the higher attine fungal symbionts was estimated to be 6.2 (95% CL 3.0 – 10.8).

Discussion Our results corroborate that the emergence of fungal gongylidia ca. 20 MYA is an even more pronounced irreversible transition on the attine fungus farming symbiosis than previously appreciated. It characterized the true domestication of fungal crops and produced four new genus-level branches in the ant phylogenetic tree, of which one split into the two extant genera of leaf-cutting ants 10 MYA (Schultz and Brady 2008; Mehdiabadi and Schultz 2010). Constant high levels of cultivar auto- and allopolyploidy throughout the higher attine ants suggests that evolutionary innovation in the crown group of the attine ants has been driven by significant improvements of crop quality related to dosage advantages and chimeric genetic diversity. It seems reasonable to hypothesize that autopolyploidy evolved first as that remains compatible with occasional sexual reproduction (Udall and Wendel 2006), but that later allopolyploidy effectively precluded most if not all meiotic cell division. The emergence of gongylidia induced about an order of magnitude increase in scale of operations, with colony sizes increasing to maximally ca. 3000 in

43 Trachymyrmex and Sericomyrmex higher attine ants, quite possibly because the truly domesticated cultivars were better able to decompose freshly shed leaves and flowers with higher nutritious value (De Fine Licht and Boomsma 2010). Leaf-cutting agriculture allowed a further increase in scale of operations by 1-2 orders of magnitude (Schultz and Brady 2008; Baer et al. 2009; Meh- diabadi and Schultz 2010), but how much of this was driven by evolutionary changes in the ants or the acquisition of a novel clade of crop symbionts 2.3 MYA (Mikheyev et al. 2010) remains unknown.

As far as mushrooms originating from attine fungus gardens have been found, they seem to be most prevalent in the lower attine ants (Mueller 2002), which still have closely related free-living sister lineages (Mueller et al. 1998; Pagnocca et al. 2001). This implies that the capacity to progress towards sexual reproduction, particularly in abandoned fungus gardens, is unlikely to have been lost (Mueller 2002). It has been suggested that higher attine fungi have retained the capacity for meiosis (Mikheyev et al. 2006), but observations of mushrooms are extremely rare, although the initiation of primordia can occasionally be observed in lab colony where they generally appear to fail (Fisher et al. 1994; Mueller 2002; P.W. Kooij, M. Schiøtt, H.H. de Fine Licht, personal observations). Whether consistent combinations of auto- and allopolyploidy would allow occasional recombination without the production of sexual mushrooms remains an open question, but seems a possibility as unrelated fungi are not always somatically incompatible when grown on the same agar plate (Poulsen and Boomsma 2005; Lind et al. 2007). However, without regular recombination fungal clones may accumulate deleterious mutations, which will tend to affect phenotypic performance in spite of polyploidy and will thus retain selection for occasional horizontal transfer (Mikheyev et al. 2007; Poulsen et al. 2009) in a system where vertical transmission via virgin queens is the norm (Weber 1972).

While the absence or rarity of recombination may be a long-term challenge for maintaining optimal phenotypic performance even in a well-protected crop symbiont, this problem is further aggravated by monoculture farming, a mutualism-stabilizing principle that convergently evolved in both the fungus-growing ants and termites via very different mechanisms (Poulsen and Booms- ma 2005; Aanen et al. 2009). Long-lived colonies thus expose the same fungal phenotypes to possible parasites, which would tend to increase disease pressure over time (Hamilton 1982). Allopolyploidy may be a helpful mechanism to alleviate this type of vulnerability, as it will allow the expression of many more resistance factors than a functionally diploid dikaryon could achieve. Analogies of this chimerism principle can be found in the expression of many MHC alleles in otherwise diploid vertebrate immune systems (Potts and Wakeland 1993) and in the evolution of

44 multiple queen-mating in hymenopteran eusocial insects, which allows the number of haplotypes expressed at the colony level to increase from three (under haplodiploid monogamy) to 10 when a queen mates with eight males. Recent studies have shown that this increase in colony-level heterozygosity indeed reduces vulnerability towards infections (Seeley and Tarpy 2007; Hughes et al. 2010).

Apart from ameliorating the absence of recombination and disease pressure, allopolyploidy in higher attine crop fungi may also have offered advantages that ultimately allowed the attine ants to abandon the brown decomposition food chain (Shik and Kaspari 2010) to become functional herbivores. This transition started with the Trachymyrmex and Sericomyrmex higher attines that can use modest amounts of soft leaves and petals to manure their fungus gardens, and culminated in the leaf-cutting ants that almost exclusively collect fresh plant material (Weber 1972; Hervey et al. 1977; De Fine Licht and Boomsma 2010). This chain of events implied progressing from litter substrate to live substrate actively defended by secondary plant compounds (De Fine Licht et al. 2013), a challenge that may have been easier to meet with an allopolyploid chimeric spectrum of fungal haplotypes that express larger arrays of complementary enzymes (Soltis et al. 1993), increase biochemical flexibility in other ways (Roose and Gottlieb 1976), or merely allow better growth (Otto and Whitton 2000). The latter may also be achieved with autopolyploidy, as gene dosis advantages may generally enhance performance of multinucleate cells even though nuclei are identical copies (Albertin and Marullo 2012). Fungal analogies of these plant examples could possibly also facilitate the degradation of more complex or recalcitrant substrates, as shown in the allopolyploid brewing yeast Saccharomyces pastorianus (Pope et al. 2007; Minato et al. 2009; Libkind et al. 2011). It is interesting that the only known other examples of fungal allopolyploidy are Heterobasidion species that attack live conifer trees (Hodges 1969; Johannesson and Stenlid 2004; James et al. 2009) and that the spectrum of leaf-cutting fungal decomposition enzymes evolved to become similar to that of many plant pathogens after the attine ants became herbivores (Schiøtt et al. 2010; De Fine Licht et al. 2013).

How the 7-17 alleles that we found are distributed across the multiple nuclei in the crop fungi of higher attine ants remains to be resolved, but two hypotheses for cases like this have been proposed (Scott et al. 2009): 1. The same primer pair amplifies the same set of multiple loci in each haploid nucleus, which would be likely if L. gongylophorus underwent recent genome duplications, and 2. There is genetic variation among the haploid nuclei in multinucleate mycelia. The consistent polyploidy across all lineages of higher attine cultivars suggests that recent duplications are unlikely to explain more than a small fraction of the variation, but explicit studies would be needed

45 to verify the validity of this, or the alternative explanation. This will not be an easy task as similar observations of multinucleate somatic cells in arbuscular mycorrhizal fungi have led to different interpretations. Also here multiple alleles can be found in single mycelia, which has been claimed to imply that a single mycelium could have genetically diverse nuclei (e.g. Kuhn et al. 2001; Hijri and Sanders 2005). However, evidence is now accumulating that this type of sequence divergence within mycelia may derive from differences within single nuclei rather than from differences between nuclei (Pawlowska and Taylor 2004; Lin et al. in press). It is interesting that the occurrence of sex has also remained ambiguous in the arbuscular mycorrhizal fungi (Sanders 2011).

Concluding remarks The adoption of fungus farming 50 MYA by the ancestor of all attine ants implied a radical aban- donment of the normal hunter-gatherer life style of all ants, but our documentation of functional allopolyploidy evolving in the ant crops 20 MYA strongly suggests that the domesticated, gongylidia-bearing fungal symbionts drove later evolutionary developments towards large scale fungus farming in complex societies of leaf-cutting ants (Bass and Cherrett 1995; Schultz and Brady 2008; Mehdiabadi and Schultz 2010). Our results do not allow us to differentiate between the various selection forces (compensation for lack of recombination, reducing disease vulnerability, performance advantages owing to multiple identical nuclei or disease resistance owing to genetic chimerism) that could have maintained high degrees of crop allopolyploidy, but it is interesting to note that the attine ants domesticated a lineage of fungi that lack clamp connections (Vellinga et al. 2003), so the main constraint for evolving polyploidy was absent, and that this potential was realized as soon as the gongylidia bearing obligate crop-lineage evolved. It is also intriguing that essentially all human agricultural crops are polyploid or originated from polyploid ancestors (Udall and Wendel 2006). Human cultural evolution of farming practices also started with largely vertically transmitted crops and gradual improvement of their quality, to be replaced by global horizontal transmission of industrially innovated cultivars only during the last half century.

Acknowledgements We thank the Smithsonian Tropical Research Institute (STRI), Panama, for providing logistic help and facilities to work in Gamboa, the Autoridad Nacional del Ambiente y el Mar (ANAM) for permission to sample ant colonies in Panama, R.M.M Adams, A. Illum, J. Liberti, B. Baer, H.H. De Fine Licht, S.P.A. Den Boer and W. Hughes for help with collecting the ant colonies or making some of their own collected colonies available, Gösta Nachmann for suggesting the statistical analysis for ploidy estimation, and the Danish National Research Foundation for funding (DNRF57).

46 References Aanen DK, De Fine Licht HH, Debets AJM, et al. (2009) High symbiont relatedness stabilizes mutualistic cooperation in fungus-growing termites. Science 326:1103–1106. doi: 10.1126/science.1173462 Adaskaveg JE, Gilbertson RL (1987) Vegetative incompatibility between intraspecific dikaryotic pairings of Ganoderma lucidum and G. tsugae. Mycologia 79:603–613. Albertin W, Marullo P (2012) Polyploidy in fungi: evolution after whole-genome duplication. P Roy Soc Lond B Bio 279:2497–2509. doi: 10.1098/rspb.2012.0434 Baer BC, Dijkstra MB, Mueller UG, et al. (2009) Sperm length evolution in the fungus-growing ants. Behav Ecol 20:38–45. doi: 10.1093/beheco/arn112 Bass M, Cherrett JM (1995) Fungal hyphae as a source of nutrients for the leaf-cutting ant Atta sexdens. Physiol Entomol 20:1–6. Buss LW (1987) The Evolution of Individuality. Princeton University Press, Princeton, New Jersey, USA, pp 201. Carvalho DB, Smith ML, Anderson JB (1995) Genetic exchange between diploid and haploid mycelia of Armillaria gallica. Mycol Res 99:641–647. doi: 10.1016/S0953-7562(09)80520-0 Comai L (2005) The advantages and disadvantages of being polyploid. Nat Rev Genet 6:836–846. doi: 10.1038/nrg1711 De Fine Licht HH, Boomsma JJ (2010) Forage collection, substrate preparation, and diet composition in fungus- growing ants. Ecol Entomol 35:259–269. doi: 10.1111/j.1365-2311.2010.01193.x De Fine Licht HH, Schiøtt M, Mueller UG, Boomsma JJ (2010) Evolutionary transitions in enzyme activity of ant fungus gardens. Evolution 64:2055–2069. doi: 10.1111/j.1558-5646.2010.00948.x De Fine Licht HH, Schiøtt M, Rogowska-Wrzesinska A, et al. (2013) Laccase detoxification mediates the nutritional alliance between leaf-cutting ants and fungus-garden symbionts. P Natl Acad Sci USA 110:583–587. doi: 10.1073/pnas.1212709110 De Mendoza A, Sebé-Pedrós A, Sestak MS, et al. (2013) Transcription factor evolution in eukaryotes and the assembly of the regulatory toolkit in multicellular lineages. P Natl Acad Sci USA 110:E4858–66. doi: 10.1073/pnas.1311818110 Ellingboe AH, Raper JR (1962) Somatic recombination in Schizophyllum commune. Genetics 47:85–98. Fisher PJ, Stradling DJ, Pegler DN (1994) Leaf cutting ants, their fungus gardens and the formation of basidiomata of Leucoagaricus gongylophorus. Mycologist 8:128–131. doi: 10.1016/S0269-915X(09)80159-6 Groth JV, Christ BJ (1992) Diversity in plant pathogenic fungi as influenced by host characteristics. In: Fritz RS, Simms EL (eds) Plant Resistance to Herbivores and Pathogens: Ecology, Evolution, and Genetics. The University of Chicago Press, Chicago, USA, pp 172–192 Hamilton WD (1982) Pathogens as causes of genetic diversity in their host population. In: Anderson RM, May RM (eds) Population Biology of Infectious Diseases. Springer-Verlag, Berlin, Heidelberg, New York, pp 269–296 Hervey A, Rogerson CT, Leong I (1977) Studies on fungi cultivated by ants. Brittonia 29:226–236. Hijri M, Sanders IR (2005) Low gene copy number shows that arbuscular mycorrhizal fungi inherit genetically different nuclei. Nature 433:160–163. doi: 10.1038/nature03069 Hintz W, Anderson JB, Horgen PA (1988) Nuclear migration and mitochondrial inheritance in the mushroom Agaricus bitorquis. Genetics 119:35–41. Hodges CS (1969) Modes of infection and spread of Fomes annosus. Annu Rev Phytopathol 7:247–266. Hughes WOH, Bot ANM, Boomsma JJ (2010) Caste-specific expression of genetic variation in the size of antibiotic-producing glands of leaf-cutting ants. P Roy Soc Lond B Bio 277:609–615. doi: 10.1073/pnas.082461999

47 Hummer KE, Nathewet P, Yanagi T (2009) Decaploidy in Fragaria iturupensis (Rosaceae). Am J Bot 96:713–716. doi: 10.3732/ajb.0800285 James TY, Johansson SBK, Johannesson H (2009) Trikaryon formation and nuclear selection in pairings between heterokaryons and homokaryons of the root rot pathogen Heterobasidion parviporum. Mycol Res 113: 583–590. doi: 10.1016/j.mycres.2009.01.006 Johannesson H, Stenlid J (2004) Nuclear reassortment between vegetative mycelia in natural populations of the basidiomycete Heterobasidion annosum. Fungal Genet Biol 41:563–570. doi: 10.1016/j.fgb.2004.01.002 Jones JR, Ranney TG, Lynch NP, Krebs SL (2007) Ploidy levels and relative genome sizes of diverse species, hybrids, and cultivars of Rhododendron. J Amer Rhododendron Soc 61:220–227. Kuhn G, Hijri M, Sanders IR (2001) Evidence for the evolution of multiple genomes in arbuscular mycorrhizal fungi. Nature 414:745–748. Langella O (2001) Populations. Available: http://bioinformatics.org/populations/ Libkind D, Hittinger CT, Valério E, et al. (2011) Microbe domestication and the identification of the wild genetic stock of lager-brewing yeast. P Natl Acad Sci USA 108:14539–14544. doi: 10.1073/pnas.1105430108 Lidzbarsky GA, Shkolnik T, Nevo E (2009) Adaptive response to DNA-damaging agents in natural Saccharomyces cerevisiae populations from “Evolution Canyon,” Mt. Carmel, Israel. PLoS ONE 4:e5914. doi: 10.1371/journal.pone.0005914 Lin K, Limpens E, Zhang Z, et al. (in press) Single nucleus genome sequencing reveals high similarity among nuclei of an endomycorrhizal fungus. PLoS Genet Lind M, Stenlid J, Olson Å (2007) Genetics and QTL mapping of somatic incompatibility and intraspecific interactions in the basidiomycete Heterobasidion annosum s.l. Fungal Genet Biol 44:1242–1251. doi: 10.1016/j.fgb.2007.04.013 Marçais B, Caël O, Delatour C (2000) Genetics of somatic incompatibility in Collybia fusipes. Mycol Res 104:304–310. Masterson J (1994) Stomatal size in fossil plants: evidence for polyploidy in majority of angiosperms. Science 264:421–424. doi: 10.1126/science.264.5157.421 Mehdiabadi NJ, Schultz TR (2010) Natural history and phylogeny of the fungus-farming ants (Hymenoptera: Formicidae: Myrmicinae: Attini). Myrmecol 13:37–55. Mikheyev AS, Mueller UG, Abbot P (2010) Comparative dating of attine ant and lepiotaceous cultivar phylogenies reveals coevolutionary synchrony and discord. Am Nat 175:E126–33. doi: 10.1086/652472 Mikheyev AS, Mueller UG, Abbot P (2006) Cryptic sex and many-to-one coevolution in the fungus-growing ant symbiosis. P Natl Acad Sci USA 103:10702–10706. doi: 10.1073/pnas.0601441103 Mikheyev AS, Mueller UG, Boomsma JJ (2007) Population genetic signatures of diffuse co-evolution between leaf- cutting ants and their cultivar fungi. Mol Ecol 16:209–216. doi: 10.1111/j.1365-294X.2006.03134.x Minato T, Yoshida S, Ishiguro T, et al. (2009) Expression profiling of the bottom fermenting yeast Saccharomyces pastorianus orthologous genes using oligonucleotide microarrays. Yeast 26:147–165. doi: 10.1002/yea.1654 Moore D, Robson GD, Trinci APJ (2011) From the haploid to the functional diploid. In: Moore D, Robson GD, Trinci APJ (eds) 21st Century Guidebook to Fungi. Cambridge University Press, Cambridge, UK, pp 179–197. Mueller UG (2002) Ant versus fungus versus mutualism: Ant-cultivar conflict and the deconstruction of the attine ant-fungus symbiosis. Am Nat 160:S67–S98. Mueller UG, Gerardo NM, Aanen DK, et al. (2005) The evolution of agriculture in insects. Annu Rev Ecol Evol S 36:563–595. doi: 10.1146/annurev.ecolsys.36.102003.152626 Mueller UG, Rehner SA, Schultz TR (1998) The evolution of agriculture in ants. Science 281:2034–2038.

48 Munkacsi AB, Pan J, Villesen P, et al. (2004) Convergent coevolution in the domestication of coral mushrooms by fungus-growing ants. P Roy Soc Lond B Bio 271:1777–1782. doi: 10.1098/rspb.2004.2759 Otto SP, Whitton J (2000) Polyploid incidence and evolution. Annu Rev Genet 34:401–437. doi: 10.1146/annurev.genet.34.1.401 Pagnocca FC, Bacci M Jr, Fungaro MH, et al. (2001) RAPD analysis of the sexual state and sterile mycelium of the fungus cultivated by the leaf-cutting ant Acromyrmex hispidus fallax. Mycol Res 105:173–176. Pawlowska TE, Taylor JW (2004) Organization of genetic variation in individuals of arbuscular mycorrhizal fungi. Nature 427:733–737. doi: 10.1038/nature02290 Pope GA, MacKenzie DA, Defernez M, et al. (2007) Metabolic footprinting as a tool for discriminating between brewing yeasts. Yeast 24:667–679. doi: 10.1002/yea.1499 Potts WK, Wakeland EK (1993) Evolution of MHC genetic diversity: a tale of incest, pestilence and sexual preference. Trends Genet 9:408–412. Poulsen M, Boomsma JJ (2005) Mutualistic fungi control crop diversity in fungus-growing ants. Science 307:741–744. doi: 10.1126/science.1106688 Poulsen M, Fernandez-Marin H, Currie CR, Boomsma JJ (2009) Ephemeral windows of opportunity for horizontal transmission of fungal symbionts in leaf-cutting ants. Evolution 63:2235–2247. doi: 10.1111/j.1558-5646.2009.00704.x Quinlan R, Cherrett JM (1978) Aspects of the symbiosis of the leaf-cutting ant Acromyrmex octospinosus (Reich) and its food fungus. Ecol Entomol 3:221–230. Quinlan R, Cherrett JM (1979) The role of fungus in the diet of the leaf-cutting ant Atta cephalotes (L.). Ecol Entomol 4:151–160. Raper CA (1976) Sexuality and life-cycle of the edible, wild Agaricus bitorquis. J Gen Microbiol 95:54–66. Raper JR (1955) Heterokaryosis and sexuality in fungi. T New York Acad Sci 17:627–635. Rizzo DM, May G (1994) Nuclear replacement during mating in Armillaria ostoyae (Basidiomycotina). Microbiol-Sgm 140:2115–2124. doi: 10.1099/13500872-140-8-2115 Roose ML, Gottlieb LD (1976) Genetic and biochemical consequences of polyploidy in Tragopogon. Evolution 30:818–830. Saksena KN, Marino R, Haller MN, Lemke PA (1976) Study on development of Agaricus bisporus by fluorescent microscopy and scanning electron microscopy. J Bacteriol 126:417–428. Sanders IR (2011) Fungal sex: meiosis machinery in ancient symbiotic fungi. Current Biology 21:R896–R897. doi: 10.1016/j.cub.2011.09.021 Schiøtt M, Rogowska-Wrzesinska A, Roepstorff P, Boomsma JJ (2010) Leaf-cutting ant fungi produce cell wall degrading pectinase complexes reminiscent of phytopathogenic fungi. BMC Biol 8:156. doi: 10.1186/1741-7007-8-156 Schultz TR, Brady SG (2008) Major evolutionary transitions in ant agriculture. P Natl Acad Sci USA 105:5435–5440. doi: 10.1073/pnas.0711024105 Scott JJ, Weskin MK, Cooper M, Mueller UG (2009) Polymorphic microsatellite markers for the symbiotic fungi cultivated by leaf cutter ants (Attini, Formicidae). Mol Ecol Resour 9:1391–1394. doi: 10.1111/j.1755-0998.2009.02684.x Seeley TD, Tarpy DR (2007) Queen promiscuity lowers disease within honeybee colonies. P R Soc B 274:67–72. doi: 10.1098/rspb.2006.3702

49 Shik JZ, Kaspari M (2010) More food, less habitat: how necromass and leaf litter decomposition combine to regulate a litter ant community. Ecol Entomol 35:158–165. doi: 10.1111/j.1365-2311.2009.01165.x Sia EA, Butler CA, Dominska M, et al. (2000) Analysis of microsatellite mutations in the mitochondrial DNA of Saccharomyces cerevisiae. P Natl Acad Sci USA 97:250–255. Soltis DE, Soltis PS, Rieseberg LH (1993) Molecular data and the dynamic nature of polyploidy. Crit Rev Plant Sci 12:243–273. doi: 10.1080/07352689309701903 Tamura KK, Peterson DD, Peterson NN, et al. (2011) MEGA5: molecular evolutionary genetics analysis using maximum likelihood, evolutionary distance, and maximum parsimony methods. Mol Biol Evol 28:2731–2739. doi: 10.1093/molbev/msr121 Udall JA, Wendel JF (2006) Polyploidy and crop improvement. Crop Sci 46:S3–S14. doi: 10.2135/cropsci2006.07.0489tpg Vellinga EC, Kok RPJ de, Bruns TD (2003) Phylogeny and taxonomy of Macrolepiota (Agaricaceae). Mycologia 95:442–456. Vo TL, Mueller UG, Mikheyev AS (2009) Free-living fungal symbionts (Lepiotaceae) of fungus-growing ants (Attini: Formicidae). Mycologia 101:206–210. doi: 10.3852/07-055 Weber NA (1972) The fungus-culturing behavior of ants. Am Zool 12:577–587. Wertheim B, Beukeboom LW, van de Zande L (2013) Polyploidy in animals: effects of gene expression on sex determination, evolution and ecology. Cytogenet Genome Res 140:256–269. doi: 10.1159/000351998 Wilson A (1991) Somatic incompatibility in dikaryotic-monokaryotic and dikarytotic pairings of Echinodontium tinctorium. Can J Bot 69:2716–2723. Wood TE, Takebayashi N, Barker MS, et al. (2009) The frequency of polyploid speciation in vascular plants. P Natl Acad Sci USA 106:13875–13879. doi: 10.1073/pnas.0811575106

50 - 32 Acro and 11 Atta Domesticated agriculture Domesticated Leaf-cutting agriculture 14 35 20 24 33 12 3 31 23 53 56 37 7 , but there is little congruence within these clades. This within these clades. congruence , but there is little 98 62 57 78 52 15 37 24 10 microsatellites Ace23 Ae372 Ao616 Ae360 Ao510 Ao512 Ae451 Tzet028 Tcor002 Sam014 Sericomyrmex As-13-BB As-2011-4 Ac-2011-3 Ace-19-BB Ace-16-BB Ac-2009-46 Ac-2006-27 and MS120529-05Tzet MS120523-01Tzet MS120523-05Tcor AI110511-05Sama JOA120525-03Sama RMMA100611-11BigRed RMMA100629-16BigRed RMMA100629-17BigRed Trachymyrmex Trachymyrmex a t t a co r 01 4 120529-05 Tz e 120523-01 Tz e S S M RMMA100611-11BigRe d RMMA100629-16BigRe d Sa m RMMA100629-17BigRe d JOA120525-03Sa m Tcor00 2 M AI110511-05Sa m Ae36 0 Ac-2006-2 7 Ae37 2 As-13-B B Ae45 1 Ac-2009-4 6 Ac-2011- 3 Ao61 6 Ao51 0 As-2011- 4 Ace-16-B B Ace2 3 Ao51 2 Ace-19-B B Tzet02 8 MS120523-05 T 9 3 6 3 3 5 5 4 3 1 10 0 ITS and LSU 8 6 4 2 7 4 2 1 0 5 6 1 1 3 9 1 values using Neighbor-Joining with 500 bootstrap replicates (right). Both trees produce the same separation between the symbionts of symbionts the between separation same the produce trees Both (right). replicates bootstrap 500 with Neighbor-Joining using values 4 st 3 10 0 5 4 8 2 3 5 7 2 leaf-cutting ants and the symbionts of the other higher attine ants of the genera ants and the symbionts of other higher attine leaf-cutting 10 0 10 0 Figure S1 Figure tree the and (left) replicates bootstrap 500 with Likelihood Maximum using LSU sequences and ITS on based symbionts fungal for tree phylogenetic the between Comparison F microsatellite on based myrmex is consistent with microsatellite loci having much higher mutation rates than ribosomal genes (Sia et al. 2000), but also shows that there is no gene flow between the two fungal type (forest, open grassland, or both), which is branches to be structured according habitat for the ribosomal RNA There is a tendency lineages at our sampling site in Panama. each within are colonies when happen only can symbionts fungal of transfers horizontal rare that rather but selection, by affected are loci non-coding these that reflect to unlikely foraging range. other’s

51 14

Mean number of nuclei per cell 4

0 um r netzi xdens imosus r co r Atta s e Atta colombica Atta cephalotes Mycocepurus smithii Trachymyrmex z eteki Cyphomyrmex Sericomyrmex amabilis Trachymyrmex Acromyrmex echinatior Myrmecocrypta ednella Cyphomyrmex costatus Trachymyrmex cf. z eteki Apterostigma dentige Acromyrmex octospinosus Cyphomyrmex longiscapus

Figure S2 The mean number of nuclei per cell was significantly different for the five different types of fungus farming with the gongylidia bearing symbionts of the higher attine ants (including leaf-cutting ants) having consistently higher numbers of nuclei per cell than the symbionts reared by the lower- and paleo-attine ants (including the yeast growers)(F4,27 =

74.46, p < 0.0001). No significant differences were found between the genera within these types of farming (F4,27 = 1.64, p = 0.1941) or between ant species within genera (F6,27 = 1.71, p = 0.1560). Bars represent means ± SE and colors are as in Figure 1.

52 15 Locus A128 Locus A1030 15

R 2 = 0.4964 R 2 = 0.7363

F1, 17 = 16.76 F1, 18 = 50.25 p = 0.0007575 p = 1.312e-06 10 10 Coef = 0.13373 Coef = 0.23212

● 5 5

Figure S3 ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● Regression plots for the number of dif- ● ● ● ● ● ● ● 0 0 ferent alleles observed per symbiont ● ● ● ● ● ● 15 15 sample against the average number of Locus A1132 Locus A1151 R 2 = 0.6018 R 2 = 0.6078 nuclei per cell for each of the ten micro- F1, 18 = 27.21 F1, 24 = 37.19 p = 5.83e-05 p = 2.678e-06 10 10 satellite loci. Samples with null alleles Coef = 0.1273 Coef = 0.29174 were excluded from the analysis, since ● ●

they provide incomplete information ● ● ● 5 5

● ● ● ● ● about heterozygosity. Regressions were ● ● ●

● ● ● ● ● ● ● ● ● forced through the ‘origin’ (in this case ● ● ● ● ● ●

● ● ● ● ● ● ● ● ● 0 0

2,1+h, the expected number of nuclei ● ● ● ● ● in a dikaryon and the expected number 15 Locus B12 Locus B447 15

2 2 of alleles to be observed for two nuclei, R = 0.2323 R = 0.5229 F1, 15 = 4.54 F1, 16 = 17.53 p = 0.05007 p = 0.0006969

which is equal to 1 plus the expected het- 10 10 Coef = 0.05812 Coef = 0.06881 erozygosity), by subtracting these values from the respective x and y values before 5 5 entering them in the analysis. Regression Number of alleles per colony ●

● ●

● ● ● ● ● ● ● slopes were expected to vary between ● ●

● ● ● ● ● ● ● ● ● ● ● ● ● ● ● 0 0

● ● zero (all additional nuclei are copies of ● ● the two unrelated nuclei of the original 15 Locus C101 Locus C126 15 dikaryon) and one (the dotted line that R 2 = 0.5391 R 2 = 0.0695 F1, 24 = 28.07 F1, 24 = 1.793 would be obtained if every additional p = 1.956e-05 p = 0.1931 10 10 Coef = 0.07107 Coef = -0.01431 nucleus would represent an independent unrelated haplotype and if a mark- 5 5 er locus would have an infinite number of alleles so that all these nuclei would ● ● ● ● ● ● ● ● ● ● ●

● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● also be scored as genetically different). 0 0 ● ● ● ● ● ● ● ● ● ● ● ●

Text boxes give locus name (Scott et al. 15 15 Locus C647 Locus D115

2009), proportion of explained variation R 2 = 0.3775 R 2 = 0.4448 F = 10.31 F = 19.23 2 1, 17 1, 24 (R ), F and p-values, and the slopes of p = 0.005126 p = 0.0001985 10 10 the regression lines. Coef = 0.11224 Coef = 0.15535 5 5

● ● ● ● ●

● ● ● ● ● ● ●

● ● ●

● ● ● ● ● ● ● ● ● ●

● ● ● ● ● ● ● ● ● 0 0

● ● ● ● ● ● ●

0 5 10 15 0 5 10 15 Average number of nuclei per colony

53 0.00E+00 0.00E+00 0.00E+00 0.00E+00 0.00E+00 0.00E+00 0.00E+00 0.00E+00 0.00E+00 0.00E+00 0.00E+00 0.00E+00 0.00E+00 0.00E+00 0.00E+00 0.00E+00 0.00E+00 0.00E+00 0.00E+00 0.00E+00 0.00E+00 0.00E+00 0.00E+00 0.00E+00 0.00E+00 E-value (LSU) sp., but the second the sp., but Lepiota rachymyrmex cornetzi rachymyrmex septentionalis rachymyrmex septentionalis rachymyrmex septentionalis rachymyrmex septentionalis rachymyrmex septentionalis rachymyrmex sp. 3 rachymyrmex zeteki rachymyrmex zeteki rachymyrmex septentionalis T Species Atta cephalotes Atta cephalotes Atta cephalotes Atta cephalotes Atta cephalotes Atta cephalotes Atta cephalotes Atta cephalotes Atta cephalotes Atta cephalotes Atta cephalotes Atta cephalotes Atta cephalotes Sericomyrmex bondari T T T T T T T T Atta cephalotes T Large subunit rRN A 1893.1 1893.1 1893.1 1893.1 1893.1 1893.1 1893.1 1893.1 1893.1 1893.1 1893.1 1893.1 1893.1 1902.1 1892.1 1892.1 1892.1 1892.1 1892.1 1893.1 1892.1 HQ191240.1 Accession number U 1 U 1 U 1 U 1 U 1 U 1 U 1 U 1 U 1 U 1 U 1 U 1 U 1 HQ191241.1 HQ191244.1 U 1 HQ191243.1 U 1 U 1 U 1 U 1 U 1 U 1 U 1 0.00E+00 0.00E+00 0.00E+00 0.00E+00 0.00E+00 0.00E+00 0.00E+00 0.00E+00 0.00E+00 0.00E+00 0.00E+00 0.00E+00 0.00E+00 0.00E+00 0.00E+00 0.00E+00 0.00E+00 0.00E+00 0.00E+00 0.00E+00 0.00E+00 0.00E+00 0.00E+00 0.00E+00 2.00E-104 E-value Closest match with GenBank ITS rachymyrmex septentionalis rachymyrmex septentionalis Atta insularis Atta insularis Atta insularis Atta insularis Atta insularis Atta insularis Atta insularis Atta insularis Atta insularis Atta insularis Atta insularis Atta insularis Atta insularis T Lepiota flammeotincta Sericomyrmex amabilis Sericomyrmex amabilis Sericomyrmex sp. Sericomyrmex sp. Sericomyrmex sp. Sericomyrmex sp. Sericomyrmex sp. T Ant species Atta insularis Sericomyrmex sp. DQ779958.1 DQ779958.1 DQ779958.1 DQ779958.1 DQ779958.1 DQ779958.1 DQ779958.1 DQ779958.1 DQ779958.1 DQ779958.1 DQ779958.1 DQ779958.1 DQ779958.1 EU561496.1 U85331.1 GU202430.1 GU202430.1 GU202429.1 GU202429.1 GU202429.1 GU202429.1 GU202429.1 EU561494.1 Accession number DQ779958.1 GU202429.1 LSU KF571943 KF571944 KF571945 KF571940 KF571941 KF571942 KF571955 KF571947 KF571948 KF571949 KF571952 KF571953 KF571954 KF571967 KF571980 KF571966 KF571970 KF571976 KF571977 KF571978 KF571950 KF571959 KF571979 KF571956 KF571975 ITS GenBank accession numbers KF571985 KF571986 KF571987 KF571982 KF571983 KF571984 KF571997 KF571989 KF571990 KF571991 KF571994 KF571995 KF571996 KF572009 KF572022 KF572008 KF572012 KF572018 KF572019 KF572020 KF571992 KF572001 KF572021 KF571998 KF572017 1BigRed cor 1- 1 Colony 1-05Sama 1-3 1-4 105 1 cor002 Ace-16-BB Ace-19-BB Ace-2010-23 Ac-2006-27 Ac-2009-46 Ac-20 1 As-13-BB As-20 1 Ae360 Ae372 Ae451 Ao510 Ao512 Ao616 T MS120523-05 T Tzet028 MS120523-01Tzet MS120529-05Tzet RMMA1006 1 RMMA100629-16BigRed RMMA100629-17BigRed Sam014 AI 1 JOA120525-03Sama “big red” Ant species omyrmex echinatior omyrmex octospinosus rachymyrmex cornetzi rachymyrmex zeteki rachymyrmex cf. zeteki T T T Sericomyrmex amabilis Atta cephalotes Atta colombica Atta sexdens Ac r Ac r Level Leaf-cutting agriculture Domesticated agriculture Table S1 Table All but one hit with Overview of the colonies used in this study with corresponding GenBank accession numbers and best matches for ITS LSU sequences. free-living a to similarities showed closest Tzet028 for ITS sequence The symbionts. fungus-growing ant other to similarity showed closest GenBank best hit was a fungus-growing ant (JQ617561.1, E-value = 2E-102).

54 0.00E+00 0.00E+00 0.00E+00 0.00E+00 0.00E+00 0.00E+00 0.00E+00 0.00E+00 0.00E+00 0.00E+00 0.00E+00 0.00E+00 0.00E+00 0.00E+00 0.00E+00 0.00E+00 0.00E+00 E-value Species Mycetophylax conformis Mycetophylax conformis Mycetophylax conformis parallelus Mycetarotes parallelus Mycetarotes parallelus Mycetarotes Myrmicocrypta ednella Mycetophylax conformis senticosus Mycetarotes Mycocepurus smithii Mycocepurus smithii Cyphomyrmex rimosus Cyphomyrmex rimosus Cyphomyrmex rimosus dentigerum Apterostigma dentigerum Apterostigma dentigerum Apterostigma Large subunit rRNA (LSU) Large subunit rRNA Accession number U11906.1 U11906.1 U11906.1 U11907.1 U11907.1 U11907.1 FJ895832.1 U11906.1 U11896.1 FJ895822.1 FJ895830.1 U11903.1 U11903.1 U11903.1 AY367625.1 AY367617.1 AY367625.1 0.00E+00 0.00E+00 0.00E+00 0.00E+00 0.00E+00 0.00E+00 0.00E+00 0.00E+00 0.00E+00 0.00E+00 0.00E+00 0.00E+00 0.00E+00 0.00E+00 0.00E+00 0.00E+00 0.00E+00 E-value Closest match with GenBank ITS Cyphomyrmex muelleri Cyphomyrmex muelleri Cyphomyrmex muelleri Cyphomyrmex longiscapus Cyphomyrmex longiscapus Cyphomyrmex longiscapus Myrmicocrypta sp. 2 Mycocepurus smithii Myrmicocrypta sp. 2 Mycocepurus smithii Mycocepurus smithii Cyphomyrmex minutus Cyphomyrmex rimosus Cyphomyrmex minutus dentigerum Apterostigma dentigerum Apterostigma dentigerum Apterostigma Ant species JQ617738.1 JQ617738.1 JQ617723.1 JQ617746.1 JQ617736.1 JQ617742.1 AF079718.1 JQ405732.1 AF079718.1 JQ405690.1 JQ405690.1 AF079682.1 AF079694.1 AF079682.1 AY367597.1 AY367597.1 AY367597.1 Accession number LSU KF571951 KF571939 KF571957 KF571958 KF571960 KF571961 KF571974 KF571971 KF571972 KF571973 KF571968 KF571963 KF571964 KF571965 KF571946 KF571962 KF571969 ITS GenBank accession numbers KF571993 KF571981 KF571999 KF572000 KF572002 KF572003 KF572016 KF572013 KF572014 KF572015 KF572010 KF572005 KF572006 KF572007 KF571988 KF572004 KF572011 Colony AI150511-04Cyco 100610-02Cyco Cyco006 Cylo005 JOA120605-02Cylo JOA120605-08Cylo MyrE002 MS120529-11Medn MS120529-12Medn MycS015 MS120523-08MycS MS120520-02Crim MS120520-03Crim MS120521-04Crim Aden003 MS120519-03Aden MS120524-03Aden Ant species Cyphomyrmex costatus Cyphomyrmex longiscapus Myrmicocrypta ednella Mycocepurus smithii Cyphomyrmex rimosus dentigerum Apterostigma Level Yeast agriculture Yeast Lower agriculture Coral fungus agriculture

55 208 208 208 208 208 208 208 208 208 208 208 208 208 A1030 206 206 206 206 206 206 206 206 206 206 A1030 204 204 204 204 204 204 A1030 212 212 212 212 212 212 212 A128 210 210 210 210 210 210 210 210 A128 208 208 208 208 208 208 A128 206 206 206 206 206 206 206 206 206 206 206 206 206 206 206 206 A128 204 204 204 204 204 204 204 204 204 204 204 204 A128 202 202 202 202 202 A128 196 196 196 A128 Level Domesticated agriculture Domesticated agriculture Domesticated agriculture Domesticated agriculture Domesticated agriculture Domesticated agriculture Domesticated agriculture Domesticated agriculture Domesticated agriculture Domesticated agriculture Domesticated agriculture Leaf-cutting agriculture Leaf-cutting agriculture Leaf-cutting agriculture Leaf-cutting agriculture Leaf-cutting agriculture Leaf-cutting agriculture Leaf-cutting agriculture Leaf-cutting agriculture Leaf-cutting agriculture Leaf-cutting agriculture Leaf-cutting agriculture Leaf-cutting agriculture Leaf-cutting agriculture Leaf-cutting agriculture Species amabilis amabilis amabilis cf. zeteki cf. zeteki cf. zeteki cornetzi cornetzi zeteki zeteki zeteki echinatior echinatior echinatior octospinosus octospinosus octospinosus cephalotes cephalotes cephalotes colombica colombica colombica sexdens sexdens rachymyrmex rachymyrmex rachymyrmex rachymyrmex rachymyrmex rachymyrmex rachymyrmex rachymyrmex Genus Sericomyrmex Sericomyrmex Sericomyrmex T T T T T T T T Acromyrmex Acromyrmex Acromyrmex Acromyrmex Acromyrmex Acromyrmex Atta Atta Atta Atta Atta Atta Atta Atta 1BigRed cor 1- 1 leaf-cutting agriculture 1-05Sama 1-3 1-4 Atta 105 1 cor002 Only domesticated agriculture Only leaf-cutting agriculture Only Colony AI 1 JOA120525-03Sama Sam014 RMMA1006 1 RMMA100629-16BigRed RMMA100629-17BigRed MS120523-05 T T MS120523-01Tzet MS120529-05Tzet Tzet028 Ae360 Ae372 Ae451 Ao510 Ao512 Ao616 Ace-16-BB Ace-19-BB Ace23 Ac-2006-27 Ac-2009-46 Ac-20 1 As-13-BB As-20 1

56 167 167 167 167 167 167 167 167 167 167 167 167 167 167 A1151 165 165 165 165 165 165 165 165 165 165 165 165 165 165 165 165 165 165 165 A1151 163 163 163 163 163 163 163 163 163 163 163 163 163 A1151 161 161 161 161 161 161 A1151 157 157 157 157 157 157 157 157 157 157 157 157 A1151 155 155 155 155 155 155 155 155 155 155 155 A1151 153 153 153 153 153 153 153 153 153 153 153 153 A1151 151 151 151 151 151 151 151 151 A1151 213 213 A1132 211 A1132 208 208 208 208 208 208 208 208 208 208 208 208 208 208 A1132 205 205 205 205 205 205 205 205 205 205 205 A1132 202 202 202 202 202 202 202 202 202 202 202 202 202 202 202 202 A1132 199 199 199 199 199 199 199 199 199 199 A1132 196 A1132 193 193 A1132 216 216 216 216 216 216 216 A1030 214 214 214 214 214 214 214 214 214 214 214 A1030 212 212 212 212 212 212 212 212 212 212 212 212 212 212 212 212 A1030 210 210 210 210 210 210 210 210 210 210 210 210 210 210 210 A1030

57 115 115 115 115 115 115 115 115 115 115 115 115 115 115 115 C101 112 112 112 112 112 112 112 112 112 112 112 112 112 C101 110 110 C101 107 C101 104 104 104 104 104 104 104 C101 101 101 101 C101 95 95 95 95 95 95 95 95 95 95 95 95 C101 89 89 89 89 89 89 89 89 89 C101 222 222 222 222 222 222 222 222 222 222 222 222 222 222 222 222 222 B447 220 220 220 220 220 220 220 220 220 220 220 220 220 220 220 B447 218 218 218 218 218 218 218 218 B447 251 B12 249 B12 245 245 B12 243 243 B12 239 239 239 239 239 B12 237 B12 235 235 235 235 235 235 235 235 235 235 235 235 235 235 235 B12 233 233 233 233 233 233 233 233 233 233 B12 231 231 B12 177 A1151 175 A1151 173 173 A1151 171 171 171 171 A1151 169 169 169 169 169 169 169 169 169 169 169 169 A1151

58 249 249 249 249 249 249 249 249 249 249 D115 243 243 243 243 243 243 243 D115 240 240 240 240 240 240 240 240 240 240 240 240 240 240 240 D115 237 237 237 237 237 D115 234 234 234 234 234 234 234 234 234 234 234 234 234 234 234 234 234 234 D115 231 231 231 231 231 231 231 231 231 231 231 231 231 231 231 231 D115 228 228 228 D115 222 D115 219 219 219 219 219 219 219 219 D115 223 223 223 C647 221 221 C647 219 C647 215 215 215 215 215 215 215 215 215 215 C647 213 213 213 213 213 213 213 213 213 213 213 213 213 213 213 213 213 213 C647 211 211 211 211 211 211 211 211 211 211 211 211 211 211 C647 209 209 C647 207 209 207 C647 245 245 245 245 245 245 245 245 245 245 245 C126 242 242 242 242 242 242 242 242 242 242 242 242 242 242 242 242 242 242 C126 239 239 C126 149 149 149 149 149 149 149 C126

59 60 Somatic incompatibility of fungal crops in sympatric Atta and Acromyrmex leaf-cutting ants

Pepijn W Kooij, Michael Poulsen, Morten Schiøtt, Jacobus J Boomsma

(Manuscript) Chapter 2 61 Somatic incompatibility of fungal crops in sympatric Atta and Acromyrmex leaf-cutting ants

Pepijn W. Kooij, Michael Poulsen, Morten Schiøtt, and Jacobus J. Boomsma

Centre for Social Evolution, Department of Biology, University of Copenhagen, Universi- tetsparken 15, 2100 Copenhagen, Denmark

Phone: +45 35321239 Fax: +45 35321250

62 Abstract Obligate mutualistic symbioses rely on mechanisms that secure host-symbiont commitments to maximize host benefits and prevent symbiont cheating. Evolutionary theory predicts that incompatibility mechanisms between clonal symbiont lineages may enforce such commitments, but the most advanced endosymbioses have exclusive uniparental and vertical transmission of symbionts so that resident symbiont lineages are never challenged. Although attine ants also transmit their fungus-garden symbionts vertically via colony founding queens, horizontal transmission is possible. Resident garden symbionts may therefore face competition by alternative symbionts, which does not serve their fitness interests and is also unlikely to offer short-term benefits to the host colony. Previous studies showed that somatic incompatibility mechanisms correlate with neutral marker genetic distance between fungal symbionts in Panamanian Acromyrmex leaf-cutting ants, but the extent to which this result applies more generally has remained unclear. We hypothesized that somatic incompatibility mechanisms are only likely to evolve between fungal genotypes that compete for the same niche, i.e. for being the clonal symbiont of one or a few ant species that share the same clade of fungal symbionts. We test this by comparing the expression of somatic incompatibility among plated fungi reared by sympatric Acromyrmex and Atta colonies in Panama. We show that genetic distances for neutral microsatellite and AFLP markers accurately predict somatic incompatibility for Acromyrmex symbionts, but that these correlations are weaker (microsatellites) or absent (AFLP) in sympatric Atta colonies. Neither set of genetic markers explains any incompatibility between combinations of symbionts from Atta and Acromyrmex on the same plates. Further analysis showed that the symbiont clades maintained by Atta and Acromyrmex are likely to represent separate gene pools, so that neutral markers are unlikely to be correlated with incompatibility loci that have experienced different selection regimes. We discuss possible reasons for somatic incompatibility among Atta symbionts being only modest, based on the likelihood of resident symbionts being challenged by competing symbionts during colony founding and when colonies are mature.

Keywords Leucoagaricus gongylophorus, commitment, mutualism, fungus-growing ants

63 Introduction Endosymbionts are normally asexual and transmitted by uniparental vertical inheritance (Sachs et al. 2011). Multicellular organisms thus have a single mitochondrial genotype and those that have photosynthesis rely on a single clone of plastids. The evolution of such obligate symbiotic mutualisms has strong elements of partner commitment driven by kin selection, because exclusive association of hosts with a single symbiont genotype ensures that any services to growth and sur- vival of the host will benefit clone mates of symbionts that are vertically transmitted when hosts reproduce (Frank 1994; Doebeli and Knowlton 1998; Sachs et al. 2004; Foster and Wenseleers 2006). The same logic implies that hosts and symbionts are potentially in conflict over the mode of symbiont transmission (Frank 1996; Douglas 2008), as symbionts would always benefit from additional horizontal transmission. However, hosts might suffer fitness losses from this form of commitment disloyalty and therefore suppress symbiont investments in sexual reproduction even if that incurs only a marginal cost (Frank 1996; Leigh 2010). When, in spite of such host efforts, symbiont lineages manage to co-infect hosts and compete for host commitment, hosts will be under selection to monitor this conflict and eliminate it when symbiont mixing implies a net loss of cumulative symbiont service to the host (Frank 1996).

The classical mitochondrial and plastid endosymbioses are so completely integrated with the host that their reduced genomes preclude any form of non-symbiotic life, and the same is true for many obligate and facultative endosymbionts with less reduced genomes (McCutcheon and Moran 2012). Many of these interactions likely represent adaptive endpoints of host-symbiont coevolution, where host-symbiont conflicts were resolved in favor of the hosts (McCutcheon and Moran 2012; Wernegreen 2012) or symbiont (Werren et al. 2008), but their advanced stage of symbiosis normally precludes direct tests of evolutionary conflict theory over symbiont mixing because symbionts cannot be reared in vitro. However, the fungus-growing ants offer a feasible model system to do such tests, because they have multiple obligate mutualisms including fungus gardens (Schultz and Brady 2008; Mikheyev et al. 2010) and cuticular Actinobacteria (Cafaro et al. 2011; Andersen et al. 2013) that are ectosymbionts for individual ants, but endosymbionts for the ant colonies. This condition implies that partners can be separated and reared in vitro without each other’s interference for sufficient periods of time to quantify antagonism between symbiont clones, monitor host reactions to alternative symbionts, and relate observed differences to the genetic characteristics of the interactants (Bot et al. 2001; Armitage et al. 2011; Seal et al. 2012).

As far as populations of fungus-growing ants have been studied, attine ant colonies have never been found to rear a multi-clone fungus garden (Acromyrmex, Poulsen and Boomsma 2005;

64 Apterostigma, Dentinger et al. 2009; Atta, Mueller et al. 2010; Cyphomyrmex, Green et al. 2002, Mehdiabadi et al. 2012). In all of these studies, there was substantial genetic variation among fungus-garden clones across sympatric colonies, consistent with the normal patterns of variation of mitochondrial and plastid organelles across individual animals and plants (e.g., Embley and Martin 2006). However, in contrast to these cellular endosymbionts, there may be considerable horizontal transfer of symbionts between attine colonies that have similar ecologies, with species belonging to the same genus often (but not always) sharing clades of symbionts (Green et al. 2002; Poulsen and Boomsma 2005; De Fine Licht and Boomsma 2011; Mehdiabadi et al. 2012), while sympatric attine ant genera normally rear distinct fungal symbiont clades (Mueller and Gerardo 2002; Vo et al. 2009; Dentinger et al. 2009; Mehdiabadi et al. 2012; Kooij et al. Chapter 1, in prep).

Horizontal swaps of fungus-garden symbionts between colonies of the same or closely related attine ant species reduce the efficiency of co-evolutionary adaptation at the lowest taxonomic level, but may allow ant lineages to replace any asexual crop symbiont that is compromised by genetic load or other forms of maladaptation to prevailing ecological conditions. However, as long as fungus-garden clones are thriving, they will also be under selection to actively defend their ant-care monopoly (Bot et al. 2001). Such defenses are expected to evolve when the threat to be replaced is real, i.e. any hostility of this kind should target non-self symbiont genotypes belonging to the clade of symbionts that can in fact partake in a viable symbiosis with a focal attine ant species.

Sympatry and co-exploitation of the same clade of fungus-garden symbionts characterize the Ac- romyrmex echinatior and Acromyrmex octospinosus populations in Gamboa, Panama (Bot et al. 2001; Richard et al. 2007a; Poulsen et al. 2009), where resident fungus gardens have been shown to maintain their clonal integrity by a combination of behavioral adaptations in the ants to remove and kill alternative fungus clones (Bot et al. 2001; Ivens et al. 2009) and the expression of somatic incompatibility reactions between clones from different Acromyrmex colonies reared together on the same agar plates (Poulsen and Boomsma 2005). These expressions of incompatibility correlated with AFLP genetic distances between pairs of fungal symbionts, a pattern that also applied to the fecal fluid of Acromyrmex large workers fed with fungus from their own and other colonies (Poulsen and Boomsma 2005). However, founding queens of these Acromyrmex species readily accepted fungal clones from other colonies, suggesting that horizontal transfers may be successful at this stage even though new combinations may imply reduced symbiont performance (Poulsen et al. 2009). This study also suggested that incompatibility mechanisms might be different for sympatric Atta leaf-cutting ants, which tend to rear different fungal symbionts (Mikheyev et al. 2006;

65 Mikheyev et al. 2007), and whose queens never forage during colony founding and therefore have negligible likelihood of encountering alternative symbionts at that stage.

In the present study, we focus explicitly on comparing sympatric mature colonies of Atta and Acromyrmex leaf-cutting ants to address the following questions: 1. Do plated fungal symbionts of Panamanian A. echinatior and Atta colombica express similar somatic incompatibility reactions when they are confronted with symbionts from other colonies? 2. To what extent does the intensity of these reactions differ within and between the two genera? 3. To what extent is the intensity of these reactions correlated with genetic distance between the fungal symbionts? 4. Are sympatric colonies of Acromyrmex and Atta rearing the same or overlapping set(s) of fungal symbionts or are they associated with segregated lineages of Leucoagaricus gongylophorus? The same set would be expected if horizontal symbiont transmission between genera is more frequent than natural divergence of lineages via mutation and genetic drift. Segregated lineages would be expected when horizontal transmission between genera is rare or absent because co-adaptations in L. gongylophorus strains reared by Acromyrmex and Atta would preclude that horizontal symbiont swaps between genera are viable. At the same time, we also compared two different sets of genetic markers (AFLP and microsatellites) and different time spans between plate inoculation and scoring of incompatibilities to evaluate the robustness of our conclusions.

Natural history summary of attine ants and their fungi Attine ants evolved ca. 50 MYA from a hunter-gatherer ancestor to become obligate farmers of fungal crops, with strict domestication of the cultivar fungus evolving ca. 20 MYA and aggressive herbivory in leaf-cutting ants ca. 10 MYA ago (Mueller et al. 1998; Schultz and Brady 2008; Me- hdiabadi and Schultz 2010). Similar to all other fungus-growing ants, leaf-cutting ants rely on the successful maintenance of their fungus garden crop as main food source (Weber 1966; Quinlan and Cherrett 1978; Quinlan and Cherrett 1979). The large workers of Panamanian Acromyrmex species are known to actively remove fungal symbiont fragments from other congeneric colonies when they are genetically different from their resident garden fungus (Bot et al. 2001; Ivens et al. 2009), but whether foragers or nurses of Atta do the same is unknown.

The fungal symbiont reared by Atta and Acromyrmex leaf-cutting ants is considered to be a single species (Leucoagaricus gongylophorus; Agaricales, Basidiomycota), specifically associated with these two higher attine genera that appear to have obtained this symbiont via a continent-wide horizontal selective sweep ca. 2-3 MYA (Mikheyev et al. 2010). As all other fungal symbionts of the higher attine ants, L. gongylophorus is vertically and asexually transmitted by newly inseminated

66 colony-founding queens (Möller 1893; Wheeler 1907; Weber 1955; Weber 1972). Horizontal genetic exchange has been documented (Mikheyev et al. 2006; Mikheyev et al. 2007) but has only been studied systematically for Panamanian Acromyrmex (Poulsen et al. 2009) and not for Atta (see Green et al. 2002 and Mehdiabadi et al. 2012 for comparable studies in the lower attine genus Cyphomyrmex).

Somatic (in)compatibilities between plated fungi of Panamanian Acromyrmex are expressed in a gradual manner that correlates with AFLP genetic distances between pairs of clones (Poulsen and Boomsma 2005). In basidiomycete fungi, such somatic incompatibilities regulate allorecognition so that strains are increasingly likely to be incompatible when they are more genetically different (May 1988; Worrall 1997). These reactions tend to be stepwise (Rayner et al. 1984; Rayner 1991; Worrall 1997) and usually involve dark pigmentation in the interaction zone (Rayner et al. 1984), changes in septal maintenance, or blockage of septa precluding the movement of cytoplasma, and can lead to programmed cell death (Rayner 1991). The underlying genetic mechanisms remain largely unknown, but multiple loci appear to be involved (Worrall 1997) and their expression may be linked to sexual incompatibility genes (Van der Nest et al. 2009; Van der Nest et al. 2011). Sex may be present but very rare in the fungal symbionts of the higher attine ants (Fisher et al. 1994a; Fisher et al. 1994b; Pagnocca et al. 2001; Mueller 2002; Mikheyev et al. 2006) and recent work indicates that these symbionts are invariably functionally polyploid (Kooij et al. Chapter 1, in prep), which likely explains why somatic incompatibility patterns between fungal symbionts of Panamanian Acromyrmex colonies appear to be gradual (Poulsen and Boomsma 2005).

Materials and methods Biological material Fungal cultivars were isolated from nine A. echinatior colonies (Ae150A, Ae160, Ae168, Ae263, Ae266, Ae322, Ae356, Ae394, Ae488) and nine A. colombica colonies (Ac-2006-27, Ac-2009- 42, Ac-2009-46, Ac-2011-2, Ac-2011-3, Ac-2012-1, Ac-2012-2, Ac-2012-8, Ac-2012-31) living sympatically in Gamboa, Panama, and grown on 39 g/L Potato Dextrose Agar (Sigma-Aldrich) with the addition of 5 g/L yeast extract, 15 mg/L Tetracycline and 12 mg/L Streptomycin. DNA of each fungal strain was extracted using the Qiagen DNeasy Plant Tissue extraction kit and stored at -20°C until further analysis. The Gamboa sampling site was the same as where most previous colonies of the Copenhagen fungus-growing ant research program have been collected, including the Acromyrmex colonies studied by Bot et al. (2001), Poulsen & Boomsma (2005), Mikheyev et al. (2007), Richard et al. (2007a), Ivens et al. (2009) and Poulsen et al. (2009).

67 Genetic analyses To calculate the genetic distance between each of the fungal strains we used two different methods: Amplified Fragment Length Polymorphism (AFLP) and ten microsatellite markers (A128, A1030, A1132, A1151, B12, B447, C101, C126, C647 and D115 developed by Scott et al. (2009)). AFLP was performed as described by Vos et al. (1995) with two selective primer combinations (Eco- ACC + Mse-CAT and Eco-ACC + Mse-CAC). Microsatellite markers were analyzed using PCR with 5 µL VWR Red Taq DNA polymerase Master Mix (VWR International, Haasrode, Belgium), 0.25 µL forward and reverse primer each, 1.5 µL ddH2O and 1 µL DNA, and a program of 5 min denaturing at 95°C, followed by 14 cycles of 30 sec denaturing at 95°C, 30 sec annealing at 68- 58°C with a touchdown of -0.5°C per cycle, and 30 sec extension at 72°C followed by 20 cycles of 30 sec denaturing at 95°C, 30 sec annealing at 58°C and 30 sec extension at 72°C, and finally a 15 min extension at 72°C.

Both AFLP and microsatellite amplification products were analyzed on an ABI 3130xl (Applied Biosystems) sequencer. Specific allele scorings were obtained by analyzing chromatograms in Genemapper 4.0 (Applied Biosystems). The program Populations 1.2.32 (Langella 2001) was used to calculate Fst values for the microsatellite data and Nei’s standard genetic distance (Ds) for the AFLP data, followed by Neighbor Joining phylogenetic analyses with 500 bootstrap replicates for each of the two types of markers.

Somatic incompatibility To test whether plated fungi showed (in)compatibility, cultures of all 18 fungi were paired in all possible (171) combinations with four replicate pairings for each combination. For each pair, small tufts of mycelium (ca. 2 mm3) were placed at a distance of 1.5 cm from each other on a 5 cm Petri dish with 39 g/L Potato Dextrose Agar (Sigma-Aldrich) with the addition of 5 g/L yeast extract and 35 g/L Agar. The growth medium was selected in a pilot study testing somatic incompatibility reactions for control (self) encounters on this and three alternative media, which showed that the used PDYA medium most consistently avoided any unexpected discolorations in controls (Figure S1).

Incompatibility reactions were assessed after six, eight and ten weeks and scored using the semi-quantitative scale described by Poulsen and Boomsma (2005): 0 = demarcation zone absent, 1 = demarcation zone weak but present, 2 = demarcation zone broad and distinct, and 3 = strong demarcation zone with consistent brown or black coloration of mycelium. These four scores are consistent with the variation in somatic (in)compatibility reactions that are typically found in

68 free-living basidiomycetes, although usually in a more stepwise manner as explained above (Wor- rall 1997). All scorings were done blindly by randomly assigning numbers to each plate, after which two of the authors did the initial assessment and the third author blindly checked combinations for which the first two authors did not agree on the score. Degrees of (in)compatibility were subsequently compared with genetic distances between pairs of fungal clones with R (R Core Team 2013) using Mantel and Partial Mantel Tests for Dissimilarity Matrices (“mantel”) with 99999 permutations in the Community Ecology Package: Ordination, Diversity and Dissimilar- ities “vegan” (Oksanen et al. 2013). Figures were created using Plot With Repeated Symbols by Size (“sizeplot”) in the Plotrix package (Lemon 2006).

Results For Acromyrmex symbionts, consistent scorings of somatic incompatibility were obtained eight weeks after agar plates were inoculated, as coloration contrasts had not fully developed after six weeks, so Mantel correlation coefficients between incompatibility and genetic distance after six weeks remained low (Figure S2). Beyond eight weeks, Mantel correlation coefficients continued to improve, but contaminations and medium desiccation problems affected scoring accuracy 10 weeks after inoculation (Figure S2) so that we lost almost 5% of the replicates. As eight weeks was approximately the same as the two-months period between inoculation and scoring in Poulsen & Boomsma (2005), we decided to present our main results for the eight weeks observation period, to remain as comparable as possible with that previous study on somatic incompatibilities between A. echinatior and A. octospinosus fungal symbionts from the same site. However, for Atta symbionts we only obtained a comparable result when using microsatellite markers, as AFLP markers produced Mantel correlation coefficients close to zero for all observation periods (Figure S2).

Using the eight weeks data, somatic incompatibilities increased with increasing AFLP genetic distances between fungi (Mantel r = 0.463, p = 0.003, Figure 1B) in Acromyrmex symbionts, but not in Atta symbionts (Mantel r = -0.164, p = 0.792, Figure 1D). When using genetic distances based on microsatellite markers, both Acromyrmex (Mantel r = 0.469, p = 0.003, Figure 1A) and Atta (Mantel r = 0.312, p = 0.032, Figure 1C) symbionts had incompatibilities that increased with genetic distance, but less of the incompatibility variance was explained in Atta than in Acromyr- mex. The A. echinatior results were consistent with the results obtained by Poulsen & Boomsma (2005), but the variance explained by the Mantel coefficient was larger in that study (r = 0.855; p < 0.0001). However, when we used our ten-weeks scorings for Acromyrmex symbiont pairings we obtained a correlation closer to the one obtained after two months by Poulsen & Boomsma (2005)

69 ●●● ●● ● ●●● ● ●● ● ●●●●● ● ●● ● 3.0 ●● A B

● ● ●●● ●●●●● ● ● ●●● ● ●● ● ● ● ●●●● ● ●● ●● ● ● 2.0 ● Acromyrmex

●●● ● ●● ● ●●● ● ●●●● ●● ● ● ●●●●●●●● ● ●●● ●● ●● ● ● 1.0

● ● ●● ● ● ● ● ● ●● ●● 0.0

● ● ● ● ● ● 3.0 C D Somatic incompatibility

● ● ● ● ● ● ●● ● ● ● ● ●● 2.0 ● Atta

● ●● ●● ● ● ● ●● ● ●●●●● ●●●●● ●●● ● ● ●● ● 1.0 ● ● ● ● ●● ● ●●

● ● ● ● ● ● ● ●

0.0 ●

0.0 0.1 0.2 0.3 0.0 0.2 0.4 0.6 Microsatellite markers AFLP Genetic distance Figure 1. Mutual somatic incompatibility reactions for Acromyrmex and Atta-associated fungal symbionts plotted against genetic distances calculated from genetic variation at ten microsatellite markers and AFLPs, with the size of each circle representing the number of times a particular combination was found. Correlations were significant for Acromyrmex with both microsatellites (Mantel r = 0.469, p = 0.003) and AFLP markers (Mantel r = 0.463, p = 0.003), and for Atta with microsatellites (Mantel r = 0.312, p = 0.032) but not AFLP markers (Mantel r = -0.164, p = 0.792). All Mantel tests were performed with 99999 permutations.

(Microsatellites: Mantel r = 0.596, p < 0.001; AFLP: Mantel r = 0.654, p < 0.001).

Overall, the Atta fungi had smaller genetic distances when calculated from the microsatellite marker data (0.09 ± 0.01 SE), but larger genetic distances when using AFLP markers (0.26 ± 0.03 SE) compared to Acromyrmex (0.13 ± 0.01 SE and 0.17 ± 0.02 SE, respectively), which is reflected in the average number of AFLP bands observed (Atta: 37.3 ± 1.8 SE; Acromyrmex 30.9 ± 1.5

SE; t15.49 = -2.724, p = 0.015). This implied that AFLP and microsatellite genetic distances (Fst)

70 ●●●● ●●●●●●●●● ●●● ● ● ● ● ● ● ● ● ●● ● ● 3.0 ● ● ● ●● ●● ● ● ● A B

●● ●● ● ●●● ● ●●● ●●●●● ● ● ●● ● ● ●●● ● ● ● ●●● ●● ● ●● 2.0 ● ● ● ●

●● ●● ●●● ● ●●●●●●●●● ●●●●●●●●●● ●● ●● ● ●●● ● ● ●● ● ●● ● ● 1.0 ● ● ● ● ●●● vs Atta Acromyrmex ●● ● ● ● ●●●●●● ● Somatic incompatibility

● ●● ● ●●● ●●● ● ● ●● ● ● ●● ●●● ● ● ● ● 0.0 ● ● ●

0.0 0.1 0.2 0.3 0.0 0.2 0.4 0.6 Microsatellite markers AFLP Genetic distance

Figure 2. Somatic incompatibilities in confrontations between fungal symbionts across the Acromyrmex and Atta ant-genus-level barrier plotted against genetic distances calculated using microsatellite and AFLP markers, respectively. Each circle represents the presence of a combination of incompatibility score and genetic distance with the size of the circles representing the number of times a particular combination was found. Correlations were not significant for microsatellites (Mantel r = 0.1518, p = 0.2064) or AFLP (Mantel r = -0.1034, p = 0.7063). All mantel tests were performed with 99999 permutations.

were only comparable for Acromyrmex symbionts (r = 0.932; Figure S3), whereas correlations decreased in comparisons with Atta symbionts and became very low in comparisons involving only colonies of A. colombica (Figure S3). Independent of the type of genetic marker used, mean genetic distances were higher in comparisons across the ant genera (Microsatellites: 0.20 ± 0.01 SE; AFLP: 0.36 ± 0.02 SE) than in comparisons within genera (means ± SEs given above; microsatellite markers: F2,168 = 45.127, p < 0.0001; AFLP: F2,168 = 22, p < 0.0001). Also the average incompatibility scores were different for comparisons within ant species and comparisons across ant species (genera) (χ2 = 12.236; Df = 2; p < 0.01). This difference was mostly due to less strongly expressed somatic incompatibilities between Atta symbionts, as the stronger mean reactions among Acromyrmex symbionts alone and between Acromyrmex and Atta symbionts were not significantly different from each other (W = 1635, p = 0.338). After pooling data across this entire range of genetic distances, the increase in somatic incompatibility with genetic distance was no longer significant (Microsatellites: Mantel r = 0.153, p = 0.213, Figure 2A) or absent (AFLP: Mantel r = -0.045, p = 0.595, Figure 2B).

71 A Ac-2012-8 B Tzet028 Ac-2009-46 Atta-associated Ae322 fungi Ae150A 49 Ac-2012-1 54 Ae266 26 Ae160 45 71 Ac-2012-2 Ac-2011-3 Ae488 43 Ac-2012-31 Ac-2011-2 Ac-2009-42 57 88 16 Ae394 23 Ae394 84 Ae168 Ac-2006-27 40 61 Ae263 90 63 Ae356

61 26 63 Ac-2012-8 Ae488 64 34 Ae168 Ac-2009-46 24 Ae263 53 Ae356 69 Ac-2012-1 Ae160 53 31 Ae322 Ac-2006-27

Ae150A Acromyrmex- 21 Ac-2011-2 associated fungi 29 Ac-2011-3

52 Ac-2009-42

85 Ac-2012-31 0.1 Ae266 47 Ac-2012-2

Figure 3. (A) Circular phylogenetic tree based on Fst values estimated from the ten microsatellite loci showing a clear separation between fungal strains associated with Acromyrmex and Atta. (B) Rooted tree of the same Acromyrmex and Atta fungal symbionts with the symbiont of a sympatric colony of Trachymyrmex zeteki as outgroup. Both trees were calculated with Neighbor Joining and the numbers at nodes are percentage consensus support for 500 bootstrap replicates.

Specific evaluation of the microsatellite genetic differences between the fungal symbionts of the sympatric Atta and Acromyrmex colonies showed that they were completely separated (Figure 3A), consistent with earlier findings of Mikheyev et al. (2007) for the same sampling site. Rooting the symbiont tree with a sympatric fungal symbiont of Trachymyrmex zeteki confirmed that the symbionts belonged to separate clades although bootstrap values were low (Figure 3B). Mirror im- aging of trees obtained by microsatellite and AFLP markers showed almost complete congruence for the Acromyrmex fungi, but more noisy correspondence for the Atta fungal symbionts (Figure S4), confirming that these markers are less reliable when used to characterizeAtta symbionts.

Discussion The results of our study show that sympatric Panamanian colonies of A. echinatior and A. colombica rear genetically different lineages of the leaf-cutting ant garden symbiont L. gongylophorus and that microsatellite markers appear to predict genetic (in)compatibility better than AFLP markers. However, even when using microsatellite markers, the correlation between somatic incompatibility and neutral marker genetic distance in Atta is noisier than in Acromyrmex. These

72 incompatibility reactions correlate only with genetic distances among fungal strains that have a realistic probability of being horizontally transferred, and not between more distant clades that are apparently unsuitable as symbionts for the sister genus of leaf-cutting ants. We discuss these findings in more detail below.

Somatic incompatibility only between fungal symbionts that can be exchanged? The results of our study confirm the earlier findings by Poulsen and Boomsma (2005) showing that somatic (in)compatibility of Panamanian Acromyrmex symbionts is predictable from pairwise genetic distances for AFLP markers (Figure 1A, B) and show that the same result can be obtained with more specific microsatellite markers. They also indicate that incompatibility reactions between separate clades of fungal symbionts, maintained by different genera of leaf-cutting ants, can no longer be predicted by neutral genetic markers. This may be due to incompatibility being ultimately caused by allelic variation at unknown loci (Worrall 1997) that only correlate with neutral markers when there is recent common ancestry. This is only likely for local lineages of fungal symbionts that are exploited by a single or a few populations of attine ants that share the same pool of symbionts because each ant colony can in principle establish a viable symbiosis with each fungal genotype.

It is only in this situation that we expect somatic incompatibility to be actively maintained, because resident fungus-garden symbionts are selected to defend their monopoly and farming ants loose fitness when maintaining multiple lineages of the same symbiont that are equally viable but compete rather than serve their hosts (Frank 1996; Bot et al. 2001). When symbionts belong to different clades that no longer mix or exchange genes, each symbiont lineage is only a viable symbiont for one lineage of ant farmers or the other, but not for both. Our finding that A. colombica and A. echinatior maintain 100% separated clades of fungal symbionts (Figure 3) suggests that L. gongylophorus has in fact been split into an Atta and Acromyrmex clade after its monophyletic origin 2-3 MYA (Mikheyev et al. 2010), and that the two species of leaf-cutting ants that we investigated rear representatives of these symbiont lineages that are adapted to being, respectively, an Atta and Acromyrmex symbiont.

These findings are consistent with the results of a recent study that experimentally swapped fungus-garden symbionts between sympatric Trachymyrmex septentrionalis and Atta texana from Texas, USA (Seal et al. 2012). Although these ants are at the northern edge of the attine ant distribution (Mueller et al. 2011a), and likely to have lower genetic variation among their symbionts, they share even less common ancestry than the Panamanian fungal symbionts of our present

73 study (Mueller et al. 2011b). The swapped fungal symbionts thus could only serve as viable mutualists for either Trachymyrmex or Atta, but not both, which was confirmed in the experiments showing that: 1. Alternative fungus gardens were not always rejected by the ants, but they were never adopted as a viable alternative symbiont, because the ants were able to grow their original symbiont back from minuscule remnants that the authors had been unable to remove. 2. None of the T. septentrionalis colonies ever produced virgin queens when they maintained an A. texana fungal symbiont, consistent with the new combination being non-viable for transmitting ant of fungal genes to future generations. A follow up study transplanting A. texana fungus to colonies of both T. septentrionalis and T. turrifex confirmed that these Trachymyrmex cannot enter into viable symbiosis with L. gongylophorus symbionts and that, even though some virgin queens were produced on swapped gardens, they had poor fat reserves making it unlikely that they could successfully found colonies (Seal and Mueller 2013).

To make further progress, it would be highly desirable to identify the genes that are directly responsible for somatic incompatibility, as this would allow direct studies on the signatures of selection and specialization across the clades of higher attine ant symbionts. In another, non-eusocial model system of fungus-growing insects, the Sirex woodwasp, a range of genes are involved in somatic incompatibility reactions, including fusion and recognition genes, and genes that mediate cellular damage, stress response, and programmed cell death (Van der Nest et al. 2011). Whether these genes have homologs in attine ant fungal symbionts remains to be seen, as the Sirex symbiont Amylostereum areolatum belongs to a distantly related clade of basidiomycete fungi (Binder and Hibbett 2002), and their respective domestication histories may have implied that recognition systems were lost and gained over evolutionary time.

Why do Atta fungal symbionts express weaker incompatibility reactions? Incompatibility reactions among Atta symbionts were significantly weaker and less predictable from neutral marker Fst values than similar reactions between Acromyrmex-associated fungi. One possible explanation for this difference could be that there is a fundamental difference in colony founding in the sense that Acromyrmex queens forage during colony founding similar to all more basal attine ants, whereas Atta queens have secondarily evolved claustral colony founding. This implies that newly-mated Atta queens close off their nest cavity to raise the first worker cohort purely on their body reserves, so that new colonies will only be opened by these workers 80 to 100 days after they were founded (Weber 1966; Fernandez-Marin and Wcislo 2005). As a con- sequence, Acromyrmex queens may not only forage for leaf fragments to manure their incipient fungus garden, but also for gardens of other incipient colonies whose queens are out foraging

74 (Poulsen et al. 2009), an option that is unavailable for founding Atta queens. Although swapping of incipient fungus gardens with a fungus garden fragment from a mature Acromyrmex colony seems relatively unconstrained during early colony founding, it is likely that stronger mutual commitment between a founding queen and her resident fungus garden builds up in a matter of weeks, including relatively strong incompatibility reactions in case a queen or one of her first workers bring in an unrelated fungus garden fragment from a neighboring nest (Poulsen et al. 2009). This can never happen in the 80-100 days during which Atta colonies remain closed, removing selection for expressing incompatibility mechanisms during colony founding.

Why neutral markers should be less reliable predictors of somatic incompatibility in Atta colonies after workers start foraging remains unclear. Imprinting of workers on the odor of a resident fungus garden is a possibility (Seal et al. 2012), but it seems unclear why such mechanisms should differ between Atta and Acromyrmex symbionts and why that should reduce selection for more direct defenses by fungal symbionts against being replaced. Fungus gardens of Panamanian Acro- myrmex colonies differ in chemical profiles (Richard et al. 2007a; Richard et al. 2007b), but these differences do not correlate with genetic distances and comparable data for sympatric Atta colonies are lacking. The study by Richard et al. (2007a) also showed that feeding Acromyrmex workers with the fungus of another colony increased their chance of being accepted by members of that other colony. A compelling hypothesis may also be that mature Atta colonies in Panama have hundreds of fungus gardens, whereas sympatric mature Acromyrmex colonies have one or a few at best. This may make a difference in the likelihood of a resident fungus garden symbiont being replaced by an accidentally imported small fragment of fungus garden from a neighboring colony, so that less accurate recognition systems suffice in mature colonies ofAtta but not in Acromyrmex.

Finally, there could also be technical explanations for the incompatibility differences between the fungal symbionts of Atta and Acromyrmex. First, specific growth conditions may be needed to switch on incompatibility genes and the agar medium that we used may have approached these conditions for Acromyrmex better than for Atta symbionts. Second, we found similar variation for the Panama-symbiont-specific microsatellite markers (Scott et al. 2009) and the general AFLP markers for Acromyrmex symbionts, but enhanced variation in AFLP peaks for Atta symbionts, relative to Acromyrmex symbionts (Figure S3). This suggests that DNA from other organisms may have been amplified with the AFLPs and that such other organisms were only present inAtta symbiont cultures. In principle, this could be viral (Pearson et al. 2009), bacterial (Suen et al. 2010) or prion (Wickner et al. 2007) DNA. However, universal bacterial 16S primers did not amplify the DNA samples (P.W. Kooij, unpublished data), making bacterial contaminations highly unlikely.

75 Acknowledgements We thank the Smithsonian Tropical Research Institute (STRI), Panama, for providing logistic help and facilities to work in Gamboa, the Autoridad Nacional del Ambiente y el Mar (ANAM) for permission to sample ant colonies in Panama, a STENO grant from The Danish Council for Independent Research | Natural Sciences to MP, and the Danish National Research Foundation (DNRF57) and the ERC (323085) for funding.

References Andersen SB, Hansen LH, Sapountzis P, et al. (2013) Specificity and stability of the Acromyrmex-Pseudonocardia symbiosis. Mol Ecol 22:4307–4321. doi: 10.1111/mec.12380 Armitage SAO, Broch JF, Marín HF, et al. (2011) Immune defense in leaf-cutting ants: a cross-fostering approach. Evolution 65:1791–1799. doi: 10.1111/j.1558-5646.2011.01241.x Binder M, Hibbett DS (2002) Higher-level phylogenetic relationships of homobasidiomycetes (mushroom-forming fungi) inferred from four rDNA regions. Mol Phylogenet Evol 22:76–90. doi: 10.1006/mpev.2001.1043 Bot A, Rehner SA, Boomsma JJ (2001) Partial incompatibility between ants and symbiotic fungi in two sympatric species of Acromyrmex leaf-cutting ants. Evolution 55:1980–1991. Cafaro MJ, Poulsen M, Little AEF, et al. (2011) Specificity in the symbiotic association between fungus-growing ants and protective Pseudonocardia bacteria. P Roy Soc Lond B Bio 278:1814–1822. doi: 10.1098/rspb.2010.2118 De Fine Licht HH, Boomsma JJ (2011) Variation in fungal enzyme spectra may affect mutualistic division of labour between ants and fungus gardens. 13th Congress of the European Society of Evolutionary Biology Dentinger BTM, Lodge DJ, Munkacsi AB, et al. (2009) Phylogenetic placement of an unusual coral mushroom challenges the classic hypothesis of strict coevolution in the Apterostigma pilosum group ant-fungus mutualism. Evolution 63:2172–2178. doi: 10.1111/j.1558-5646.2009.00697.x Doebeli M, Knowlton N (1998) The evolution of interspecific mutualisms. P Natl Acad Sci USA 95:8676–8680. Douglas AE (2008) Conflict, cheats and the persistence of symbioses. New Phytol 177:849–858. doi: 10.1111/j.1469-8137.2007.02326.x Embley TM, Martin W (2006) Eukaryotic evolution, changes and challenges. Nature 440:623–630. doi: 10.1038/nature04546 Fernandez-Marin H, Wcislo WT (2005) Production of minima workers by gynes of Atta colombica Guérin-Ménéville (Hymenoptera: Formicidae: Attini) that lack a fungal pellet. J Kansas Entomol Soc 78:290–292. doi: 10.2317/0402.19.1 Fisher PJ, Stradling DJ, Pegler DN (1994a) Leaf cutting ants, their fungus gardens and the formation of basidiomata of Leucoagaricus gongylophorus. Mycologist 8:128–131. doi: 10.1016/S0269-915X(09)80159-6 Fisher PJ, Stradling DJ, Pegler DN (1994b) Leucoagaricus basidiomata from a live nest of the leaf-cutting ant Atta cephalotes. Mycol Res 98:884–888. doi: 10.1016/S0953-7562(09)80259-1 Foster KR, Wenseleers T (2006) A general model for the evolution of mutualisms. J Evolution Biol 19:1283–1293. doi: 10.1111/j.1420-9101.2005.01073.x Frank SA (1994) Genetics of mutualism: the evolution of altruism between species. J Theor Biol 170:393–400. doi: 10.1006/jtbi.1994.1200 Frank SA (1996) Host-symbiont conflict over the mixing of symbiotic lineages. P Roy Soc Lond B Bio 263:339–344. Green AM, Mueller UG, Adams RMM (2002) Extensive exchange of fungal cultivars between sympatric species of fungus-growing ants. Mol Ecol 11:191–195. doi: 10.1046/j.1365-294X.2002.01433.x

76 Ivens AB, Nash DR, Poulsen M, Boomsma JJ (2009) Caste-specific symbiont policing by workers of Acromyrmex fungus-growing ants. Behav Ecol 20:378–384. doi: 10.1093/beheco/arn150 Kooij PW, Aanen DK, Schiøtt M, Boomsma JJ (Chapter 1) Advanced farming ants rear polyploid crop fungi. Langella O (2001) Populations. Available: http://bioinformatics.org/populations/ Leigh EG (2010) The evolution of mutualism. J Evolution Biol 23:2507–2528. doi: 10.1111/j.1420-9101.2010.02114.x Lemon J (2006) Plotrix: a package in the red light district of R. R-News 8–12. Available: http://cran.r-project.org/web/packages/plotrix/index.html May G (1988) Somatic incompatibility and individualism in the coprophilous Basidiomycete, Coprinus cinereus. T Brit Mycol Soc 91:443–451. doi: 10.1016/S0007-1536(88)80121-9 McCutcheon JP, Moran NA (2012) Extreme genome reduction in symbiotic bacteria. Nat Rev Microbiol 10:13–26. doi: 10.1038/nrmicro2670 Mehdiabadi NJ, Mueller UG, Brady SG, et al. (2012) Symbiont fidelity and the origin of species in fungus-growing ants. Nat Commun 3:840. doi: 10.1038/ncomms1844 Mehdiabadi NJ, Schultz TR (2010) Natural history and phylogeny of the fungus-farming ants (Hymenoptera: Formicidae: Myrmicinae: Attini). Myrmecol 13:37–55. Mikheyev AS, Mueller UG, Abbot P (2010) Comparative dating of attine ant and lepiotaceous cultivar phylogenies reveals coevolutionary synchrony and discord. Am Nat 175:E126–33. doi: 10.1086/652472 Mikheyev AS, Mueller UG, Abbot P (2006) Cryptic sex and many-to-one coevolution in the fungus-growing ant symbiosis. P Natl Acad Sci USA 103:10702–10706. doi: 10.1073/pnas.0601441103 Mikheyev AS, Mueller UG, Boomsma JJ (2007) Population genetic signatures of diffuse co-evolution between leaf-cutting ants and their cultivar fungi. Mol Ecol 16:209–216. doi: 10.1111/j.1365-294X.2006.03134.x Möller A (1893) Die Pilzgärten einiger südamerikanischer Ameisen. 1–127. Mueller UG (2002) Ant versus fungus versus mutualism: Ant-cultivar conflict and the deconstruction of the attine ant-fungus symbiosis. Am Nat 160:S67–S98. Mueller UG, Gerardo N (2002) Fungus-farming insects: multiple origins and diverse evolutionary histories. P Natl Acad Sci USA 99:15247–15249. doi: 10.1073/pnas.242594799 Mueller UG, Mikheyev AS, Hong E, et al. (2011a) Evolution of cold-tolerant fungal symbionts permits winter fungiculture by leafcutter ants at the northern frontier of a tropical ant-fungus symbiosis. P Natl Acad Sci USA 108:4053–4056. doi: 10.1073/pnas.1015806108 Mueller UG, Mikheyev AS, Solomon SE, Cooper M (2011b) Frontier mutualism: coevolutionary patterns at the northern range limit of the leaf-cutter ant-fungus symbiosis. P Roy Soc Lond B Bio 278:3050–3059. doi: 10.1098/rspb.2011.0125 Mueller UG, Rehner SA, Schultz TR (1998) The evolution of agriculture in ants. Science 281:2034–2038. Mueller UG, Scott JJ, Ishak HD, et al. (2010) Monoculture of leafcutter ant gardens. PLoS ONE 5:e12668. doi: 10.1371/journal.pone.0012668 Oksanen J, Blanchet FG, Kindt R, et al. (2013) vegan: Community Ecology Package. Available: http://CRAN.R-project.org/package=vegan Pagnocca FC, Bacci M Jr, Fungaro MH, et al. (2001) RAPD analysis of the sexual state and sterile mycelium of the fungus cultivated by the leaf-cutting ant Acromyrmex hispidus fallax. Mycol Res 105:173–176. Pearson MN, Beever RE, Boine B, Arthur K (2009) Mycoviruses of filamentous fungi and their relevance to plant pathology. Mol Plant Pathol 10:115–128. doi: 10.1111/j.1364-3703.2008.00503.x

77 Poulsen M, Boomsma JJ (2005) Mutualistic fungi control crop diversity in fungus-growing ants. Science 307:741–744. doi: 10.1126/science.1106688 Poulsen M, Fernandez-Marin H, Currie CR, Boomsma JJ (2009) Ephemeral windows of opportunity for horizontal transmission of fungal symbionts in leaf-cutting ants. Evolution 63:2235–2247. doi: 10.1111/j.1558-5646.2009.00704.x Quinlan R, Cherrett JM (1978) Aspects of the symbiosis of the leaf-cutting ant Acromyrmex octospinosus (Reich) and its food fungus. Ecol Entomol 3:221–230. Quinlan R, Cherrett JM (1979) The role of fungus in the diet of the leaf-cutting ant Atta cephalotes (L.). Ecol Entomol 4:151–160. R Core Team (2013) R: A language and environment for statistical computing. Available: http://www.R-project.org/ Rayner ADM (1991) The challenge of the individualistic mycelium. Mycologia 83:48–71. Rayner ADM, Coates D, Ainsworth AM, et al. (1984) The biological consequences of the individualistic mycelium. In: Jennings DH, Rayner ADM (eds) The ecology and physiology of the fungal mycelium. Cambridge University Press, pp 509–540 Richard F-J, Poulsen M, Hefetz A, et al. (2007a) The origin of the chemical profiles of fungal symbionts and their significance for nestmate recognition in Acromyrmex leaf-cutting ants. Behav Ecol Sociobiol 61:1637–1649. doi: 10.1007/s00265-007-0395-1 Richard F-J, Poulsen M, Drijfhout F, et al. (2007b) Specificity in chemical profiles of workers, brood and mutualistic fungi in Atta, Acromyrmex, and Sericomyrmex fungus-growing ants. J Chem Ecol 33:2281–2292. doi: 10.1007/s10886-007-9385-z Sachs JL, Essenberg CJ, Turcotte MM (2011) New paradigms for the evolution of beneficial infections. Trends Ecol Evol 26:202–209. doi: 10.1016/j.tree.2011.01.010 Sachs JL, Mueller UG, Wilcox TP, Bull JJ (2004) The evolution of cooperation. Q Rev Biol 79:135–160. doi: 10.1086/383541 Schultz TR, Brady SG (2008) Major evolutionary transitions in ant agriculture. P Natl Acad Sci USA 105:5435–5440. doi: 10.1073/pnas.0711024105 Scott JJ, Weskin MK, Cooper M, Mueller UG (2009) Polymorphic microsatellite markers for the symbiotic fungi cultivated by leaf cutter ants (Attini, Formicidae). Mol Ecol Resour 9:1391–1394. doi: 10.1111/j.1755-0998.2009.02684.x Seal JN, Gus J, Mueller UG (2012) Fungus-gardening ants prefer native fungal species: do ants control their crops? Behav Ecol 23:1250–1256. doi: 10.1093/beheco/ars109 Seal JN, Mueller UG (2013) Instability of novel ant-fungal associations constrains horizontal exchange of fungal symbionts. Evol Ecol. doi: 10.1007/s10682-013-9665-8 Suen G, Scott JJ, Aylward FO, et al. (2010) An insect herbivore microbiome with high plant biomass-degrading capacity. PLoS Genet. doi: 10.1371/journal.pgen.1001129 Van der Nest MA, Slippers B, Steenkamp ET, et al. (2009) Genetic linkage map for Amylostereum areolatum reveals an association between vegetative growth and sexual and self-recognition. Fungal Genet Biol 46:632–641. doi: 10.1016/j.fgb.2009.06.002 Van der Nest MA, Steenkamp ET, Slippers B, et al. (2011) Gene expression associated with vegetative incompatibility in Amylostereum areolatum. Fungal Genet Biol 48:1034–1043. doi: 10.1016/j.fgb.2011.08.001 Vo TL, Mueller UG, Mikheyev AS (2009) Free-living fungal symbionts (Lepiotaceae) of fungus-growing ants (Attini: Formicidae). Mycologia 101:206–210. doi: 10.3852/07-055

78 Vos P, Hogers R, Bleeker M, et al. (1995) AFLP: a new technique for DNA fingerprinting. Nucleic Acids Res 23:4407– 4414. Weber NA (1955) Pure cultures of fungi produced by ants. Science 121:109–109. Weber NA (1966) Fungus-growing ants. Science 153:587–604. Weber NA (1972) The fungus-culturing behavior of ants. Am Zool 12:577–587. Wernegreen JJ (2012) Strategies of genomic integration within insect-bacterial mutualisms. Biol Bull 223:112–122. Werren JH, Baldo L, Clark ME (2008) Wolbachia: master manipulators of invertebrate biology. Nat Rev Microbiol 6:741–751. doi: 10.1038/nrmicro1969 Wheeler WM (1907) The fungus-growing ants of North America. B Am Mus Nat Hist 23:669–U130. Wickner RB, Edskes HK, Shewmaker F, Nakayashiki T (2007) Prions of fungi: inherited structures and biological roles. Nat Rev Microbiol 5:611–618. doi: 10.1038/nrmicro1708 Worrall JJ (1997) Somatic incompatibility in basidiomycetes. Mycologia 89:24–36.

79 3.0

2.5

2.0

1.5

1.0 Somatic incompatibility score

0.5

0.0 A B C A B C A B C A B C PDA PDA+ PDYA PDYA+

Figure S1. Average somatic incompatibilities on different media for the two intraspecific controls,Atta vs Atta (A) and Acromyrmex vs Acromyrmex (C), and the cross-confrontations between Atta and Acromyrmex symbionts (B). All media were Potato Dextrose Agar (PDA) with either 5 g/L Yeast Extract (PDYA) or 35 g/L Agar (PDA+) or both (PDYA+). The average incompatibilities of the intraspecific controls were significantly

lower than the scores obtained in the interspecific test combinations Ac-Ae (F2, 21 = 5.88, p < 0.01). The PDYA medium was selected for the remaining experiments because it had consistent zero scores for the in-

traspecific controls, even though there were no significant overall differences between the four media 3,(F 21 =

1.49, p = 0.25). Also the medium x type-of-confrontation interaction term was not significant (F6, 21 = 1.19, p = 0.35). Although we selected the PDYA for our further experiments, the results that we obtained should be reproducible in other media, with the possible exception of PDA.

80 Microsatellite markers AFLP 0.8 Acromyrmex Acromyrmex Atta Atta 0.6 0.4 0.2 0.0 −0.2

0.8 all all cross cross Mantel r coefficient 0.6 0.4 0.2 0.0 −0.2

6 8 10 6 8 10 Number of weeks between inoculation and scoring

Figure S2. Mantel correlation coefficients (r) for the level of incompatibility against genetic distance for each of the three time points (six, eight and 10 weeks) at which incompatibilities were scored. Correlations increased over time for the microsatellite markers, indicating stronger relationships between genetic distance and level of incompatibility, but this was only true for Acromyrmex when AFLP markers were used, and cross-confrontations never produced significant correlations. We decided to base our main analyses on the week 8 data, because in week 10 a total of 32 plates (4.7%) had to be discarded due to infections (6) or because they had dried out (26), so that fungal interactions could no longer be assessed.

81 0.8 A B ●

● ● ●

● ● ● ● 0.6

● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ●

0.4 ● ● ● ● ● ● ● ● ● ●● ● ● ● ● ● ● ● ● ●●● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ●

0.2 ●● ●● ● ● ●● ● ● ● ● ● ●● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ●● ● ● ● ● ● ● ● ● ● ● All ● Acromyrmex ● ● 0.0 AFLP 0.8 C D ●

● ● ●

● ● ● ● 0.6

● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ●

● ● ● ●

0.4 ● ● ● ● ● ●● ● ● ● ● ● ● ● ●● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ●

0.2 ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ●

Atta Acromyrmex vs Atta ● 0.0 0.0 0.1 0.2 0.3 0.4 0.0 0.1 0.2 0.3 0.4 Microsatellite markers

Figure S3. Comparisons of genetic distances calculated with AFLP and microsatellite markers. Pearson’s correlation coefficients were 0.580 for all comparisons (A), 0.932 for comparisons between Acromyrmex symbionts (B), 0.400 for comparisons between Atta symbionts (C), and 0.418 for the cross-comparisons between Acromyrmex and Atta symbionts (D).

82 Microsatellite markers AFLP

Ac-2012-31 Ac-2012-31

Ac-2012-2 Ac-2012-2 25 65 Ac-2009-42 Ac-2009-42

Ac-2011-3 Ac-2009-46 23 6 57 Ac-2012-1 Ac-2012-1 45 Ac-2009-46 Ac-2011-3 43 88 60 Ac-2012-8 Ac-2011-2 23 44

3 Ac-2011-2 Ac-2012-8 19

Ac-2006-27 16 Ae394 15

Ae394 Ac-2006-27

90 40 Ae168 Ae168 3 63 19 Ae263 Ae263 64 46

60 Ae356 Ae356 24 Ae322 Ae322

Ae150A Ae150A 61 67 53 74 Ae266 Ae266 24 24 Ae160 Ae160 69 43 Ae488 Ae488

Figure S4.

Comparison of unrooted phylogenetic trees based on Fst values estimated from the ten microsatellite loci (left; the same data as in Figure 3) and Nei’s standard genetic distances (Ds) calculated from AFLP markers (right), indicating that phylogenies are almost identical for Acromyrmex-associated fungal strains, but less consistent between the two markers for Atta-associated strains.

83 84 Differences in forage-acquisition and fungal enzyme activity contribute to niche segregation in Panamanian leaf-cutting ants

Pepijn W. Kooij, Joanito Liberti, Konstantinos Giampoudakis, Morten Schiøtt,

Jacobus J. Boomsma

(Submitted) Chapter 3 85 Differences in forage-acquisition and fungal enzyme activity contribute to niche segregation in Panamanian leaf-cutting ants

Pepijn W. Kooij*, Joanito Liberti, Konstantinos Giampoudakis, Morten Schiøtt, Jacobus J. Boomsma

Centre for Social Evolution, Department of Biology, University of Copenhagen, Universi- tetsparken 15, 2100 Copenhagen, Denmark

Phone: +45 35321239 Fax: +45 35321250

*Corresponding author: [email protected]

86 Abstract The genera Atta and Acromyrmex are often grouped as leaf-cutting ants for pest management assessments and ecological surveys, although their mature colony sizes and foraging niches may differ substantially. Few studies have addressed such interspecific differences at the same site, which prompted us to conduct a comparative study across six sympatric leaf-cutting ant species in Central Panama. We show that foraging rates during the transition between dry and wet season differ about 60 fold between genera, but are relatively constant across species within genera. These differences appear to match overall differences in colony size, especially when Atta workers that return to their nests without leaves are assumed to carry liquid food. We confirm that Panamanian Atta specialize primarily on tree-leaves whereas Acromyrmex focus on collecting flowers and herbal leaves and that species within genera are similar in these overall foraging strategies. Spe- cies within genera tended to be spaced out over the three habitat categories that we distinguished in Central Panama (forest, forest edge, open grassland), but each of these habitats normally had only a single predominant Atta and Acromyrmex species. We measured activities of twelve fungus garden decomposition enzymes, belonging to the amylases, cellulases, hemicellulases, pectinases and proteinases, and show that average enzyme activity per unit of fungal mass in Atta gardens is lower than in Acromyrmex gardens. Expression profiles of fungal enzymes in Atta also appeared to be more specialized than in Acromyrmex, possibly reflecting variation in forage material. Our results suggest that species- and genus-level identities of leaf-cutting ants and habitat-specific foraging profiles may give predictable differences in the expression of fungal genes coding for decomposition enzymes.

Keywords AZCL-assay, behavior, mutualism, plant degradation, Acromyrmex, Atta, Leucoagaricus gongylophorus

87 Introduction Maximizing the acquisition of high quality food under varying ecological conditions is expected to be under continuous natural selection. This notion has inspired many studies addressing optimal foraging strategies [1] and the extent to which related species realize niche segregation [2] and character displacement [3] to avoid interspecific competition. These processes often lead to (sub)habitat segregation [4-6] or food specialization, but few comparative studies have focused on generalist insect herbivores as it often remains unclear whether specialization within generalist strategies does in fact occur and what the decisive axes are along which niches and habitats may segregate [7, 8]. This question is particularly relevant for social insects, as they are central place foragers and often have a large impact on their surrounding communities. For wood eating termites that live in their food, pest management agencies will automatically accumulate comparative data on habitat and niche segregation among species and genera [9, 10], but such comparative studies have remained rare in the leaf-cutting ants.

Atta [FABRICIUS, 1804] and Acromyrmex [MAYR, 1865] leaf-cutting ants originated between 8 and 12 million years ago as the most specialized crown-group of the fungus growing ants (Attini [EMERY, 1913]) [11]. Their extant distribution ranges from warm-temperate South America up to the southern regions of the United States [12-14]. Throughout this range these ants are important (often dominant) herbivores and significant accelerators of nitrogen and phosphorus cycling [15]. They decompose the harvested live plant material through the mutualistic services provided by their fungus-garden symbiont Leucoagaricus gongylophorus [SINGER, 1986], which feeds the ants in exchange for the plant substrate provided [16]. Weber [14] estimated that ca. two kg of fresh plant material is needed to build one fungus garden in an Atta cephalotes [LINNAEUS, 1758] colony and that almost 6000 kg of fresh vegetation had been processed by the collective fungus gardens of a 6.5 year old colony of Atta sexdens [LINNAEUS, 1758].

Many species of leaf-cutting ants are considered pests in agricultural and urban areas [17]. For economic damage assessments, the genera Atta and Acromyrmex are often considered indiscrim- inately, in spite of large differences in colony size [14, 18], degree of worker polymorphism [18- 20], fungus garden enzyme activity [21], and foraging behavior [14, 19, 22, 23]. For example, Cherrett [24] showed that forage material of an Atta cephalotes colony in Guyana consisted mostly of leaves with flowers as a distinct minority class, similar to a later studied colony ofAtta colombica [GUÉRIN-MÉNEVILLE, 1844] in Panama [25], whereas forage of Costa Rican Acromyrmex octospinosus [REICH, 1793], Acromyrmex coronatus [FABRICIUS, 1804] and Acromyrmex volcanus [WHEELER, 1937] is known to consist of leaves, flowers and some fruit fragments [26].

88 However, to our knowledge no studies have been done to quantify differences of this kind simultaneously at the same site for an entire local guild of leaf-cutting ants. This implies that habitat-specificity, foraging efficiency, and leaf-processing in fungus gardens have not been compared with formal statistical analyses.

So far, 40 species of leaf-cutting ants have been described [11, 18], and they have all been hypothesized to rear the same polymorphic species of L. gongylophorus as fungal symbiont [27], despite the enormous distribution range mentioned above [12, 28] and the highly variable habitats and forage availabilities [18, 22, 24, 28-32]. A recent study [33] has indicated that the extant L. gongylophorus species is only 2-3 million years old, inferring that it must have swept through all leaf-cutting ant species while replacing the original fungus garden symbiont(s) that they had retained after coming into existence 8-12 million years ago. Other recent studies have shown that the L. gongylophorus fungus garden symbiont is highly plastic in its enzymatic responses to the various leaf-substrates that the ants deposit on their fungus gardens [21, 34], suggesting that forage type may systematically affect the expression of decomposition enzymes.

The objective of our study was to design a sampling scheme that allows the key characteristics of forage acquisition and processing to be compared across an entire guild of leaf-cutting ants. To achieve that goal, we quantified the diversity of forage material and the absolute and relative foraging rates for six sympatric leaf-cutting ants in the month of May, around the start of the rainy season, in Gamboa, Panama: Atta cephalotes, Atta sexdens, Atta colombica, Acromyrmex echinatior [Schultz, Bekkevold & Boomsma, 1998], Acromyrmex octospinosus, and Acromyrmex volcanus. Based on field surveys at this mosaic landscape of secondary growth forest and subur- ban areas for a period of two decades the following generalizations of habitat differentiation [35] appear to apply in Gamboa: Atta cephalotes and Acromyrmex volcanus are forest canopy foragers, whereas Acromyrmex octospinosus forages on the forest-floor. Atta colombica occurs both in the forest (usually at lower elevations) and in moist open grassland habitats, while Atta sexdens and Acromyrmex echinatior prefer open and sunlit nesting habitats for foraging. The latter two species extend their distributions towards the Pacific coast where annual rainfall is less than in Gamboa and natural habitat resembles savannas rather than a mosaic of forest patches [12, 28], matching their preference for open habitat in Gamboa. We supplemented our comparative data on foraging rates and substrate diversity with field measurements on the activity of extracellular enzymes in the fungus gardens maintained by the six leaf-cutting ant species to assess whether foraging preferences might be related to specific garden processing activities.

89 Material and methods Ant foraging behavior In May 2011 we located 9 – 11 foraging trails each for five of the six ant species (Table 1), always < 30m from the nest for Atta and < 5 m from the nest for Acromyrmex. The sixth species, Acro- myrmex volcanus, was so rare that only one trail was found. We observed Acromyrmex trails for 15 to 30 min and Atta trails for 2 min (or 4 times 0.5 min when trails were very busy) to obtain comparable data when counting ants that passed an imaginary line perpendicular to the trail. We replicated observations by sampling either trails of different colonies or multiple trails of the same colony going in different directions so they could be considered as independent samples of foraging habitat (Table S1). Diversity of forage material was classified in six categories: (parts of) flowers, (pieces of) fruit, herbaceous leaves, tree-leaves, other material (always rare), and ants carrying nothing on their way back to the colony. When in doubt, we verified the origin of forage particles by back-tracking the trail to the source. Observations were repeated across parts of the day (morning 9 AM-12 PM, afternoon 12 PM-5 PM, evening in the dark 10 PM – 11 PM) and compared statistically to see whether this made any difference.

To assess rates of foraging, we expressed our records in numbers of workers on trails per hour, counting both workers carrying material back to the nest and those without. Data were log-transformed to approximately equalize variances and analyzed with R [36], using Linear Mixed-Ef- fects Models (“lme”) [37] and generating p-values with General Linear Hypotheses (“glht”) in the package “multcomp” [38]. Proportional distributions of forage types (tree-leaves, herbaceous leaves, flowers, fruit, other) were analyzed with the same tests, with separate trails being considered as a random factor in both analyses. To test for heterogeneity across trails within species we created a sub dataset consisting of four Atta trails from two colonies (one Atta colombica and one Atta cephalotes) and compared the foraging category scores between trails and species (with trails nested within species, and 0.5 min replicate observations for each trail). This was only possible for these Atta species as we did not have replicate samples within single trails for Acromyrmex (Table S1).

Results for the proportional distribution of forage types were visualized using “heatmap.2” in the R package “gplots” [39]. Dendrograms were calculated with the “pvclust” package [40] using 1000000 bootstrap iterations. Final clustering plots were based on the overall similarities in mean proportions (p) between species and supplemented by estimates of the inverse Simpson Diversity 2 index (D = 1/[Σpi ]) to allow an explicit analysis of the degree of evenness (high D) between the different forage or expressed enzyme categories across species and genera of leaf-cutting ants.

90 The denominator of the index decreases when more categories (p) enter the equation, but when the number of categories is constant (as in our analyses) more even distributions will give lower sums in the denominator and thus higher values of D. For the purpose of our study, D therefore functions as an index of generalist foraging or equal enzyme expression, so that high values indicate that all categories are important and low values indicate specialization either on a subset of forage categories or on a subset of expressed enzymes that were most active.

AZCL enzyme activity assays For each of the six ant species the garden enzyme profiles were analyzed for five different colonies, with the exception of Acromyrmex volcanus for which only one colony was available, but where we could add data for another colony obtained in the previous year by H. H. De Fine Licht (pers. com.). For each colony, fungus gardens were dug up, and about equal size fragments (ca 80 mg) from top, middle and bottom layers of fungus gardens were collected and immediately ho- mogenized together to obtain representative average enzyme activity measures per colony. These measurements were subsequently performed using previously published methods, which are easily applicable in the field and give repeatable results [21, 34, 41]. In short, fungus garden material (240 mg) was crushed with a pestle in a 1.5 ml eppendorf-tube containing 1000 µl 0.05 M TRIS- HCl buffer (pH 7.0), vortexed immediately and then centrifuged for 15 minutes (15000g) after which the supernatant was removed and applied immediately to each of 12 different assay-plates containing 0.1g/L of the Azurine-Crosslinked (AZCL) substrates: amylose, arabinoxylan, barley ß-glucan, casein, collagen, debranched arabinan, galactan, galactomannan, HE-cellulose, rhamnogalacturonan, xylan and xyloglucan that were chosen because they yielded positive enzyme activities in an earlier study [21].

The assay-plates of 6 cm diameter were prepared separately for each substrate using an agarose medium (1% agarose, 23 mM phosphoric acid, 23 mM acetic acid, 23 mM boric acid), and pH adjusted according to the manufacturer’s description (Megazyme, Bray, Ireland). After the medium had solidified, round wells (area of ca. 0.1 cm²) were made in each plate with a cut-off pipette tip and 12 µl of the supernatant was applied to each well in triplicate. After 22 hours of incubation at 25°C the plates were photographed and the area of the blue halo surrounding each well (a quantitative measure for the absolute amount of enzyme activity [21, 34]) was measured using the software program ImageJ ver. 1.43u for Macintosh. Enzyme activity measurements were grouped into categories based on which plant cell wall component is the main target of the enzymes ([42] and Megazyme, Bray, Ireland): amylases (measured with amylose), cellulases (measured with barley ß-glucan and HE-cellulose), hemicellulases (measured with arabinoxylan, galactomannan,

91 xylan and xyloglucan), pectinases (measured with debranched arabinan, galactan and rhamnogalacturonan), and proteases (measured with casein and collagen). We present these data grouped for five categories of enzymes, implying that each of these categories had data from 2-3 enzymes, except for amylases that were represented by only a single enzyme amylose.

The enzyme activity scores were analyzed using a General Linear Model in SAS, with “colony” nested within “species” and “species” nested within “genus”. Colony was then treated as having a random enzyme class activity, composed of specific enzyme activities nested within enzyme activity class. This procedure implied that, we had to omit amylase, because we only had a single substrate (amylose = starch) for testing activity and because starch is not a primary challenge in the degradation of plant material [34]. We also left out rare Acromyrmex volcanus and thus report original mean values for enzyme activity in Figure 2 for all species and enzyme classes, whereas statistics given in the text refer to the reduced data set of four enzyme classes and the gardens of five ant species. In our final analyses (Figure 3) we combined the enzyme and foraging datasets and visualized patterns of association with the plot.PCA function after Principal Component Anal- ysis (“PCA”) using the “FactoMineR” package [43].

Results The three Atta species had an average foraging rate of 5014 ± 555 SE ants/h, with 3374 ± 476 SE ants (67%) returning to the nest carrying forage material and 1640 ± 203 returning unloaded, whereas the three Acromyrmex species had an average foraging rate of 80 ± 14 SE ants/h (t52.444 = -19.347, p < 0.0001) and no workers returning without forage (Table 1). Separate analyses, using the probability of a foraging trail having loaded ants, showed that the genus-level differences in 2 loaded and unloaded returning foragers per hour were highly significant (χ 1 = 7.567, p < 0.01), but that the differences in loaded returning workers between species within genera were not sig- 2 nificant (χ 4 = 1.257, p = 0.87).

Across genera, Atta foragers harvested significantly more tree-leaves thanAcromyrmex workers (z = 6.420, p < 0.0001), while Acromyrmex foragers collected significantly more (pieces of) flowers than Atta workers (z = -4.894, p < 0.0001). However, there were also differences in the most abundant forage category within genera. All Acromyrmex species preferred some combination of flowers and herbaceous leaves, but Acromyrmex volcanus was more flower-biased and Acromyrmex echinatior more herbaceous-leaves-biased (Figure 1A). Similarly, while Atta cephalotes primarily harvested tree-leaves (sexdens vs cephalotes, z = -4.987, p < 0.001; cephalotes vs colombica, z = 6.782, p < 0.0001), Atta colombica brought in more herbaceous leaves (cephalotes vs colombica,

92 Genus Species Trails Observation Total ants Loaded ants per hour Unloaded ants per hour echinatior 10 (10) 285 min 265 58 ± 10 0 ± 0 Acromyrmex octospinosus 10 (10) 285 min 458 80 ± 14 102 ± 26 0 ± 0 0 ± 0 volcanus 1 (1) 30 min 43 86 ± NA 0 ± NA cephalotes 11 (6) 26 min 1519 2740 ± 456 2408 ± 373 3374 ± 10 (7) 24.5 min 1724 5383 ± 1003 1640 ± 203 1115 ± 188 Atta colombica 476 sexdens 9 (7) 22 min 801 1173 ± 447 1012 ± 301

Table 1. Differences in foraging rate between loaded and unloaded foragers. Differences in foraging rate between loaded and unloaded foragers of Atta and Acromyrmex species in Gamboa, Pan- ama, with summary statistics on the number of trails observed per species (number of colonies in brackets), the total number of minutes of observation per species, the total number of ants counted while returning to their nests, and the foraging rates for loaded and unloaded returning workers: means (± SE) per genus and per species (see Table S1 for details).

z = -7.109, p < 0.0001), and Atta sexdens had approximately equal shares of all forage categories (Figure 1A), which confirmed earlier findings by De Vasconcelos [44].

We validated the statistical independence of our trail samples, using a subset of two Atta colonies (one Atta colombica and one Atta cephalotes), for which we had four replicated samples of the same trails (0.5 min each) and two separate trails per colony (Table S1). This recovered our ear-

lier result that the two Atta species have different fractions of forage categories (F4,60 = 89.48, p < 0.0001), but also showed that different trails of the same colony yielded similar results in spite of

covering non-overlapping fractions of the colony’s foraging habitat (F8,60 = 0.75, p = 0.65). Fur- ther ANOVA showed that frequencies of forage types between the different times of the day were

significantly different for Atta species (F8,218 = 4.451, p = 0.0001), and a post-hoc test indicated this was due to a higher share of herbaceous leaves in the afternoon compared to the evening (z = -3.929, p = 0.009). Acromyrmex species did not show any activity in the dark (evening) and

frequencies of forage types between morning and afternoon observations were not different (F4,76 = 1.457, p = 0.224).

Ranking the six species according to the diversity of forage material (Figure 1B) gave no signifi-

cant difference between species within genera in evenness of forage category use (F5,45 = 1.776, p = 0.137), but the pooled Atta species had a lower evenness in forage category use (D = 1.42±0.08

SE) than the pooled Acromyrmex species (D = 1.86 ± 0.13 SE; F1,49 = 5.435, p = 0.024). After excluding Acromyrmex volcanus where sample size was very small and two forage categories completely missing, the evenness trends in Figure 1B corresponded fairly well with the relative proportional forage acquisition data in Figure 1A with, from left to right, a clearly increasing

93 au bp B au bp A edge # Forest edge # 76 41 Forest edge 4 92 30 Open area 3 89 75 100 21 0 0.3 0.6 0.9 4 2 Value 100 83 1 volc octo echi col sex cep 98 71 3 0.491 ± 0.352 ± 0.027 ± 0.149 ± 0.054 ± 0.512 99 83 0.104 0.105 0.008 0.093 0.032 flowers 67 71 2 1

0.142 ± 0.115 ± 0.121 ± 0.089 ± 0.153 ± 0.000 0.082 0.073 0.057 0.058 0.051 fruit xden s olcanus

0.298 ± 0.417 ± 0.565 ± 0.251 ± 0.000 ± 0.000 0.081 0.100 0.098 0.093 0.000 herbaceous Atta se Atta colombic a Atta cephalote s

0.007 ± 0.000 ± 0.252 ± 0.323 ± 0.790 ± 0.349 tree

0.007 0.000 0.099 0.116 0.054 Acromyrmex v Acromyrmex echinatior Acromyrmex octospinosus 0.062 ± 0.115 ± 0.036 ± 0.188 ± 0.002 ± 0.140 D = 2.48 1.88 ± 0.18 1.77 ± 0.19 1.63 ± 0.221.37 ± 0.14 1.36 ± 0.08 0.031 0.083 0.016 0.121 0.001 other D = 1.86 ± 0.13 D = 1.42 ± 0.08

Figure 1. Differences in forage diversity. Differences in forage diversity between leaf-cutting ant species (nested within genera), using solid lines for Atta and dotted lines for Acromyrmex, and with typical foraging habitat indicated with dark green (forest), yellow (forest edge), and orange (open sunlit areas): (A) Heatmap showing differences between species and genera in the use of forage categories, with numbers representing the mean proportions ± SE of the forage types. Darker colors indicate higher mean acquisition proportions, with the top-dendrogram illustrating similarities between species/genera across means of the five forage categories (vertical axis). Ant species names are given as abbreviations (volc, octo, echi, col, sex, cep). (B) Dendrogram based on the Inverse Simpson Diversity Index of the five forage categories, indicating the degree of evenness across foraging categories (numbers below the branches are mean D-values ± SE per species and means per genus), showing that Acromyrmex has a broader (more even) spectrum (D = 1.86 ± 0.08 SE) of forage material than

Atta (D = 1.42 ± 0.13 SE; F1,49 = 5.435, p < 0.05). Numbers above the branch nodes represent Approximately Unbi- ased p-values (AU, red) and Bootstrap Probability values (BP, green). trend in tree-leaf use, a decreasing trend in the use of flowers, and hump-shaped trends in the use of herbaceous leaves and fruit. These inferences were supported by moderately high overall Ap- proximately Unbiased (AU) p-values and Bootstrap Probability (BP) values (Figure 1). As night foraging tended to decrease the acquisition of herbaceous leaves by Atta species, we probably underestimated the difference in dependence on tree leaves between Acromyrmex and Atta by obtaining most of our Atta observations at daytime.

In absolute quantities, fungus gardens of Acromyrmex showed a higher overall enzyme activity than gardens of Atta (F1,20 = 8.54, p < 0.01)(Figure 2A), but species within genera did not show significant additional differences (F3,20 = 1.32, p = 0.29). To further investigate the differences in

94 au bp A au bp B edge # edge #

66 41

4 28 21 0.5 1 1.5 2 3 Value 66 59 99 51 99 22 66 59 4 3 1 2 echi volc octo cep col sex 100 44 100 44 1 2 1.581 ± 1.716 ± 1.938 ± 1.454 ± 1.823 ± 1.382 ± amylases 0.109 0.031 0.113 0.150 0.109 0.081

1.038 ± 1.372 ± 1.426 ± 0.907 ± 0.968 ± 0.650 ± cellulases xdens 0.094 0.163 0.103 0.086 0.097 0.097 Atta se

1.014 ± 1.261 ± 1.205 ± 0.741 ± 0.736 ± 0.611 ± Atta colombic a hemicellulases Atta cephalotes 0.074 0.159 0.087 0.065 0.076 0.064 Acromyrmex v olcanus 0.942 ± 1.670 ± 1.652 ± 0.992 ± 1.081 ± 0.826 ± pectinases Acromyrmex echinatior 0.073 0.134 0.086 0.047 0.071 0.054 Acromyrmex octospinosus

D =44.64 ± 0.25 4.54 ± 0.09 4.54 ± 0.05 4.42 ± 0.07 .15 ± 0.10 3.98 ± 0.14 1.074 ± 1.322 ± 0.980 ± 0.551 ± 0.915 ± 0.549 ± proteinases 0.097 0.140 0.108 0.075 0.135 0.116 D = 4.55 ± 0.05 D = 4.18 ± 0.07

Figure 2. Differences in fungus garden enzyme activity. Differences in fungus garden enzyme activity between species grouped as in Figure 1 with solid lines for Atta and dotted lines for Acromyrmex, and with dark green, yellow and orange indicating the same habitat categories: (A) Heatmap showing differences between species and genera in fungus garden activity of enzyme classes, expressed as the mean area in cm2 ± SE of colored halos on AZCL plates across all assays for enzymes belonging to the amylases (1), cellulases (2), hemicellulases (4), pectinases (3) and proteinases (2). Darker colors in the heatmap indicate higher means, and the top-dendrogram illustrates similarities between species across all means for the five groups of enzymes, estimated by “pvclust” with 1000000 bootstraps. (B) Dendrogram based on the inverse Simpson Diversity Index of proportional enzyme activity showing that Acromyrmex fungus gardens have more even secretions across enzyme categories (D =

4.55 ± 0.05 SE) than Atta (D = 4.18±0.07 SE, F1,52 = 15.006, p < 0.0001). Numbers above the branch nodes represent Approximately Unbiased p-values (AU, red) and Bootstrap Probability values (BP, green). enzyme activity between the two genera, we decomposed the significant interaction term of ant genus and AZCL category (F10,780 = 4.95, p < 0.0001). This revealed significant differences between Acromyrmex and Atta for all substrates except rhamnogalacturonan. The spectrum of relative enzyme activities, as expressed by the inverse Simpson indices (Figure 2B), showed that Acromyrmex species tend to have more evenly distributed enzyme activities (D = 4.55 ± 0.05 SE) than Atta species who tend to specialize more on the expression of specific classes of enzymes

(D = 4.18 ± 0.07 SE; F1,52 = 15.006, p < 0.0001). No significant differences were observed for the evenness of the enzyme activity spectra for species within genera (Atta: F2,12 = 1.618, p = 0.239;

Acromyrmex: F2,9 = 0.111, p = 0.896).

95 Comparative analyses (PCA), with either fungus garden enzyme expression as a predictor variable and forage diversity as a response variable (Figure 3A and B) or vice versa (Figure 3C and D), confirmed a separation between the genera Atta and Acromyrmex (Figure 3B and D). Taking the fungus garden enzyme activities as predictor variables produced a first axis explaining 69.04% of the variation and a second axis explaining 12.02% of the variation. The first axis corresponded to overall enzyme activity and illustrates that general fungus garden enzyme activity is lower towards the left (predominantly Atta) and higher towards the right (predominantly Acromyrmex) (Figure 3A and B), confirming the results given in Figure 2. The vertical axis reflects higher amounts of pectinases (positive scores) versus higher amounts of proteases (negative scores). However, this does not correspond in any obvious way with genus-level differences, but may correlate with colony-level differences in the proportions of flowers and fruit in the forage (Figure 3A and B).

A similar pattern was obtained when the predictor and response variables were reversed, (Figure 3C and D). The first axis (explaining 33.11% of the variation) illustrates a preference for tree- leaves (mostly Atta) towards the left and a preference for herbaceous leaves (mostly Acromyrmex) towards the right (Figure 3C and D). The second axis (explaining 25.54% of the variation) indicates higher intake of fruit (negative scores) in weak association with cellulases, and of flowers (positive scores) mostly in association with pectinases. Here the leaf-cutting genera are also separated to some extent (centroid squares) with Acromyrmex having some preference for flowers and Atta for fruit, confirming the results depicted in Figures 1 and 2.

The PCA comparisons also revealed correlations between forage preference and fungus garden enzyme expression, as both PCA analyses showed a negative correlation between extent of acquisition of tree-leaves and overall intensity of enzyme activity. Kendall’s rank test for correlation showed that this negative trend was significant for most enzymes: amylases (z = -3.185, p < 0.01); pectinases (z = -2.608, p < 0.01); proteases (z = -3.465, p < 0.001), but not cellulases (z = -1.429, p = 0.153) and hemicellulases (z = -1.953, p = 0.051). The same analyses also found a negative correlation between fruit foraging and expression of amylases (z = -2.311, p < 0.05) and positive correlations between foraging on herbaceous leaves and expression of amylases (z = 3.643, p < 0.001), pectinases (z = 2.130, p < 0.05) and proteases (z = 3.404, p < 0.001) and between flower foraging and expression of the same enzymes: amylases (z = 2.712, p < 0.01), pectinases (z = 3.066, p < 0.01) and proteases (z = 3.769, p < 0.001). Foraging on other materials was only (pos- itively) correlated with the expression of pectinases (z = 2.366, p < 0.05).

96 ABVariables factor map Individuals factor map

Acromyrmex 1. 0 4 Atta

Pectinases .5 ) ) Amylases Flowers Other Herbacous 0. 00 Tree Hemicellulases 02 Acromyrmex Atta Cellulases Dim 2 (12.02%

Dim 2 (12.02% Fruit

Enzyme activit y Proteinases -0.5 -2 -1.0 -4 -1.0 -0.5 0.0 0.5 1.0 -2 024 Dim 1 (69.04%) Dim 1 (69.04%)

CDVariables factor map Individuals factor map

Acromyrmex 1. 0

Flowers 3 Atta .5

Other ) 12 Pectinases Acromyrmex Amylases Proteinases 0 0. 00 Tree Hemicellulases Atta

Forag e Cellulases -1 Dim 2 (25.54% Dim 2 (25.54%) Herbacous -0.5 -2 Fruit -3 -1.0

-1.0 -0.5 0.0 0.5 1.0 -4 -2 02 Dim 1 (33.11%) Dim 1 (33.11%)

Figure 3. Correlations between forage diversity and enzyme activity. Principal Component Analyses (PCA) using either the five enzyme groups of Figure 2 (A and B) or the five forage material categories of Figure 1 (C and D) as predictor variables (black arrows). Red arrows represent response vectors for forage material (A) or enzymes (C). The B and D panels complement the respective A and C panels by plotting PCA’s scores across the fungus garden measurements (B, 45 for Atta and 32 for Acromyrmex) the sampled ant trails (D, 30 for Atta and 21 for Acromyrmex; Table 1), largely separating the ant genera along the x-axes, confirming that Atta primarily focuses on tree-leaf material (compare panels C and D) and Acromyrmex on herbaceous leaves, flowers and (less pronounced) fruit. Comparison of the A and B panels illustrates that enzyme activity was generally higher in Acromyrmex (towards the right).

97 Discussion Genus-level niche segregation between Atta and Acromyrmex Although it is widely appreciated that Atta and Acromyrmex differ by more than two orders of magnitude in their scale of operations (see e.g. [14, 18, 26, 45]), systematic comparative studies similar to the present analyses have to our knowledge not been done. Although our snapshot results for the month of May cannot be generalized, we believe to have achieved our objective of demonstrating that larger scale studies like this can be done in principle. Our analyses illustrate that the statistical tools to analyze such data are available and can easily be expanded for use in more encompassing field surveys, as extra seasons and/or sampling sites can be added as new factors within which genera, species, and colonies can be nested. In the sections below, we offer tentative interpretations and compare them with available literature.

Our results confirm that the two genera of leaf-cutting ants operate at different scales, show that their foraging niches are different and that the enzymatic processing activities of fungus gardens appear to reflect these differences. The differences in foraging preferences quantified our intuitive expectations based on two decades of fieldwork in Gamboa, but the enzymatic activity differences were more substantial than we expected, because the two genera rear fungus-garden symbionts that belong to a single species L. gongylophorus [33]. This suggests that studies of phenotypic plasticity in enzyme gene expression will be worthwhile to enhance our understanding of the ver- satility of the leaf-cutting ant symbiosis. We will return to this in more detail below.

The finding that average genus-specific foraging rates show a 42 fold difference in loaded-worker return rates and a 63 fold difference in total worker return rates per hour, seems to match the ca. two order of magnitude difference in colony size between Atta and Acromyrmex. The fact that the differences do not quite reach 100 fold [14, 18, 26, 46] may be due to our primary focus on the largest Acromyrmex colonies (smaller colonies have too little foraging activity making the type of sampling that we did less feasible), whereas our selection of Atta colonies mostly contained medium size colonies. It is also conceivable that the Atta workers that returned to their nests without carrying plant material may have had their crops filled with plant sap as suggested by Littledyke & Cherrett [47], Quinlan & Cherret [48] and Hölldobler & Wilson [19], but the present setup did not allow any measurements on this. This suggests that considering only loaded workers may underestimate foraging effort, and that larger scale comparative studies should include sampling of liquid food in the crops of returning foragers. In spite of these limitations, we will also return to tentative inferences on species- and genus-level niche segregation that our snapshot data for Gamboa may allow.

98 Garden enzyme activity and forage material – is there a connection? It has long been known that the fungus is a major producer of decomposition enzymes for the plant material that leaf-cutting ant foragers provide, and recent work has shown that these decomposition services are supplemented by several other microorganisms that live in attine gardens [49, 50]. Other recent studies have emphasized that the expression of enzymes can be remarkably plastic and substrate dependent [21, 34]. This is consistent with earlier notions that there are active feedback loops between forager supply and symbiont demand, such that foragers may discard some forage material under specific conditions where its excess processing would not be optimal [51].

Our present results suggest that Atta and Acromyrmex represent ecologically distinct ant genera, both with regard to forage acquisition/diversity and garden enzyme activity/diversity. The latter confirms recent results showing differences in proteomes between Acromyrmex echinatior and Atta cephalotes fungus gardens [52]. It is remarkable that our results indicate that Atta gardens generally produce lower amounts of enzymes, even though these ants forage mostly on tree-leaves (Figure 3), which one would expect to be more demanding to decompose. It also appears that Atta gardens tend to overproduce two classes of enzymes, cellulases and pectinases, in addition to amylases (Figure 2), whereas Acromyrmex tended gardens produce higher amounts of all enzyme categories. This suggests that Atta gardens may somehow extract necessary nutrients more effi- ciently, but further work will be needed to understand the details of these processes. An additional factor to consider in this context is that Atta colonies produce conspicuous waste heaps or underground compost chambers, whereas this is rare for Acromyrmex (J.J. Boomsma & P.W. Kooij pers. obs.; CSE logbooks). This is consistent with Panamanian Atta discarding a much larger fraction of not fully degraded older fungus garden biomass than Acromyrmex [53, 54], perhaps because average enzyme activity per unit of fungus garden mass is lower and fresh tree leaves are more abundantly available than flower parts.

An earlier study [21] has hypothesized that Atta species focus on the rapid degradation of starch and proteins, but discard fungus garden material before most of the cellulose and hemicellulose is degraded. This is consistent with other recent studies showing that high amounts of cellulose and hemicellulose are still present in the bottom layer of the fungus gardens [50, 55, 56] and that only cellulases from L. gongylophorus remain highly active in this bottom layer [52, 57]. The larger scale and more wasteful substrate processing practiced by Atta may thus leave more substantial niches for additional bacterial and/or yeast [49] decomposition, similar to the domestication of specialized gut bacteria in large ungulates [58] that rely on residues of leaf-material that were

99 hard to digest even for ruminants. Focused, comparative transcriptomics to investigate conditional gene expression in fungus gardens of the two leaf-cutting ant genera could shed further light on possible differences of this kind and metagenome sequencing could identify the microbial communities involved, similar an earlier yeast study on the fungus gardens of Acromyrmex and Atta [49].

Niche partitioning in Panamanian Atta and Acromyrmex The data provided in our study are a snapshot of year-round foraging, which is known to vary across the seasons [25]. This implies that we cannot be sure that sampling in other seasons or at other sites would have yielded similar results. However, we also note that the five forage categories that we distinguished are very general and likely to be available throughout the year and that medium-size colonies of Atta and large colonies of Acromyrmex are unlikely to move over substantial distances (but see [59]), so their central place for foraging would tend to cover the same (sub) habitat over time. In spite of these caveats, our study shows that genus- and species-level differences across leaf-cutting ants can be quantified with the statistical tools we developed during this study.

The results of our study suggest that direct competition for forage material between the two genera of leaf-cutting ants is likely to remain limited, because Atta and Acromyrmex species target rather different types of forage, in spite of some overlap consistent with earlier reports that mostly report allopatrically collected data[18, 22, 24, 28-30, 32]. The correlations between garden enzyme activity and genus-level difference in forage use that we uncovered for the Gamboa community of leaf-cutting ants, may be reinforced or supplemented differences in salivary gland secretions between the two ant genera [60], a variable we were unable to measure. However, comparisons at species level suggested that both Atta and Acromyrmex species tend to have optimal habitats that are largely mutually exclusive, with Acromyrmex volcanus and Atta cephalotes foraging in the canopy, Acromyrmex octospinosus and (somewhat less specifically) Atta colombica foraging on the forest floor, and Acromyrmex echinatior and Atta sexdens foraging in the open landscape. Although it is possible that these differences are less pronounced in other seasons or sites, these results seem consistent with ecological theory predicting that interspecific competition is more pronounced when species are more similar, so that habitat partitioning may evolve [4-6].

The only case in which habitat segregation was somewhat less pronounced was between Atta sexdens and Atta colombica, which often overlap in park-like and man-made habitats. Although there is a clear gradient across the isthmus of Panama, Atta sexdens is the dominant Atta species

100 along the Pacific coast and becomes less abundant towards Gamboa in central Panama, whereas the pattern is opposite for Atta colombica [28]. It is interesting that these are the only two species for which we once observed neighboring trails and active avoidance behavior, i.e. trails of a colony stopping ca. one meter from the trail of another colony (P.W. Kooij, pers. obs.), behavior expected for all Atta spp. For the two common Panamanian Acromyrmex species, of which our research group has dug up ca. 500 colonies over the last two decades, habitat segregation (forest for A. octospinosus and open grassland areas for A. echinatior) is so pronounced that they will rarely encounter each other, similar to what is seen in Costa Rica [61]. In this way mature colonies of these species are unlikely to compete for the same type of plant forage. As far as we are aware distributions of incipient (founding) colonies are similar in Gamboa, but this is harder to quantify.

Acknowledgments We thank the Smithsonian Tropical Research Institute (STRI), Panama, for providing logistic help and facilities to work in Gamboa, the Autoridad Nacional del Ambiente y el Mar (ANAM) for permission to sample ant colonies in Panama, Anders Illum, Mattias Lange Nielsen, Sämi Schär and Nicholas Westwood for help with collecting the foraging data, Henrik De Fine Licht for supplying data on enzyme activity in an Acromyrmex volcanus colony and for helping to interpret the results, Gøsta Nachmann for suggesting the statistical analysis of the enzyme experiment, and the Danish National Research Foundation for funding (DNRF57). The collection of the observational data reported in this paper was facilitated by the Gamboa Graduate Course in Tropical Behavioral Ecology and Evolution, hosted by the Copenha- gen Centre for Social Evolution and the Smithsonian Tropical Research Institute.

References 1. Stephens DW, Krebs JR (1986) Foraging Theory. Princeton: Princeton University Press. 247 pp. 2. Rohde K (2006) Nonequilibrium Ecology. Cambridge: Cambridge University Press. 236 pp. 3. Dayan T, Simberloff D (2005) Ecological and community-wide character displacement: the next generation. Ecol Lett 8: 875–894. doi:10.1111/j.1461-0248.2005.00791.x. 4. Florencio M, Gómez-Rodríguez C, Serrano L, Díaz-Paniagua C (2013) Competitive exclusion and habitat segregation in seasonal macroinvertebrate assemblages in temporary ponds. Freshwater Sci 32: 650–662. doi:10.1899/12-105.1. 5. Guo Z, Liu J, Lek S, Li Z, Ye S, et al. (2012) Habitat segregation between two congeneric and introduced goby species. Fund Appl Limnol 181: 241–251. doi:10.1127/1863-9135/2012/0397. 6. Ortega S (1987) Habitat segregation and temporal variation in some tropical intertidal populations. J Exp Mar Biol Ecol 113: 247–265. 7. Ali JG, Agrawal AA (2012) Specialist versus generalist insect herbivores and plant defense. Trends Plant Sci 17: 293–302. doi:10.1016/j.tplants.2012.02.006. 8. Bernays EA, Minkenberg O (1997) Insect herbivores: different reasons for being a generalist. Ecology 78: 1157–1169. 9. Houseman RM, Gold RE, Pawson BM (2001) Resource partitioning in two sympatric species of subterra nean termites, Reticulitermes flavipes and Reticulitermes hageni (Isoptera: Rhinotermitidae). Environ Ento mol 30: 673–685. doi:10.1603/0046-225X-30.4.673.

101 10. Kambara K, Takematsu Y (2009) Field habitat selection of two coexisting species of Reticulitermes, R. speratus and R. kanmonensis (Isoptera, Rhinotermitidae). Sociobiology 54: 65–75. 11. Schultz TR, Brady SG (2008) Major evolutionary transitions in ant agriculture. P Natl Acad Sci USA 105: 5435–5440. doi:10.1073/pnas.0711024105. 12. Farji-Brener AG (2000) Leaf-cutting ant nests in temperate environments: mounds, mound damages and nest mortality rate in Acromyrmex lobicornis. Stud Neotrop Fauna E 35: 131–138. 13. Mayhé-Nunes AJ, Jaffe K (1998) On the biogeography of Attini (Hymenoptera: Formicidae). Ecotropicos 11: 45–54. 14. Weber NA (1966) Fungus-growing ants. Science 153: 587–604. 15. Fowler HG, Pagani MI, Da Silva OA, Forti LC, Da Silva VP, et al. (1989) A pest is a pest is a pest? The dilemma of Neotropical leaf-cutting ants: keystone taxa of natural ecosystems. Environ Manage 13: 671–675. 16. Bass M, Cherrett JM (1995) Fungal hyphae as a source of nutrients for the leaf-cutting ant Atta sexdens. Physiol Entomol 20: 1–6. 17. Cherrett JM, Peregrine D (1976) A review of the status of leaf-cutting ants and their control. Ann Appl Biol 84: 124–128. 18. Mehdiabadi NJ, Schultz TR (2010) Natural history and phylogeny of the fungus-farming ants (Hymenop- tera: Formicidae: Myrmicinae: Attini). Myrmecol 13: 37–55. 19. Hölldobler B, Wilson EO (2011) The leafcutter ants. New York, London: W. W. Norton & Company, Inc. 160 pp. 20. Weber NA (1958) Evolution in fungus-growing ants. Proceedings of the Tenth International Congress of Entomology, Vol. 2. pp. 459–473. 21. De Fine Licht HH, Schiøtt M, Mueller UG, Boomsma JJ (2010) Evolutionary transitions in enzyme activity of ant fungus gardens. Evolution 64: 2055–2069. doi:10.1111/j.1558-5646.2010.00948.x. 22. De Fine Licht HH, Boomsma JJ (2010) Forage collection, substrate preparation, and diet composition in fungus-growing ants. Ecol Entomol 35: 259–269. doi:10.1111/j.1365-2311.2010.01193.x. 23. Weber NA (1972) The fungus-culturing behavior of ants. Am Zool 12: 577–587. 24. Cherrett JM (1968) The Foraging Behaviour of Atta cephalotes L. (Hymenoptera, Formicidae) I. Foraging pattern and plant species attacked in tropical rain forest. J Anim Ecol 37: 387–403. 25. Wirth R, Herz H, Ryel RJ, Beyschlag W, Hölldobler B (2003) Herbivory of leaf-cutting ants: a case study on Atta colombica in the tropical rainforest of Panama. Berlin, Heidelberg, New York: Springer-Verlag. 230 pp. 26. Wetterer JK (1995) Forager size and ecology of Acromyrmex coronatus and other leaf-cutting ants in Costa Rica. Oecologia 104: 409–415. 27. Mikheyev AS, Mueller UG, Boomsma JJ (2007) Population genetic signatures of diffuse co-evolution between leaf-cutting ants and their cultivar fungi. Mol Ecol 16: 209–216. doi:10.1111/j.1365- 294X.2006.03134.x. 28. Weber NA (1969) Ecological relations of three Atta species in Panama. Ecology 50: 141–147. 29. Berish C (1986) Leaf-cutting ants (Atta cephalotes) select nitrogen-rich forage. Am Midl Nat 115: 268–276. 30. Cherrett JM (1972) Some factors involved in the selection of vegetable substrate by Atta cephalotes (L.) (Hymenoptera: Formicidae) in Tropical Rain Forest. J Anim Ecol 41: 647–660.

102 31. Farji-Brener AG (2001) Why are leaf-cutting ants more common in early secondary forests than in old- growth tropical forests? An evaluation of the palatable forage hypothesis. Oikos 92: 169–177. 32. Howard JJ (1988) Leafcutting and diet selection: relative influence of leaf chemistry and physical features. Ecology 69: 250–260. 33. Mikheyev AS, Mueller UG, Abbot P (2010) Comparative dating of attine ant and lepiotaceous cultivar phylogenies reveals coevolutionary synchrony and discord. Am Nat 175: E126–E133. doi:10.1086/652472. 34. Kooij PW, Schiøtt M, Boomsma JJ, De Fine Licht HH (2010) Rapid shifts in Atta cephalotes fungus-garden enzyme activity after a change in fungal substrate (Attini, Formicidae). Insect Soc 58: 145–151. doi:10.1007/s00040-010-0127-9. 35. Ibáñez R, Condit RS, Angehr G, Aguilar S, García T, et al. (2002) An ecosystem report on the Panama Canal: monitoring the status of the forest communities and the watershed. Environ Monit Assess 80: 65–95. 36. R Core Team (2013) R: A language and environment for statistical computing. Available: http://cran.r-project.org/ 37. Pinheiro J, Bates D, DebRoy S, Sarkar D, R Core Team (2013) nlme: Linear and nonlinear mixed effects models. Available: http://cran.r-project.org/web/packages/nlme/index.html 38. Hothorn T, Bretz F, Westfall P (2008) Simultaneous inference in general parametric models. Biometrical J 50: 346–363. 39. Warnes GR, Ben Bolker, Bonebakker L, Gentleman R, Liaw WHA, et al. (2012) gplots: Various R program ming tools for plotting data. Available: http://cran.r-project.org/web/packages/gplots/index.html 40. Suzuki R, Shimodaira H (2011) pvclust: Hierarchical Clustering with P-Values via Multiscale Bootstrap Resampling. Available: http://cran.r-project.org/package=pvclust. 41. Rønhede S, Boomsma JJ, Rosendahl S (2004) Fungal enzymes transferred by leaf-cutting ants in their fungus gardens. Mycol Res 108: 101–106. doi:10.1017/S0953756203008931. 42. Kacuráková M, Capek P, Sasinková V, Wellner N, Ebringerová A (2000) FT-IR study of plant cell wall model compounds: pectic polysaccharides and hemicelluloses. Carbohyd Polym 43: 195–203. doi:10.1016/ S0144-8617(00)00151-X. 43. Husson F, Josse J, Le S, Mazet J (2013) FactoMineR: multivariate exploratory data analysis and data mining with R. Available: http://cran.r-project.org/package=FactoMineR. 44. De Vasconcelos HL (1990) Foraging activity of two species of leaf-cutting ants (Atta) in a primary forest of the central amazon. Insect Soc 37: 131–145. doi:10.1007/BF02224026. 45. Villesen P, Murakami T, Schultz TR, Boomsma JJ (2002) Identifying the transition between single and multiple mating of queens in fungus-growing ants. P R Soc B 269: 1541–1548. doi:10.1098/rspb.2002.2044. 46. Murakami T, Higashi S, Windsor D (2000) Mating frequency, colony size, polyethism and sex ratio in fungus-growing ants (Attini). Behav Ecol Sociobiol 48: 276–284. 47. Littledyke M, Cherrett JM (1976) Direct ingestion of plant sap from cut leaves by the leaf-cutting ants Atta cephalotes (L.) and Acromyrmex octospinosus (Reich)(Formicidae, Attini). B Entomol Res 66: 205–217. 48. Quinlan R, Cherrett JM (1979) The role of fungus in the diet of the leaf-cutting ant Atta cephalotes (L.). Ecol Entomol 4: 151–160. 49. Mendes TD, Rodrigues A, Dayo-Owoyemi I, Marson FAL, Pagnocca FC (2012) Generation of nutrients and detoxification: possible roles of yeasts in leaf-cutting ant nests. Insects 3: 228–245. doi:10.3390/insects3010228.

103 50. Suen G, Scott JJ, Aylward FO, Adams SM, Tringe SG, et al. (2010) An insect herbivore microbiome with high plant biomass-degrading capacity. PLoS Genet 6. doi:10.1371/journal.pgen.1001129. 51. North RD, Jackson CW, Howse PE (1999) Communication between the fungus garden and workers of the leaf-cutting ant, Atta sexdens rubropilosa, regarding choice of substrate for the fungus. Physiol Entomol 24: 127–133. doi:10.1046/j.1365-3032.1999.00122.x. 52. Aylward FO, Burnum-Johnson KE, Tringe SG, Teiling C, Tremmel DM, et al. (2013) Leucoagaricus gongylophorus produces diverse enzymes for the degradation of recalcitrant plant polymers in leaf-cutter ant fungus gardens. Appl Environ Microb 79: 3770–3778. doi:10.1128/AEM.03833-12. 53. Bot A, Currie CR, Hart AG, Boomsma JJ (2001) Waste management in leaf-cutting ants. Ethol Ecol Evol 13: 225–237. 54. Hart AG, Ratnieks FLW (2002) Waste management in the leaf-cutting ant Atta colombica. Behav Ecol 13: 224–231. 55. Moller IE, De Fine Licht HH, Harholt J, Willats WGT, Boomsma JJ (2011) The dynamics of plant cell- wall polysaccharide decomposition in leaf-cutting ant fungus gardens. PLoS ONE 6: e17506. doi:10.1371/ journal.pone.0017506. 56. Nagamoto NS, Garcia MG, Forti LC, Verza SS, Noronha NC, et al. (2011) Microscopic evidence supports the hypothesis of high cellulose degradation capacity by the symbiotic fungus of leaf-cutting ants. J Biol Res-Thessalon 16: 308–312. 57. Grell MN, Nielsen KL, Linde T, Nygaard S, Boomsma JJ, et al. (in press) The fungal symbiont of leaf-cutter ants expresses the full enzyme complement to degrade cellulose and other plant polymers. BMC Genomics. 58. Tyrrell HF, Moe PW (1975) Effect of intake on digestive efficiency. J Dairy Sci 58: 1151–1163. 59. Porter SD, Bowers MA (1980) Emigration of an Atta colony. Biotropica 12: 232–233. 60. Crewe RM, Blum MS (1972) Alarm pheromones of the Attini: their phylogenetic significance. J Insect Physiol 18: 31–42. 61. Wetterer JK, Gruner D, Lopez J (1998) Foraging and nesting ecology of Acromyrmex octospinosus (Hyme- noptera : Formicidae) in a Costa Rican tropical dry forest. Fla Entomol 81: 61–67.

104 Table S1. Sample sizes for each colony. 41 15 10 25 21 21 21 24 25 62 14 65 23 14 23 42 50 73 19 43 43 265 135 458 Number of ants counted Number 30 30 30 15 30 30 30 30 30 30 30 30 30 30 30 30 15 30 30 30 30 30 285 285 Minutes observation per trail 30 30 30 15 30 30 30 30 30 30 30 30 30 30 30 30 15 30 30 30 30 30 285 285 Minutes observation per colony 30 30 285 285 285 285 Minutes observation per species 30 30 30 15 30 30 30 30 30 30 30 30 15 30 30 30 30 30 30 30 30 30 285 285 Observation time (min) 2:55 PM 3:35 PM 4:10 PM 1:00 PM 2:00 PM 3:00 PM 3:00 PM 2:15 PM 9:40 AM 9:55 AM 9:45 AM 9:10 AM 10:30 AM 11:30 AM 11:30 AM 10:20 AM 10:30 AM 11:00 AM 10:15 AM 10:45 AM 11:00 AM TOTAL TOTAL TOTAL Time of day 51 17 18 32 28 29 16 30 40 31 50 14 15 42 43 44 45 46 48 49 47 Trail Av1 Ae1 Ae2 Ae3 Ae4 Ae5 Ae6 Ae7 Ae8 Ae9 Ao1 Ao2 Ao3 Ao4 Ao5 Ao6 Ao7 Ao8 Ao9 Ae10 Ao10 Colony Species Acromyrmex volcanus Acromyrmex Acromyrmex echinatior Acromyrmex Acromymrex octospinosus Acromymrex

105 48 60 53 50 24 48 43 49 71 90 68 82 94 91 91 42 43 45 99 72 98 158 1519 Number of ants counted Number 4 4 2 2 2 2 2 2 2 2 2 26 Minutes observation per trail 4 4 4 6 6 2 26 Minutes observation per colony 26 26 Minutes observation per species 2 2 2 2 2 2 2 2 2 2 26 0.5 0.5 0.5 0.5 0.5 0.5 0.5 0.5 0.5 0.5 0.5 0.5 Observation time (min) 3:00 PM 3:00 PM 3:00 PM 5:00 PM 5:00 PM 5:00 PM 10:30 PM 10:00 PM 10:00 PM 10:30 PM 10:30 PM 10:30 PM 10:00 PM 10:00 PM 10:00 PM 11:15 AM 11:15 AM 10:50 AM 10:50 AM 10:50 AM 10:50 AM 10:30 AM Time of day TOTAL 7 8 23 22 26 27 21 25 13 12 11 Trail Ace3 Ace4 Ace5 Ace2 Ace1 Ace6 Colony Species Atta cephalotes Atta

106 801 5 7 69 20 88 53 72 81 61 22 39 66 58 72 29 97 94 99 87 95 98 105 172 107 103 118 263 102 113 130 1724 Number of ants counted Number 22 4 2 2 2 2 2 2 2 2 2 4 4 4 4 2 2 2 2 0.5 24.5 Minutes observation per trail 22 4 2 2 2 2 2 8 8 4 4 4 2 2 0.5 24.5 Minutes observation per colony 22 22 24.5 24.5 Minutes observation per species 22 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 0.5 0.5 0.5 0.5 0.5 0.5 0.5 0.5 0.5 24.5 Observation time (min) 3:15 PM 3:40 PM 4:15 PM 3:00 PM 4:00 PM 1:30 PM 3:50 PM 4:00 PM 4:00 PM 4:00 PM 4:00 PM 4:00 PM 4:00 PM 4:00 PM 4:00 PM 9:30 AM 9:30 AM 9:30 AM 9:00 AM 9:15 AM 9:15 AM 12:30 PM 11:45 AM 11:45 AM 10:00 AM 10:00 AM 10:30 AM 10:30 AM 10:30 AM 10:30 AM TOTAL TOTAL Time of day 4 9 2 1 3 6 5 34 35 38 39 41 10 19 20 24 37 33 36 Trail As3 As4 As5 As6 As7 As1 As2 Acol1 Acol2 Acol3 Acol4 Acol7 Acol5 Acol6 Colony Species Atta sexdens Atta Atta colombica

107 108 Leucoagaricus gongylophorus uses leaf-cutting ants to vector proteolytic enzymes towards new plant substrate

Pepijn W Kooij, Adelina Rogowska-Wrzesinska, Daniel Hoffmann, Peter Roepstorff,

Jacobus J Boomsma, Morten Schiøtt

(In press at the ISME Journal) Chapter 4 109 Leucoagaricus gongylophorus uses leaf-cutting ants to vector proteolytic enzymes towards new plant substrate

Running title: Protease activity in ant fungus gardens

Pepijn W Kooij*a, Adelina Rogowska-Wrzesinskab, Daniel Hoffmanna, Peter Roepstorffb, Jacobus J Boomsmaa, Morten Schiøtt*a

aCentre for Social Evolution, Department of Biology, University of Copenhagen, Universi- tetsparken 15, DK-2100 Copenhagen, Denmark bProtein Research Group, Department of Biochemistry and Molecular Biology, University of Southern Denmark, Campusvej 55, DK-5230 Odense, Denmark.

*Corresponding authors: [email protected] and [email protected] at the above address

Accepted for publication at The ISME Journal, December 2013

110 Abstract The mutualism between leaf-cutting ants and their fungal symbionts revolves around processing and inoculation of fresh leaf pulp in underground fungus gardens, mediated by ant fecal fluid deposited on the newly added plant substrate. As herbivorous feeding often implies that growth is nitrogen-limited, we cloned and sequenced six fungal proteases found in the fecal fluid of the leaf-cutting ant Acromyrmex echinatior and identified them as two metalloendoproteases, two serine proteases and two aspartic proteases. The metalloendoproteases and serine proteases showed significant activity in fecal fluid at pH values of 5-7, but the aspartic proteases were inactive across a pH range of 3-10. Protease activity disappeared when the ants were kept on a sugar water diet without fungus. Relative to normal mycelium, both metalloendoproteases, both serine proteases and one aspartic protease were upregulated in the gongylidia, specialized hyphal tips whose only known function is to provide food to the ants. These combined results indicate that the enzymes are derived from the ingested fungal tissues. We infer that the five proteases are likely to accelerate protein extraction from plant cells in the leaf pulp that the ants add to the fungus garden, but regulatory functions such as activation of proenzymes are also possible, particularly for the aspartic proteases that were present but without showing activity. The proteases had high sequence similarities to proteolytic enzymes of phytopathogenic fungi, consistent with previous indications of convergent evolution of decomposition enzymes in attine ant fungal symbionts and phytopathogenic fungi.

Keywords: Acromyrmex echinatior/nutrition/mutualism/phytopathogens/proteases

111 Introduction Obligate symbioses are characterized by high degrees of partner commitment and functional complementarity (Janzen 1985). Such interactions normally show only minor remnants of the potential reproductive conflicts that needed to be overcome when these partnerships evolved, allowing them to ultimately become single adaptive units with organismal properties (Queller & Strassmann 2009; Boomsma 2013). Such high levels of interdependency are most characteristic for endosymbioses where the symbiont lives in cells or tissues of the host, but may also occur in ectosymbioses like the eusocial fungus-farming mutualisms of ants and termites where the fungus gardens are ectosymbionts for the individual ants but endosymbionts for the colony (Poulsen & Boomsma 2005; Aanen et al. 2009). We are intuitively inclined to consider the insect colony as the active farming host and the fungus gardens as passive crops, but the contention that fungal clones have contracted ant families to propagate them across generations appears to be equally valid, especially in the attine ants where the fungal symbiont is vertically transmitted (Poulsen & Boomsma 2005).

In terms of biomass conversion efficiency, the functional complementarity between attine ants and their fungus garden symbionts achieved spectacular progression, moving from initial stages of ant-driven litter-based decomposition farming that evolved 50 Million Years Ago (MYA) to advanced farming of irreversibly domesticated crop fungi 20 MYA (Mueller et al. 1998; Schultz & Brady 2008; Mehdiabadi & Schultz 2010), culminating in aggressive herbivorous leaf-cutting farming 10 MYA (De Fine Licht et al. 2010; Schiøtt et al. 2010). Particularly the latter transition allowed the ants to evolve much larger and complex societies (Mueller et al. 1998; Villesen et al. 2002; Schultz & Brady 2008; Fernandez-Marin et al. 2009; Mehdiabadi & Schultz 2010) after they had managed to overcome a series of challenges that are associated with a herbivorous lifestyle (Schiøtt et al. 2008; 2010; De Fine Licht et al. 2010; 2013). Although depending on live vegetation rather than dead leaf litter undoubtedly implied better access to the proteins that normally limit insect growth, an almost exclusive fungal diet is generally poorer in protein than the carnivo- rous diets of many hunter-gatherer species in the Myrmicinae subfamily of ants (Davidson 2004). Hosts and symbionts can thus be assumed to have been under consistent selection to maximize protein yield as this will determine colony growth rate and the production of winged virgin queens that transmit both their own genes and clonal copies of the fungal symbiont to future generations.

Specialized foraging on live plant material as growth substrate for the fungal symbiont Leucoag- aricus gongylophorus implies that leaf-cutting ants are major herbivores in the Neo(sub)tropics with substantial roles in recycling nitrogen and phosphorus (Fowler et al. 1989). A recent study

112 has confirmed that the leaf-cutting fungus-farming mutualism is likely to be nitrogen-limited as Atta fungus gardens harbour substantial levels of nitrogen fixing bacteria (Pinto-Tomás et al. 2009). Studies on the other genus of leaf-cutting ants, Acromyrmex, have shown that the fungal symbiont uses the ants to vector pectinases and a laccase from the most productive middle layers of fungus gardens to the top where they are most useful for the symbiosis but where fungal biomass to produce these enzymes is in short supply (Schiøtt et al. 2010; De Fine Licht et al. 2013). The fungal gongylidia, special hyphal tips produced only to feed the ants and their brood (Bass & Cherrett 1995), are instrumental in this process as they upregulate the expression of genes producing these enzymes (Schiøtt et al. 2010; De Fine Licht et al. 2013). These studies indicate that the fungal symbiont uses the ants for vector enzymes for the neutralization of phenolic compounds (laccase) and for loosening up the cell wall matrix (pectinases) to new garden sections. However, whether these initial decomposition steps are followed by a similar boost in protease activity to decompose intracellular proteins has remained unknown.

The ants process freshly cut leaf fragments into a mass of leaf pulp while mixing it with droplets of fecal fluid and subsequently deposit the new substrate on fungal ridges in the top of the garden where they inoculate it with small tufts of mycelium from the older part of the garden (Weber 1966). In the 1970’s, leaf-cutting ant fecal fluid was shown to contain active proteases (Martin 1970; Martin & Martin 1970a; Martin & Martin 1970b; Martin & Martin 1971; Martin 1974) with similar chemical properties as enzymes originating from the fungal symbiont (Boyd & Martin 1975a; 1975b), and recent work has shown that fungus garden endo-protease activity in evolutionarily derived leaf-cutting ants is much higher than in sister ant lineages that do not use fresh leaves to make their gardens grow (De Fine Licht et al. 2010), activity that could subsequently be assigned to metalloproteases and serine proteases (Semenova et al. 2011).

Recently, we obtained the proteome of Acromyrmex echinatior fecal fluid and were able to identify 33 proteins. Among these were seven pectinases (Schiøtt et al. 2010), one laccase (De Fine Licht et al. 2013), and seven proteases. Only a single of these proteases is known to be produced by the ants, whereas the other six are contributed by the fungal symbiont. We hypothesized that a number of these proteases might serve to either process the intracellular proteins of freshly opened plant cells or to assist in the breakdown of secondary plant defences (Christeller et al. 1992; Mer- cado-Flores et al. 2003; Ievleva et al. 2006). The latter is a distinct possibility as plants are known to use protease inhibitors to counteract the harmful effect of proteases produced by phytopathogic fungi and to generally decrease the digestibility of their tissues to discourage herbivores (Habib & Fazili 2007). The objectives of the present study were to: 1. measure the specific protease activity

113 in fecal fluid of gardeningA. echinatior workers, 2. confirm that they are derived from the fungal symbiont and have different expression levels and pH optima, 3. assess the extent to which the expression of genes coding for these enzymes was enhanced in the fungal gongylidia, as would be expected when vectoring these enzymes to the top of fungus gardens is a specific adaptation of the symbiosis, and 4. discuss the implications of our results for understanding the co-adaptations between partners that has allowed this symbiosis to evolve its substantial ecological footprint in the Neo(sub)tropics.

Materials and methods Biological material We used seven colonies of A. echinatior (Ae263, Ae280, Ae322, Ae332, Ae335, Ae349 and Ae370), collected in and around Gamboa, Panama, between 2004 and 2007 and kept in rearing facilities in Copenhagen under controlled conditions of ca. 25°C and ca. 70% humidity, where they were fed twice a week with fresh bramble leaves, apple pieces and dry rice. Fecal fluid was obtained by gently squeezing the abdomen of large workers with a forceps on a microscope slide. Each fecal droplet was then mixed with 0.5 µl demineralised water, collected with a micropipette, and stored in an Eppendorf tube on ice. Sixty droplets from five colonies each (Ae263, Ae280, Ae322, Ae332 and Ae349) were collected this way, pooled per colony and diluted with demineralised water to a final volume of 250 µl.

For the gene expression measurements, gongylidia clusters (staphylae) and normal mycelium were collected separately under a stereomicroscope at 40x magnification from each of five colonies (Ae263, Ae280, Ae322, Ae335 and Ae370) in 2 mL Eppendorf tubes floating in liquid nitrogen. After collecting app. 100 µL for each type of tissue, samples were stored at -80°C for subsequent RNA extraction.

Protein identification and gene cloning SDS-PAGE and mass spectrometry were performed as described previously (Schiøtt et al. 2010; De Fine Licht et al. 2013). RNA extraction from the fungal symbiont, followed by cDNA con- struction, was done as in Schiøtt et al. (2010). Aspartic protease 2 (AspII), Metallopeptidase 1 (MetI), Serine protease 1 (SerI) and Serine protease 2 (SerII) were initially identified by PCR amplification from cDNA using degenerate primers constructed from the mass spectrometry data using the PCR scheme: one cycle of 95°C for 5 min, 35 cycles of 94°C for 20 sec, 50°C for 30 sec and 72°C for 2 min, and ending with one cycle of 72°C for 7 min. Aspartic protease 1 (AspI) and Metallopeptidase 2 (MetII) were identified at a later stage by a Blast search of the ca. 100x

114 coverage A. echinatior genome (Nygaard et al. 2011) and a low coverage genome sequence of the A. echinatior fungal symbiont (De Fine Licht et al. 2013) using the mass spectrometry data as queries. Sequencing of full-length gene transcripts was done using a RACE (Rapid Amplification of cDNA Ends) strategy. 3’ and 5’-RACE libraries were made from ca. 1 µg of purified RNA with the SMART RACE cDNA kit (CLONTECH, Mountain View, California, USA), and gene sequences were PCR amplified from these libraries using specific primers designed from the PCR amplified gene fragments (AspII, MetI, SerI and SerII) or BLAST search identified sequence reads (AspI and MetII) along with the primers enclosed in the SMART RACE cDNA kit. In some cases the 5’RACE had to be done in several steps to get a full transcript sequence. The following PCR scheme was used in the RACE experiments: one cycle of 95°C for 5 min, 10 cycles of 94°C for 20 sec, 72°C for 30 sec (with a decrease in temperature of 0.5°C in every cycle) and 72°C for 2 min, followed by 35 cycles of 94°C for 20 sec, 67°C for 30 sec and 72°C for 2 min, and ending with one cycle of 72°C for 7 min. All PCR products were cloned in pCR4-TOPO using the TOPO TA cloning method (Invitrogen, Carlsbad, California, USA) and sequenced at Eurofins MWG Op- eron (Ebersberg, Germany). Gene sequences are deposited in GenBank with accession numbers KF571927-KF571932. Primer sequences for cloning of the genes are provided in Table S1.

Enzyme assays Protease activity was measured for four different protease inhibitors and two controls (without inhibitors, one of which incubated and the other not incubated) at eight different pH levels. The inhibitors used in the experiment were 5mM 1,10-phenanthroline to inhibit metalloendoproteases, 2µM pepstatin to inhibit aspartic proteases, 10mM phenylmethylsulfonyl fluoride to inhibit serine proteases and 10µM trans-epoxysuccinyl-L-leucylamido(4-guanidino)butane to inhibit cysteine proteases. In the assays, 4 µl of the diluted fecal fluid was mixed with 1 µl of inhibitor or distilled demineralised water and incubated on ice for one hour. Next, 35 µl of 0.1M Britton Robinson buffer (Britton & Robinson 1931) with 2% azoalbumin was added and the mix was incubated at 30°C for one hour. For the blank control (not incubated) the azoalbumin mix was added at the last moment without incubation to prevent any protease activity to occur. To stop the activity of the proteases, 120 µl 10% trichloroacetic acid was added, after which the mix was vortexed and incubated for 15 minutes before centrifugation for 5 minutes at 8000g. The supernatant (86 µl) was added to 100 µl freshly prepared 1M NaOH on a 96 well cell culture plate.

Absorbance was measured at 440 nm on a Versamax ELISA microplate reader (Molecular Devic- es, Sunnyvale, USA) and scored after subtracting the corresponding blank-control measurements to correct for background absorbance. To find the enzyme activity in each of the four protease

115 classes, measurements obtained with specific inhibitors for each protease class were subtracted from the total protease activity in identical assays without addition of inhibitors. Protease activity was defined as the amount of enzyme required to cause a unit increase at 440 nm across a 1 cm path length (Sarath et al. 2001). The number of units of protease activity data were plotted and analyzed with R (R Core Team 2013), using Linear Mixed-Effects Models (“lme”) (Pinheiro et al. 2013), with Colony as random factor, and generating p-values with General Linear Hypoth- eses (“glht”) in the Simultaneous Inference in General Parametric Models package “multcomp” (Hothorn et al. 2008).

For measuring protease activity in control ants, 100-200 workers of the colonies Ae322, Ae332 and Ae349 were transferred to subcolonies with no access to fungus garden material, but with an ad libitum diet of 10% sucrose and bramble leaves. After ca. two weeks protease activity was measured in pooled samples of fecal droplets from two ants for each colony at pH 6 in three replicates per colony.

Quantitive real-time PCR Primers for the six different genes (Table S2) were designed by matching the obtained cDNA sequences to a database of a partially sequenced genome of the A. echinatior fungal symbiont (De Fine Licht et al. 2013) using BLASTn to identify intron and exon sequences. Primers were designed to span an intron to avoid amplification of genomic DNA. The qPCRs were run on an Mx3000P QPCR System (Agilent, Santa Clara, CA, USA) in 20 µL reactions (0.5 µL cDNA, 10 µL 2x SYBR Premix Ex Taq [TaKaRa Bio Inc., Otsu, Japan] and 0.4 µL of each primer [10µM]) with the following PCR conditions: one cycle of 95°C for 2 min, followed by 40 cycles of 95°C for 30 sec, 55°C (AspI, AspII, MetII, SerI, SerII, and EF1-α) or 57°C (MetI, GAPDH, Ubc) for 30 sec, and 72°C for 30 sec, and ending with a melting curve cycle of 95°C for 30 sec, 55°C or 57°C for 30 sec and 95°C for 30 sec.

For each of the ten samples (one gongylidia and one mycelium sample from five colonies), three technical qPCR replicates were done for nine genes: three housekeeping genes (GAPDH, Ubc and EF1-α) and six target genes (MetI, MetII, AspI, AspII, SerI and SerII). All Ct values from the RT-qPCR analysis were analyzed in R with packages “ReadqPCR” (Perkins & Kohl 2011) and “NormqPCR” (Kohl & Perkins 2011) using averages across the three technical replicates. Stabili- ty of housekeeping genes was assessed across the entire data set, which showed that GAPDH and Ubc were more stable than EF1-α, so the latter gene was discarded in the further analyses. GAP- DH and Ubc were then used to calculate the normalized expression (2∆Ct) of the target genes in

116 Table 1 Peptide sequences of proteases in fecal fluid of A. echinatior and corresponding molecular weights (Da) of full-length proteins. The letter J symbolises the amino acids leucin and isoleucin, which have the same molecular mass and are therefore indiscernible by mass spectrometry. The single discrepancy between mass spectrometry data and sequence data (letters D and H in bold) may have been caused by deamination of asparagine into aspartic acid.

Protein Mass spectrometry data Sequence data MW (Da)

Aspartic protease 1 DENDGGEATFGGJNPSSYR DENDGGEATFGGINPSSYR 44788

NAYWEV NAYWEV

YFTVYDJGR YFTVYDLGR

Aspartic protease 2 FTGAJNFTPR FTGALNFTPR 43158

SSGFEGTDGJJGJGPVDJTR SSGFEGTDGILGIGPVDLTR

GTJSPAVNSJVPTVTDNJFSSGR GTLSPAVNSLVPTVTDNLFSSGR

TSTSPSSJFWGJNQSVR TSTSPSSLFWGLNQSVR

VJDNNTGJJR VLDNNTGLLR

Metallopeptidase 1 TYASNAATYJSSHSSSSTR TYASNAATYLSSHSSSSTR 37207

YTTWFGTYTSAR YTTWFGTYTSAR

Metallopeptidase 2 TYASNAJJYJR TYASNALIYLR 37511

Serine protease 1 AWAGJHFAVAAGNDNR AADAGLHFAVAAGNDNR 50114

GSDGSGSTSDVJAGVQWAAR GSDGSGSTSDVIAGVQWAAR

Serine protease 2 ADJQTFFR ADLQTFFR 63388

AGYJDEFANR AGYLDEFANR

AJYNTVNYVPSQTSR ALYNTVNYVPSQTSR

DGJGVAGYJDEFANR NGLGVAGYLDEFANR

GVAGYJDEFANR GVAGYLDEFANR

GSVGGTSASSPTVAGVFAJJNDFR GSVGGTSASSPTVAGVFALLNDFR

FRPDAAGSSFTTVR FRPDAAGSSFTTVR

YFSTPSYQSAAVSR YFSTPSYQSAAVSR

Carboxypeptidase A QNNPGVFFESGJHAR QNNPGVFFESGIHAR 46943 (EGI65848)

117 gongylidia and mycelium. Values of 2-∆∆Ct (Livak & Schmittgen 2001) for each gene subsequently produced estimates of fold changes in relative gene expression between gongylidia and mycelium. This sequence of procedures allowed us to identify genes with significantly different expression levels and to obtain estimates of their normalized and relative expression.

Results The mass spectrometry data of fecal fluid proteins (Table 1) described previously (Schiøtt et al. 2010) allowed us to clone and sequence six protease genes encoded by the fungal symbiont genome, and to identify an ant encoded protease (EGI65848) belonging to a family of M14 carboxy- peptidases, which is significantly expanded in the genome of A. echinatior (Nygaard et al. 2011). Comparison with the MEROPS database release 9.8 (http://merops.sanger.ac.uk/ (Rawlings et al. 2012)) (Table S3) showed that all six proteases (AspI, AspII, MetI, MetII, SerI and SerII) were similar to peptidases known from other fungi; AspI with saccharopepsin (A01.018) from Laccar- ia bicolor (78.57%, E-value = 1.10e-142); AspII with polyporopepsin (A01.019) from L. bicolor (71.60%, E-value = 1.80e-125); MetI with peptidyl-Lys metallopeptidase (M35.004) from Pleuro- tus ostreatus (56.55%, E-value = 1.50e-53); MetII with peptidyl-Lys metallopeptidase (M35.004) from P. ostreatus (69.23%, E-value = 2.60e-42); SerI with an unassigned peptidase from the S8A subfamily from L. bicolor (61.28%, E-value = 6.90e-111); SerII with grifolisin (S53.010) from L. bicolor (66.67%, E-value = 4.40e-136).

To test the overall protease activity in the ant fecal fluid and to determine whether the fungus garden is responsible for this activity, protease activity was measured in fecal droplets of worker ants taken directly from their colonies and from purged ants that had been kept on sugar water and bramble leaves for ca. two weeks with no access to fungal garden material. This showed that protease activity is present in fecal fluid when the ants have their normal diet, but disappears almost completely when the ants are deprived of fungus garden material (t17 = 14.8972, p < 0.0001), indicating that the protease activity in the ant fecal fluid is caused by ingested proteases from the fungal symbiont and not by proteases produced by the ants themselves (Figure 1).

Next, we tested the activity of the four main types of proteases using different inhibitors specifically targeted at each category of proteases: metalloendoproteases, serine proteases, aspartic proteases and cysteine proteases. Only the metalloendoproteases and serine proteases had significant activities, with respective peaks at pH 6 (z = -10.824, p < 0.0001) and pH 7 (z = 4.654, p < 0.0001) and serine protease activity also being significantly enhanced at pH 6 (z = 2.004, p < 0.05), consistent with fecal fluid also having pH 6 (Figure 2). In contrast to serine proteases,

118 Figure 1 Enzyme activity in milliunits of proteases (mU) in fecal fluid of workers of the leaf-cutting ant A. echinatior kept on natural fungus-garden-diet and on a diet of only sugar

water and possibly plant sap as control (t17 = 14.8972, p < 0.0001). Whiskers are SEs of the mean.

metalloendoproteases were active at a very broad range of pH values and retained activities up to pH 10. We found no activity for aspartic proteases (present in fecal fluid), and cysteine proteases (absent in fecal fluid).

Relative changes in gene expression in gongylidia versus mycelium (fold-changes: 2-∆∆Ct) showed that the expression of five out of the six fungal protease genes was upregulated in the gongylidia compared to the mycelium (Figure 3a): peptidyl-Lys metallopeptidase I and II (MetI and MetII) (z = -3.979, p = 0.00004; z = -20.010, p = 0.00001), saccharopepsin (AspI) (z = -3.946, p = 0.00004), subtilisin (SerI) (z = -8.927, p = 0.00001) and grifolisin (SerII) (z = -9.041, p = 0.00001), whereas the expression of polyporopepsin (AspII) remained unchanged (z = 2.580, p = 0.99492). This is consistent with all but one of these fungal proteases having their main function after being ingested by the ants, i.e. after they pass the ant guts unharmed to be deposited with the fecal fluid.

Plotting relative gene expression (fold change in gongylidia vs. mycelium) as a function of normalized gene expression in the gongylidia (i.e. expression relative to housekeeping genes) (Figure 3b) showed proportionality for four of the proteases (saccharopepsin, subtilisin, grifolisin and polyporopepsin). However, the two metalloendoproteases deviated from the 1:1 proportionality represented by the diagonal in Figure 3b. One of them (MetII) combined a fairly low normalized expression (Table S4) with a 17x upregulation in gongylidia, and the other (MetI) combined a high normalized expression with a relatively low 5x fold upregulation in gongylidia. This implies that

119 Figure 2 Activity of metalloendoproteases, serine proteases, aspartic proteases, and cysteine proteases in milliunits (mU). Both metalloendoproteases (z = 10.824, p < 0.0001) and serine proteases (z = 2.004, p < 0.05) showed significantly enhanced activity at pH 6 relative to the zero base line. Serine proteases had their peak activity at pH 7 (z = 4.654, p < 0.0001) and activities of metalloendoproteases at pH levels >7 remained equally significant as those at pH 6. Whiskers are SEs of the mean. the expression of MetII is much more gongylidia-specific than the expression of MetI, which is apparently also needed in high quantities in the undifferentiated mycelium. It also appeared that the single non-differentially expressed protease (AspII) had a low normalized expression level (Figure 3b), indicating that it may not play a very important role under normal conditions, but might have an unknown adaptive function under specific field conditions that did not apply to our lab colonies.

Discussion In the present study we used mass spectrometry to identify seven proteases in the fecal fluid of A. echinatior leaf-cutting ants. For all but one of these we could exclude that they are produced by the ants, because we have a high quality reference genome for the ants (Nygaard et al. 2011), to which only one protease matched, and a poorly assembled genome for the fungal symbiont (De Fine Licht et al. 2013) to which the remaining six sequences matched. Our protease assays showed both metalloendoprotease activity and serine protease activity in the ant fecal fluid, but when we deprived the ants of their fungal symbiont by giving them a diet of sugar water and bramble leaves, essentially all protease activity disappeared. We further showed that five out of the six protease genes were overexpressed in the fungal gongylidia that the ants eat. In combination, the results strongly suggest that these proteases mostly belong to a distinct but restricted group of proteins that have been selected to be ingested by the ants for the sole purpose of being passed on to the fecal fluid where they provide functional mutualistic benefits for the farming symbiosis.

120 Figure 3 Average relative gene expression (fold change) across gongylidia collected from fungus gardens of five colonies ofA. echinatior: dark grey = metalloendoproteases, grey = aspartic proteases, light grey = serine proteases. A) Relative expression fold-changes (mean ± SE) in gongylidia compared to mycelium, with the dashed horizontal line representing no change in expression. Target genes were significantly upregulated in the gongylidia (P = 0.00004, 0.00001, 0.00004, 0.99492, 0.00001 and 0.00001 for MetI, MetII, AspI, AspII, SerI and SerII, respectively, marked with asterisk (*)), except for polyporopepsin (AspII). B) Relative expression fold-change plotted against gene expression normalized for the housekeeping genes GAPDH and Ubc (means ± SE). The three upregulated serine (SerI and SerII) and aspartic (AspI) proteases had fold-changes proportional to normalized gene expression (being very close to the dashed line diagonal), but peptidyl-Lys metallopeptidase I (MetI) combined a high normalized gene expression with a relatively low gene fold-change, and peptidyl-Lys metallopeptidase II (MetII) combined a low normalized gene expression with a relatively high gene fold-change.

The most parsimonious hypothesis to explain this remarkable phenomenon is that the stationary fungus uses the ants to vector crucial enzymes produced in the central parts of the garden, where the ratio between fungal biomass and remaining substrate allows high gongylidia production, to the upper parts of the garden where an abundance of new substrate is available but as yet without much hyphal growth. These findings and interpretations are consistent with earlier results obtained for seven pectinases (Schiøtt et al. 2010) and a laccase enzyme (De Fine Licht et al. 2013). It remains possible that the ingested fungal enzymes also interact with unknown other compounds in the ant gut, produced either by the ants or their microbiome, but such additional complexities appear unnecessary to explain our current or previous results obtained for other enzymes in the gongylidia and fecal fluid (Rønhede et al. 2004; Schiøtt et al. 2010; De Fine Licht et al. 2013).

121 It remains to be seen whether the proteases that we identified may also have functions that are restricted to the gongylidia. If that would be the case, the non-adaptive null hypothesis would be that finding them in the fecal fluid may merely imply that digestion was incomplete, but this seems unlikely given that SDS-PAGE banding patterns of fecal fluid were always consistent, showing a very specific subset of proteins that escape digestion (Schiøtt et al. 2010). Together with our previous work (Schiøtt et al. 2010; De Fine Licht et al. 2013) we have now shown that 12 of the 14 identified fungal pectinase, laccase and protease enzymes in the A. echinatior fecal fluid have upregulated gene expression in the gongylidia, which we believe underlines the key significance of these special hyphal tips as fungal organs mediating a series of symbiotic adaptations.

As it appears, the main function of the fecal fluid proteases is enhanced digestion of the most valuable nitrogen-containing molecules of the fresh plant material harvested by the leaf-cutting ants. Peptidyl-Lys metallopeptidases (MetI and MetII) and serine carboxyl-peptidases (SerII) have been suggested to be involved in the extraction of nutrients from extracellular proteins (Heck 2012; Oda 2012). As both MetI, MetII and SerII were active in the fecal fluid (Figure 2) and upregulated in the gongylidia (Figure 3), it seems reasonable to hypothesize that these enzymes increase nitrogen extraction from the chewed-up leaf pulp after other enzymes such as the pectinases (Schiøtt et al. 2010) have partially broken down the cell walls. The other active serine protease, subtilisin, is structurally similar to several unassigned fungal proteases in the subfamily S8A (Table S3). Most of the matches in protein databases for this enzyme were proteases from general pathogenic or phytopathogenic fungi. This finding is intriguing, because most of the subtilisins in subfamily S8A are part of a large gene-family expansion in Angiosperms with functions roles in seed or fruit development or manipulating cell walls in response to changes in biotic and abiotic factors (Schaller 2012; Schaller et al. 2012).

While subtilisins likely have defensive functions in plants, other studies have shown that subtilisins have important aggressive functions in saprotrophs, phytopathogens and other infectious pathogens (Sreedhar et al. 1999; Dunaevskii et al. 2006; Ievleva et al. 2006; Bryant et al. 2009; Li et al. 2010). One explanation for this could be that using proteins very similar to those present in the hosts may allow pathogens to avoid detection by immune defences, a phenomenon referred to as molecular mimicry (Elde & Malik 2009; Armijos Jaramillo et al. 2013). Although the fungal symbionts of leaf-cutting ants hardly face any active immune defense in the severed plant material that the ants bring in, they will be challenged by substantial amounts of secondary (defensive) compounds, similar to any functional insect herbivores whose digestive systems need to cope with plant defences that decrease digestibility (Chen 2008). Gongylidial subtilisin might therefore

122 have some role in the penetration and digestion of newly harvested plant material, consistent with serine protease activity being also high in the gardens of other higher attine ants. Panamanian representatives of the genera Trachymyrmex and Sericomyrmex that readily use soft fresh leaves as substrate in the laboratory, tend to have very high levels of serine protease activity in their fungus gardens, whereas these levels tend to be low in attine ant species that do not or only reluctantly use fresh leaves as fungal substrate (Semenova et al. 2011, P. Kooij and M. Schiøtt, unpublished observations).

We found two aspartic proteases in the fecal fluid, but none of them showed activity (Figure 2), even though one of them, saccharopepsin, was upregulated in the gongylidia when compared to mycelium (Figure 3). This suggests that this protein has a function, either for the gongylidia or after ingestion by the ants. Saccharopepsin is known to be important in the activation of yeast vacuolar hydrolases (Parr et al. 2007). This could explain the abundance of this enzyme in the gongylidia as they always have a large vacuole, but it does not explain why this protease is passed on to the fecal fluid. As non-upregulated polyporopepsin is also found in fecal fluid, it might be that these proteases cleave very specific target sequences so they would not show activity in our general activity assays. Possible target sequences could be other fungal enzymes that are produced as inactive precursors (zymogens or proenzymes), which are subsequently activated when regulatory proteases cleave off part of the precursor amino acid chain. Such an activation mechanism would be particularly appropriate for abundant plant degrading enzymes that are also harmful to the fungus garden but not to the ants. These enzymes could then be produced in large quantities in the gongylidia, but only be activated in the ant gut to be mixed with the plant substrate via fecal fluid before the fungal hyphae are inoculated.

The possibility of leaf-cutting ant fungi having independently evolved adaptations reminiscent of phytopathogens is intriguing, so it is of interest to briefly evaluate recent studies addressing evolutionary arms-races between phytopathogens and their host plants (Misas-Villamil & van der Hoorn 2008). Also here, metalloproteases appear to have a role in the breakdown of cell wall stabilizing proteins (Rauscher et al. 1995) and in targeting specific plant resistance proteins (Xia 2004). Among serine proteases, trypsins appear to be most prevalent in phytopathogens, whereas subtilisins are more important in saprotrophs (Dunaevskii et al. 2006; 2008). However, when they occur, subtilisins may have an important role in the breakdown of cell wall structural proteins (Murphy & Walton 1996), as in infections of Kentucky bluegrass, Poa pratensis (Sreedhar et al. 1999) and in Colletotricum fungi that apparently acquired a subtilisin gene by horizontal transfer from their host plant, encoding a protease which is active during infection (Armijos Jaramillo et al. 2013).

123 The possible similarities in attine fungal proteases with those of phytopathogens (St Leger et al. 1997) has interesting parallels to our earlier studies on pectinases (Schiøtt et al. 2010), and a laccase (De Fine Licht et al. 2013) in the fecal fluid of A. echinatior ants. Also here, indications of convergent evolution were found, although the evidence remained largely indirect as in the present evaluation of our proteases results. Although many of the molecular mechanisms remain unknown (but see De Fine Licht et al. 2013), it seems beyond doubt that selection pressure on herbivorous leaf-cutting ant fungi must have been largely in the same direction as in unrelated lineages of necrotrophic fungi. The relatively simple fecal fluid proteome (a total of ca. 33 identified proteins so far in A. echinatior) appears to be a valuable and relatively transparent window through which the molecular adaptations that have shaped fungus-farming in the monophyletic clade of the attine ants can be unravelled.

Acknowledgements

The authors thank the Smithsonian Tropical Research Institute (STRI), Panama, for providing logistic support and facilities to work in Gamboa, the Autoridad Nacional del Ambiente y el Mar (ANAM) for permission to sample ant colonies in Panama in to export them to Denmark, and the Danish National Research Foundation for funding (DNRF57).

Abbreviations

EF1α - Elongation factor 1-α GAPDH - Glyceraldehyde 3-phosphate dehydrogenase SDS-PAGE - Sodium dodecyl sulphate – polyacrylamide gel electrophoresis Ubc – Ubiquitin

Conflict of interest

The authors declare that they have no competing interests related to this manuscript

Supplementary Information accompanies the paper on The ISME Journal website (http://www.natur.com/ismej)

124 References Aanen DK, De Fine Licht HH, Debets AJM, Kerstes NAG, Hoekstra RF, Boomsma JJ. (2009). High symbiont relatedness stabilizes mutualistic cooperation in fungus-growing termites. Science 326:1103–1106.

Armijos Jaramillo VD, Vargas WA, Sukno SA, Thon MR. (2013). Horizontal transfer of a subtilisin gene from plants into an ancestor of the plant pathogenic fungal genus Colletotrichum. PLoS ONE 8:e59078.

Bass M, Cherrett JM. (1995). Fungal hyphae as a source of nutrients for the leaf-cutting ant Atta sexdens. Physiol En- tomol 20:1–6.

Boomsma JJ. (2013). Beyond promiscuity: mate-choice commitments in social breeding. Philos T Roy Soc B 368:20120050.

Boyd ND, Martin MM. (1975a). Faecal proteinases of the fungus-growing ant, Atta texana: properties, significance and possible origin. Insect Biochem 5:619–635.

Boyd ND, Martin MM. (1975b). Faecal proteinases of the fungus-growing ant, Atta texana: their fungal origin and ecological significanca. J Insect Physiol 21:1815–1820.

Britton HTS, Robinson RA. (1931). Universal buffer solutions and the dissociation constant of veronal. J Chem Soc 1:1456-1462

Bryant MK, Schardl CL, Hesse U, Scott B. (2009). Evolution of a subtilisin-like protease gene family in the grass en- dophytic fungus Epichloë festucae. BMC Evol Biol 9:168.

Chen M-S. (2008). Inducible direct plant defense against insect herbivores: A review. Insect Sci 15:101–114.

Christeller JT, Laing WA, Markwick NP, Burgess E. (1992). Midgut protease activities in 12 phytophagous lepidopteran larvae: dietary and protease inhibitor interactions. Insect Biochem Molec 22:735–746.

Davidson DW. (2004). Ecological stoichiometry of ants in a New World rain forest. Oecologia 142:221–231.

De Fine Licht HH, Schiøtt M, Mueller UG, Boomsma JJ. (2010). Evolutionary transitions in enzyme activity of ant fungus gardens. Evolution 64:2055–2069.

De Fine Licht HH, Schiøtt M, Rogowska-Wrzesinska A, Nygaard S, Roepstorff P, Boomsma JJ. (2013). Laccase detoxification mediates the nutritional alliance between leaf-cutting ants and fungus-garden symbionts. P Natl Acad Sci USA 110:583–587.

Dunaevskii YE, Gruban TN, Belyakova GA, Belozerskii MA. (2006). Extracellular proteinases of filamentous fungi as potential markers of phytopathogenesis. Microbiology+ 75:649–652.

Dunaevskii YE, Matveeva AR, Fatkhullina GN, Belyakova GA, Kolomiets TM, Kovalenko ED, et al. (2008). Extracel- lular proteases of mycelial fungi as participants of pathogenic process. Russ J Bioorg Chem+ 34:286–289.

Elde NC, Malik HS. (2009). The evolutionary conundrum of pathogen mimicry. Nat Rev Microbiol 7:787–797.

Fernandez-Marin H, Zimmerman JK, Nash DR, Boomsma JJ, Wcislo WT. (2009). Reduced biological control and enhanced chemical pest management in the evolution of fungus farming in ants. P R Soc B 276:2263–2269.

Fowler HG, Pagani MI, Da Silva OA, Forti LC, Da Silva VP, De Vasconcelos HL. (1989). A pest is a pest is a pest? The dilemma of Neotropical leaf-cutting ants: keystone taxa of natural ecosystems. Environ Manage 13:671–675.

Habib H, Fazili KM. (2007). Plant protease inhibitors: a defense strategy in plants. Biotechnol Mol Biol Rev 2:68–85.

Heck AJR. (2012). Chapter 284 - Peptidyl-Lys Metalloendopeptidase. In:Handbook of proteolytic enzymes, Rawlings, ND & Salvesen, GS (eds). Vol. 1, Academic Press: New York, pp. 1266–1270.

Hothorn T, Bretz F, Westfall P. (2008). Simultaneous Inference in General Parametric Models. Biometrical J 50:346– 363.

Ievleva EV, Revina TA, Kudryavtseva NN, Sof’in AV, Valueva TA. (2006). Extracellular proteinases from the phytopathogenic fungus Fusarium culmorum. Appl Biochem Micro+ 42:298–303.

125 Janzen DH. (1985). The natural history of mutualisms. In:The biology of mutualism, Bucher, DH (ed). Oxford Univer- sity Press: New York, pp. 40–99.

Kohl M, Perkins J. (2011). NormqPCR: Functions for normalisation of RT-qPCR data. Available at http://www.bioconductor.org/packages/devel/bioc/html/NormqPCR.html

Li J, Yu L, Yang J, Dong L, Tian B, Yu Z, et al. (2010). New insights into the evolution of subtilisin-like serine protease genes in Pezizomycotina. BMC Evol Biol 10:68.

Livak KJ, Schmittgen TD. (2001). Analysis of relative gene expression data using real-time quantitative PCR and the 2(-Delta Delta C(T)) Method. Methods 25:402–408.

Martin JS, Martin MM. (1970a). The presence of protease activity in the rectal fluid of attine ants. J Insect Physiol 16:227–232.

Martin MM. (1974). Biochemical ecology of the attine ants. Accounts Chem Res 7:1–5.

Martin MM. (1970). The biochemical basis of the fungus-attine ant symbiosis. Science 169:16–20.

Martin MM, Martin JS. (1970b). The biochemical basis for the symbiosis between the ant, Atta colombica tonsipes and its food fungus. J Insect Physiol 16:109–119.

Martin MM, Martin JS. (1971). The presence of protease activity in the rectal fluid of primitive attine ants. J Insect Physiol 17:1897–1906.

Mehdiabadi NJ, Schultz TR. (2010). Natural history and phylogeny of the fungus-farming ants (Hymenoptera: Formi- cidae: Myrmicinae: Attini). Myrmecol 13:37–55.

Mercado-Flores Y, Hernández-Rodríguez C, Ruiz-Herrera J, Villa-Tanaca L. (2003). Proteinases and exopeptidases from the phytopathogenic fungus Ustilago maydis. Mycologia 95:327–339.

Misas-Villamil JC, van der Hoorn RAL. (2008). Enzyme-inhibitor interactions at the plant-pathogen interface. Curr Opin Plant Biol 11:380–388.

Mueller UG, Rehner SA, Schultz TR. (1998). The evolution of agriculture in ants. Science 281:2034–2038.

Murphy JM, Walton JD. (1996). Three extracellular proteases from Cochliobolus carbonum: cloning and targeted dis- ruption of ALP1. Mol Plant Microbe In 9:290–297.

Nygaard S, Zhang G, Schiøtt M, Li C, Wurm Y, Hu H, et al. (2011). The genome of the leaf-cutting ant Acromyrmex echinatior suggests key adaptations to advanced social life and fungus farming. Genome Res 21:1339–1348.

Oda K. (2012). Chapter 735 - Sedolisin. In:Handbook of proteolytic enzymes, Rawlings, ND & Salvesen, GS (eds). Academic Press: New York, pp. 3331–3337.

Parr CL, Keates RAB, Bryksa BC, Ogawa M, Yada RY. (2007). The structure and function of Saccharomyces cerevisiae proteinase A. Yeast 24:467–480.

Perkins J, Kohl M. (2011). ReadqPCR: Read qPCR data. Available at http://www.bioconductor.org/packages/2.14/bioc/ html/ReadqPCR.html

Pinheiro J, Bates D, DebRoy S, Sarkar D, R Core Team. (2013). nlme: Linear and nonlinear mixed effects models. Available at http://cran.r-project.org/web/packages/nlme/index.html

Pinto-Tomás AA, Anderson MA, Suen G, Stevenson DM, Chu FST, Cleland WW, et al. (2009). Symbiotic nitrogen fixation in the fungus gardens of leaf-cutter ants. Science 326:1120–1123.

Poulsen M, Boomsma JJ. (2005). Mutualistic fungi control crop diversity in fungus-growing ants. Science 307:741–744.

Queller DC, Strassmann JE. (2009). Beyond society: the evolution of organismality. Philos T Roy Soc B 364:3143– 3155.

R Core Team. (2013). R: A language and environment for statistical computing. Available at http://cran.r-project.org/

126 Rauscher M, Mendgen K, Deising H. (1995). Extracellular proteases of the rust fungus Uromyces viciae-fabae. Exp Mycol 19:26–34.

Rawlings ND, Barrett AJ, Bateman A. (2012). MEROPS: the database of proteolytic enzymes, their substrates and inhibitors. Nucleic Acids Res 40:D343–50.

Rønhede S, Boomsma JJ, Rosendahl S. (2004). Fungal enzymes transferred by leaf-cutting ants in their fungus gardens. Mycol Res 108:101–106.

Sarath G, Zeece MG, Penheiter AR. (2001). Protease assay methods. In:Proteolytic enzymes: A practical approach, Beynon, R & Bond, JS (eds). Oxford University Press: Oxford, pp. 45–76.

Schaller A. (2012). Chapter 717 - Plant Subtilisins. In:Handbook of proteolytic enzymes, Rawlings, ND & Salvesen, GS (eds). Academic Press: New York, pp. 3247–3254.

Schaller A, Stintzi A, Graff L. (2012). Subtilases - versatile tools for protein turnover, plant development, and interactions with the environment. Physiol Plantarum 145:52–66.

Schiøtt M, De Fine Licht HH, Lange L, Boomsma JJ. (2008). Towards a molecular understanding of symbiont function: Identification of a fungal gene for the degradation of xylan in the fungus gardens of leaf-cutting ants. BMC Microbiol 8:40.

Schiøtt M, Rogowska-Wrzesinska A, Roepstorff P, Boomsma JJ. (2010). Leaf-cutting ant fungi produce cell wall degrading pectinase complexes reminiscent of phytopathogenic fungi. BMC Biol 8:156.

Schultz TR, Brady SG. (2008). Major evolutionary transitions in ant agriculture. P Natl Acad Sci USA 105:5435–5440.

Semenova TA, Hughes DP, Boomsma JJ, Schiøtt M. (2011). Evolutionary patterns of proteinase activity in attine ant fungus gardens. BMC Microbiol 11:15.

Sreedhar L, Kobayashi DY, Bunting TE, Hillman BI, Belanger FC. (1999). Fungal proteinase expression in the interaction of the plant pathogen Magnaporthe poae with its host. Gene 235:121–129.

St Leger RJ, Joshi L, Roberts DW. (1997). Adaptation of proteases and carbohydrases of saprophytic, phytopathogenic and entomopathogenic fungi to the requirements of their ecological niches. Microbiol-Sgm 143:1983–1992.

Villesen P, Murakami T, Schultz TR, Boomsma JJ. (2002). Identifying the transition between single and multiple mating of queens in fungus-growing ants. P R Soc B 269:1541–1548.

Weber NA. (1966). Fungus-growing ants. Science 153:587–604.

Xia Y. (2004). Proteases in pathogenesis and plant defence. Cell Microbiol 6:905–913.

127 Table S1. List of primers used for gene cloning.

Primer Gene Primer function Primer sequences name AspI Asp-F1 3’ RACE primer 5’-CGTTCCCTTCGTTGTTGGCCTCC-3’ Asp-R1 5’ RACE primer 5’- GCAAGCTTGTGAAGTCTGAGGGTGTG-3’ AspII Pro-F4 Forward degenerate primer 5’-ACNGGXAGNTCXAAYACNTGG-3’ Pro-R2 Reverse degenerate primer 5’-TTRTCNGTIACNGTIGGDAT-3’ Pro-F5 3’ RACE primer 5’-CCGTCGATCTCACGCGAGGAACGCTG-3’ Pro-R5 5’ RACE primer 5’-GCGACGCTGTTGCCGGTCTGCTGAC-3’ MetI Forward 3’RACE degenerate Sp21-F2 5’-GGTGGTACYTAYGTNCAYGA-3’ primer* Met-R1 5’ RACE primer 5’-AGCGAGTGCCTTGGCGCCAGATTG-3’ Met-R2 5’ RACE primer 5’-CGTAGTCTTGTGTTCCGCCATTTCTGAGG-3’ Met-R3 5’ RACE primer 5’-AGGGCTTGAGAGGCTCGATTGAGTAAG-3’ Met-R4 5’ RACE primer 5’-CCGGAGTTGGTGAAGTTGTACATTGTTGAC-3’ Met-R5 5’ RACE primer 5’-CAAGAGCAACAGCACATCGC-3’ MetII MetII-full-F3 Forward primer 5’-AATGTCAATTCTATCGAATTCAAG-3’ MetII-R1 Reverse qPCR primer 5’-CGTGGACGATTGTTCCACCTC-3’ MetII-F1 Forward qPCR primer 5’- GGCTATTACTGCAGCGAAGACC -3’ MetII-full-R Reverse primer 5’-CAATTCAAGAATTATATGGCAGG-3’ SerI Sub-F1 Forward degenerate primer 5’-GTTGARTTYGARGGXMGNGC-3’ Sub-R1 Reverse degenerate primer 5’-TTRTCRTTACCXGCNGCNAC-3’ Sub1-F1 3’ RACE primer 5’-GATCGCCGACAACAAGTATGAAGATGGTG-3’ Sub1-R1 5’ RACE primer 5’-GCACAGTGAGTACCATGACCAACACC-3’ SerII Sp24-F3 Forward degenerate primer 5’-GCTGACCTYCARACNTTYTT-3’

Sp16-R2 Reverse degenerate primer 5’-GAGAAACCACCNCCNGARAA-3’ Ser-F1 3’ RACE primers 5’-TAGGCGGTGGCGACTGCCGTACTAATGAC-3’ Ser-R2 5’ RACE primer 5’-GAAACTAGAACCTGCAGCATCGGGGCGG-3’ Ser-R3 5’ RACE primer 5’-CGGAATCTGGGTCAAAAGGAGGGGAACCG-3’ Ser-R4 5’ RACE primer 5’-GTCCATACCGTGCATGACCAGGATCGC-3’ * This primer was used together with the UPM primers from the SMART RACE cDNA kit

Note: X symbolizes inosines

128 Table S2

The Real Time qPCR primers used in this study.

Annealing Enzyme Primer 5’- sequence -3’ Size temp °C Saccharopepsin Pep-F2 CCG CCA TTG ACA CTG GTA CC 53.8 289bp (AspI) Pep-R2 GGC GGA GGA AAA CAT CAC CAA 54.4

Polyporopepsin Pol-F2 CTC GTT GAC ACC GGG AGT TC 55.9 243bp (AspII) Pol-R3 GAC GGG GCC GAT GCC TAA 54.9 Peptidyl-Lys Met-F2 CGG GAC CGA CTC TAA GGG 54.9 metallopeptidase 169bp (MetI) Met-R6 TTC ATT GTG AAT ACA TCA TAG ACT TC 51.7 Peptidyl-Lys MetII-F1 GGC TAT TAC TGC AGC GAA GAC C 56.7 metallopeptidase 307bp (MetII) MetII-R1 CGT GGA CGA TTG TTC CAC CTC 56.3 Sub-F2 AGA AAT CTG CTC CAT GGG GC 53.8 Subtilisin (SerI) 300bp Sub-R1 CTA CCA TCA GAA CCA AGC ACC 54.4 Ser-F3P GCT ACC TTC AAT GGT CTA TAC AAT AAA AC 56.0 Grifolisin (SerII) 310bp Ser-R4P TAC CCA GGC CAG TGA CAG GAT 55.4

Elongation factor EF1a-F1 TTG GAG GAA TCT CCC AAC ATG 57.9 273bp 1-a (EF1-a) EF1a-R1 AAC GGA CTT GAC TTC AGT AGT C 58.4 Glyceraldehyde 3-phosphate GAPDH-F4 TCA ACG GCA AGC TCA CTG GT 53.8 253bp dehydrogenase GAPDH-R4 ACA AAA TTC CCG TTC AAA GGA ATC 52.3 (GAPDH) Ubc-F4 AAC GAT AAT GGG ACC CGG TG 53.8 Ubiquitin (Ubc) 239bp Ubc-R4 GAT TGG GAT CAG TCA GCA TTG 52.4

129 Table S3

Comparison of the sequenced proteases with the MEROPS database showing similarities with known fungal peptidases. For each protease the top ten blast results are shown.

Protease MEROPS MEROPS Protease name Species % identity E-value code classification AspI MER122335 A01.018 saccharopepsin Laccaria bicolor 78.57% 1.10e-142 MER309581 A01.018 saccharopepsin Serpula lacrymans 79.56% 4.40e-141 MER107385 A01.018 saccharopepsin Coprinus cinereus 78.02% 1.70e-139 MER309629 A01.018 saccharopepsin Schizophyllum 77.36% 4.10e-138 commune MER382986 A01.018 saccharopepsin Stereum hirsutum 76.88% 5.20e-138 MER382978 A01.018 saccharopepsin Coniophora puteana 76.42% 8.40e-138 MER382983 A01.018 saccharopepsin Punctularia 77.67% 1.80e-137 strigosozonata MER382989 A01.018 saccharopepsin Trametes versicolor 77.12% 180e-135 MER382982 A01.018 saccharopepsin Dichomitus squalens 76.42% 4.80e-135 MER382984 A01.018 saccharopepsin Fomitiporia 74.53% 1.30e-132 mediterranea AspII MER118464 A01.019 polyporopepsin Laccaria bicolor 71.60% 1.80e-125 MER383691 A01.019 polyporopepsin Pholiota nameko 74.53% 2.70e-122 MER383440 A01.019 polyporopepsin Trametes versicolor 71.60% 2.80e-120 MER383572 A01.019 polyporopepsin Dichomitus squalens 71.74% 3.20e-190 MER000940 A01.019 polyporopepsin Polyporus tulipiferea 69.04% 9.70e-118 MER383537 A01.019 polyporopepsin Stereum hirsutum 70.15% 2.00e-117 MER383829 A01.019 polyporopepsin Punctularia 70.37% 4.20e-117 strigosozonata MER062431 A01.019 polyporopepsin Phanerochaete 67.70% 4.50e-113 chrysosporium MER172923 A01.019 polyporopepsin Laccaria bicolor 65.43% 5.30e-110

MER383761 A01.019 polyporopepsin Stereum hirsutum 64.71% 1.40e-109 MetI MER003546 M35.004 Peptidyl-Lys Pleurotus ostreatus 56.55% 1.50e-53 metallopeptidase MER005595 M35.004 Peptidyl-Lys Armillaria mellea 57.49% 1.20e-51 metallopeptidase MER003545 M35.004 Peptidyl-Lys Grifola frondosa 51.50% 7.90e-50 metallopeptidase MER187451 M35.003 EcpA peptidase Corallococcus 50.00% 4.30e-42 coralloides MER019534 M35.003 EcpA peptidase Xanthomonas 48.82% 3.60e-38 axonopodis MER255520 M35.003 EcpA peptidase Collimonas 45.51% 5.80e-38 fungivorans MER089593 M35.003 EcpA peptidase Shewanella loihica 47.56% 2.50e-37 MER071568 M35.003 EcpA peptidase Shewanella 48.45% 4.10e-37 denitrificans MER075809 M35.003 EcpA peptidase Shewanella 48.50% 5.20e-37 amazonensis MER062468 M35.003 EcpA peptidase Xanthomonas 47.62% 2.90e-36 campestris

130 MetII MER003546 M35.004 Peptidyl-Lys Pleurotus ostreatus 69.23% 2.60e-42 metallopeptidase MER003546 M35.004 Peptidyl-Lys Pleurotus ostreatus 54.90% 2.60e-42 metallopeptidase MER003546 M35.004 Peptidyl-Lys Pleurotus ostreatus 47.06% 2.60e-42 metallopeptidase MER003546 M35.004 Peptidyl-Lys Pleurotus ostreatus 71.43% 2.60e-42 metallopeptidase MER003546 M35.004 Peptidyl-Lys Pleurotus ostreatus 35.48% 1.60e-21 metallopeptidase MER005595 M35.004 Peptidyl-Lys Armillaria mellea 69.23% 6.80e-33 metallopeptidase MER005595 M35.004 Peptidyl-Lys Armillaria mellea 53.19% 6.80e-33 metallopeptidase MER005595 M35.004 Peptidyl-Lys Armillaria mellea 48.48% 6.80e-33 metallopeptidase MER005595 M35.004 Peptidyl-Lys Armillaria mellea 54.17% 4.60e-24 metallopeptidase MER255520 M35.003 EcpA peptidase Collimonas 59.18% 2.60e-32 fungivorans SerI MER169844 S08.UPA Subfamily S8A Laccaria bicolor 61.28% 6.90e-111 unassigned peptidases MER034994 S08.UPA Subfamily S8A Coprinus cinereus 60.17% 3.10e-108 unassigned peptidases MER032520 S08.UPA Subfamily S8A Aspergillus oryzae 52.47% 3.30e-100 unassigned peptidases MER032520 S08.UPA Subfamily S8A Aspergillus oryzae 36.11% 3.30e-100 unassigned peptidases MER236481 S08.UPA Subfamily S8A Cryptococcus gattii 56.90% 5.80e-98 unassigned peptidases MER087598 S08.UPA Subfamily S8A Neosartorya fischeri 61.65% 9.00e-94 unassigned peptidases MER087598 S08.UPA Subfamily S8A Neosartorya fischeri 36.11% 9.00e-94 unassigned peptidases MER003475 S08.UPA Subfamily S8A Aspergillus fumigatus 61.65% 1.10e-93 unassigned peptidases MER003475 S08.UPA Subfamily S8A Aspergillus fumigatus 36.11% 1.10e-93 unassigned peptidases MER006088 S08.UPA Subfamily S8A Penicillium 60.43% 4.90e-93 unassigned chrysogenum peptidases SerII MER137045 S53.010 grifolisin Laccaria bicolor 66.67% 4.40e-136 MER137535 S53.010 grifolisin Laccaria bicolor 60.05% 1.90e-119 MER389943 S53.010 grifolisin Stereum hirsutum 56.77% 6.00e-93 MER125099 S53.010 grifolisin Coprinus cinereus 46.44% 1.20e-89 MER178229 S53.010 grifolisin Sclerotinia 44.78% 1.30e-85 sclerotiorum MER178229 S53.010 grifolisin Sclerotinia 40.80% 1.30e-85 sclerotiorum MER405414 S53.UPW Family S53 Agaricus bisporus 66.24% 1.50e-82 unassigned peptidases MER078639 S53.010 grifolisin Grifola frondosa 45.70% 1.90e-80 MER032166 S53.010 grifolisin Emericella nidulans 44.53% 1.10e-79 MER030252 S53.010 grifolisin Aspergillus oryzea 43.91% 1.50e-78

131 Table S4

Results of statistical analyses in R on normalized (relative to the housekeeping genes GAPDH and Ubc) and relative gene expression measurements (gongylidia relative to mycelium) 0.9403 1.2406 0.6121 0.9072 2.6928 2.1765 SD mycelium ) Ct - ession 4.598 0.1446 0.9162 1.0086 0.1767 1.0875 exp r Normalized mycelium (2 13 3.25 1.7 1 1.1513 2.2386 2.8525 1.2502 SD gongylidia ) Ct - ession 2.4346 6.3952 2.2548 0.1307 4.2134 23.1296 exp r Normalized gongylidia (2 p value 0.00001 0.00004 0.00001 0.00001 0.99492 0.00004 2.5797 z value -3.9793 -9.0412 -8.9269 -3.9462 -20.0102 1.173 min - max SD Ct 1.6288 - 9.217 0.2258 - 2.422 2.2648 - 1 1.4791 - 32.944 0.3095 - 16.146 1.7703 - 160.232 - 2 ) Ct - 3.8745 5.0304 6.9805 2.2356 0.7395 16.8421 Relative fold change (2 otease SerI AspI MetI SerII AspII MetII P r

132 Cellulose degradation patterns in Acromyrmex echinatior colonies

Pepijn W Kooij, Jeroen W M Pullens, Jacobus J Boomsma , Morten Schiøtt

(Manuscript) Chapter 5 133 Cellulose degradation patterns in Acromyrmex echinatior colonies

Pepijn W. Kooij1, Jeroen W.M. Pullens1,2, Jacobus J. Boomsma1, Morten Schiøtt1

1Centre for Social Evolution, Department of Biology, Copenhagen University, Universitetsparken 15, DK-2100 Copenhagen, Denmark

2Laboratory of Genetics, Wageningen University and Research Centre, P.O. Box 309, 6700 AH Wageningen, The Netherlands

134 Abstract Cellulose is the dominant component in plant cell walls, but only few herbivorous animals can digest this compound without help of microbial symbionts. Leaf-cutting ants are special in feeding on a domesticated fungus that is provisioned with freshly harvested live plant material that the ants chew up and mix with fecal fluid. In recent studies we have shown that fungal-derived fecal fluid pectinases, laccases and proteases are particularly produced in the gongylidia, inflated hyphal tips that are the main food source for the ants. Here we complement these results by investigating the presence, activity and gene regulation of a putative fungal cellulase enzyme (LgCel12) in the ant fecal fluid. We show that this cellulase is only active in the fecal fluid when the ants have access to their normal fungal diet, but remains less active than the fecal fluid pectinases and proteases. The Lgcel12 gene is ca. 3-fold upregulated in the gongylidia that the ants ingest compared to normal mycelium, consistent with overall gene expression being highest in the middle layer of fungus gardens where most gongylidia are produced. However, overall cellulase activity was higher in the older, bottom layers of gardens and in discarded debris material than in the upper and middle layers of gardens. These results indicate that LgCel12 endocellulase plays a role in the initial breakdown of plant substrate but without being a major participant in the decomposition process, as overall cellulolytic activity is most prevalent in the bottom layer of the gardens, consisting of material that is about to be discarded by the ants. We offer a stoichiometric interpretation of these results emphasizing that the symbiosis between leaf-cutting ants and their garden symbiont is almost certainly protein limited, so that the acquisition of nitrogen and phosphorus is prioritized by the joint efforts of the mutualistic partners and excess carbon-based resources discarded in considerable quantities.

Keywords: Leaf-cutting ants, Leucoagaricus gongylophorus, gene expression, stoichiometry

135 Introduction Cellulose is the dominant constituent of plant cell walls and is considered the most abundant organic compound on earth, but proteins with the capacity to degrade cellulose are encoded in the genomes of very few animals (Watanabe and Tokuda 2001). Instead, most animals that specialize on cellulose-rich food live in symbiosis with cellulose degrading microorganisms allowing them to tap into this enormous nutritional resource. Symbioses of this kind are particularly prevalent among the insects where feeding on dead wood and other plant material has evolved multiple times independently (Breznak 1982; Douglas 2013). These gut symbiont microorganisms are usually bacteria, but protists fulfil similar roles in the lower wood-dwelling termites (Ohkuma 2003; Noda et al. 2007). However, in some other cases, these services are provided by fungi, with gut yeasts in several families of wood-ingesting beetles (Zhang et al. 2003; Grünwald et al. 2010), ambrosia fungi in wood-boring Xyleborine beetles (Biedermann et al. 2009), and Termitomyces fungi in fungus-growing termites (Aanen et al. 2002).

Also the lower attine fungus-growing ants use dead plant material to manure their fungus-gardens, which likely implies that these fungi must have considerable cellulose degrading capacities (De Fine Licht and Boomsma 2010). However, in contrast to any other insect-fungus symbiosis, the attine ants evolved an evolutionarily derived lineage of active herbivores, which gained access to vast amounts of fresh leaf substrate with an abundance of intracellular sugars and proteins that would typically no longer be available in dead plant cells. Although the fungi of these leaf-cutting ants have retained the capacity to degrade cellulose (De Fine Licht et al. 2010; Moller et al. 2011; Aylward et al. 2013; Grell et al. 2013), the relative importance of cellulolytic enzymes for the herbivorous leaf-cutting ant symbiosis has remained controversial (Abril and Bucher 2002; Erthal et al. 2009; Nagamoto et al. 2011).

The two genera of leaf-cutting ants, Atta and Acromyrmex, live in obligate mutualistic symbiosis with a single, genetically variable fungal species: Leucoagaricus gongylophorus (Weber 1966; Mueller et al. 2005; Kooij et al., Chapter 1). These ants are dominant herbivores in the New World tropics (Della Lucia et al. 2014; Kooij et al., Chapter 3, submitted) and significant accelerators of nitrogen and phosphorus cycling (Fowler et al. 1989). Similar to their lower attine ant relatives rearing saprotrophic fungus gardens, the leaf-cutting ants cultivate their symbiotic fungus in special underground chambers, but their fungal crop belongs to a unique domesticated lineage that has evolved special symbiotic organs, staphylae, that grow bundles of inflated hyphal tips (gongylidia) that provide essentially all nutrition for the ants and their brood (Quinlan and Cherrett 1978; Quinlan and Cherrett 1979; Bass and Cherrett 1995). After the larger foraging worker ants

136 have cut the leaf fragments and transported them back to the nest, smaller workers chew these pieces into a pulp, reducing the crystallinity of the plant cellulose (Martin et al. 1991; Nagamoto et al. 2011), and place this material on top of the fungus garden before manuring it with fecal fluid and inoculating it with fresh tufts of mycelium.

The fecal fluid is used for the rapid maceration of cut leaves (Martin et al. 1975), and contains a wide range of fungal plant degrading enzymes such as pectinases (Schiøtt et al. 2010), laccases (De Fine Licht et al. 2013) and proteases (Kooij et al. in press). These enzymes have retained activity in the fecal fluid after gut passage, and the genes encoding them were almost always upregulated in the gongylidia when compared to the normal mycelium. However, gongylidia upregulation and fecal fluid expression for cellulases has not been investigated, as cellulases were not particulatly common among the ca. 33 proteins found in the Acromyrmex echinatior fecal fluid (Schiøtt et al. 2010). Only a single protein (LgCel12) belonging to the glycoside hydrolase family 12 (GH12) was identified in the fecal fluid of Panamanian Acromyrmex echinatior (Schiøtt et al. 2010), but remained unconfirmed in a recent study on lignocellolytic enzymes in Leucoagaricus gongylophorus from other sympatric colonies of the same ant species (Aylward et al. 2013).

Substrates targeted by the members of the GH12 enzyme family are cellulose, xyloglucan and β-1,3-1,4-glucan. As already indicated above, the importance of cellulose degradation for the ant-fungus mutualism has been much debated (e.g. Martin and Weber 1969; Abril and Bucher 2002; Richard et al. 2005; Erthal et al. 2009; Moller et al. 2011; Nagamoto et al. 2011; Aylward et al. 2013; Grell et al. 2013), which makes finding a putative non-digested fungal cellulase in the ant fecal fluid (but missed in the proteome of the fungus garden; Aylward et al. 2013) of interest, as it may suggest that some steps in cellulose degradation in the top of fungus gardens (where fecal fluid is deposited) may be important for the symbiosis even in low quantities. The present study therefore set out to: 1. Measure cellulase activity in both fecal fluid and fungus gardens of Acromyrmex echinatior, 2. Confirm that most of the cellulase activity in the fecal fluid is fungal derived, 3. Measure gene expression levels of Lgcel12 in the different layers of the fungus garden and in the fungal gongylidia relative to normal mycelium, and 4. Discuss the implications of our findings to the ongoing discussion on cellulose degradation by leaf-cutting ant fungus gardens.

Material and methods Biological material A total seven ant colonies of Acromyrmex echinatior (Ae226, Ae263, Ae266, Ae280, Ae322, Ae335 and Ae370), collected between 2004 and 2007 in Gamboa, Panama, were used for this study.

137 All colonies were kept under controlled conditions of ca. 25˚C and a relative humidity of ca. 70%, and were fed twice a week with fresh bramble leaves (Rubus sp.), parboiled rice and pieces of apple. For enzyme activity measurements ant fecal fluid and fungus garden material were collected from five of these ant colonies (Ae226, Ae263, Ae266, Ae280 and Ae322).

For the collection of fecal fluid, the gaster of a large worker was gently pressed to a glass microscope slide using a forceps (Rodrigues et al. 2008). Each fecal droplet was mixed with 0.5µl demineralized ddH2O (the fecal fluid “enzyme extract”) and collected with a micropipette in an Eppendorf tube kept on ice until further processing. Fecal fluid was collected from two different groups of large workers in each colony, one having access to the fungus garden and thus likely having ingested fungal gongylidia, and the other being deprived of their fungus garden and provided only with 10% sucrose water for 20 days. We collected 120 mg fungus garden material from each of the three visible layers of fungus garden (Kooij et al. 2011), the fresh and dark top layer, the gongylidia-rich middle layer (De Fine Licht et al. 2013) and the older bottom layer. We supplemented this material with samples from the debris material, i.e. old fungus garden material discarded by the ants, similar to a previous study (Kooij et al. 2011) and added 500 µl of ddH2O to each sample and crushed material with a pestle before vortexing and centrifuging for 15 min (15000 g). The fungus garden supernatant and the fecal fluid extract were then kept on ice until further processing the same day.

For gene expression measurements in gongylidia and mycelium, we collected from five colonies (Ae263, Ae280, Ae322, Ae335 and Ae370) ca. 100 µl of gongylidia clusters (staphylae) and mycelium in separate 2 ml Eppendorf tubes floating in liquid nitrogen using a stereomicroscope at 40x magnification. For gene expression measurements in the different fungus garden layers and the debris pile, these samples were supplemented by 120 mg fungal material from colonies Ae226, Ae263, Ae266, Ae280 and Ae322, after carefully removing any visible ants or ant parts, larvae and eggs. All these gene expression samples were stored at -80˚C for RNA extraction at a later point.

Cellulase activity Fecal fluid and fungus garden cellulase activities were measured for β-glucosidase witha pNPG-assay and for endo-1,4-β-glucanase with both a CMC/DNS-assay (Zhang et al. 2009) and the AZO-CM-Cellulose-assay (Megazyme, Wicklow, Ireland). Exoglucanase assays require purified enzymes to avoid interference with other cellulases and could therefore not be performed. β-Glucosidase activity was determined by incubating 3 µl enzyme extract from fecal fluid or fungus garden in 42 µl mM 4-nitrophenyl-β-D-glucopyranoside (Sigma N7006) and 0.1 M sodium

138 acetate buffer, pH 4.8, for 30 minutes at 30°C. We added 60 µl 0.4 M glycine buffer, pH 10.8 and read absorbances at 430nm using a Versamax ELISA microplate reader (Molecular Devices, Sunnyvale, USA). Absorbance measurements were converted to enzyme units using a standard curve made with 4-nitrophenol (Sigma 241326), defining one unit as the amount of enzyme required to release 1 µg nitrophenol per minute per fecal droplet or mg fungus garden.

Using the CMC/DNS-assay, endo-1,4-β-glucanase activity was determined by incubating 2.5 µl enzyme extract in 2 % (w/v) Carboxymethylcellulose (Sigma 21902) and 50 mM citrate buffer, pH 4.8 for 30 minutes at 30°C. A quantity of 15 µl DNS solution (0.4 M NaOH, 0.04M 3,5-dini- trosalicylic acid, 1 M potassium sodium tartrate) was added, and the mixture was heated at 99.9°C for 5 minutes in a PCR machine. Samples were put on ice to terminate the reaction and 100 µl distilled water was added, after which the absorbance at 540 nm was read in a Versamax ELISA microplate reader (Molecular Devices, Sunnyvale, California, USA). The absorbance measurements were converted to enzyme units using a standard curve made with glucose (Sigma G8270). One unit was defined as the amount of enzyme required to release 1 µg of glucose per minute per fecal droplet or mg fungus garden.

Endo-1,4-β-glucanase activity was also determined using the AZO-CM-Cellulose-assay (Mega- zyme, Wicklow, Ireland) using 50 µl of enzyme extract added to 50 µl pre-equilibrated substrate solution, pH 5. The solution was vortexed and incubated for 30 minutes at 30˚C and the reaction terminated by adding 250 µl of precipitant solution. Samples were then vortexed and centrifuged for 10 minutes at 1000 g, after which the supernatant was measured for absorbance at 590 nm in the Versamax ELISA micro plate reader and converted to enzyme units using the standard curve included in the protocol, once more defining a unit as the amount of enzyme required to release 1 µg of glucose per minute per fecal droplet or mg fungus garden. All data for cellulase activity were analysed with R (R Core Team 2013) using a Wilcoxon test for the fecal fluid data and a Kru- skal-Wallis test for the garden-layer data followed by the multiple comparison test “kruskalmc” from the package “pgirmess” (Giraudoux 2013).

AZCL enzyme activity assays Fecal fluid enzyme profiles were performed similar to previously published methods (Rønhede et al. 2004; De Fine Licht et al. 2010). In short, 50 fecal droplets per colony were diluted 20x with ddH2O and 1 ml of this suspension was applied immediately to 16 different assay-plates containing 0.1 g/l of the Azurine-Crosslinked (AZCL) substrates: amylose, arabinoxylan, barley ß-glucan, casein, chitosan, collagen, curdlan, debranched arabinan, dextran, galactan, galactomannan,

139 HE-cellulose, pachyman, pullulan, xylan and xyloglucan.

The assay-plates of 6 cm diameter were prepared separately for each substrate using an agarose medium (1 % agarose, 23 mM phosphoric acid, 23 mM acetic acid, 23 mM boric acid) and were pH adjusted according to the manufacturer’s description (Megazyme, Bray, Ireland). After the medium had solidified, round wells of ca. 0.1 cm² were made in each plate with a cut-off pipette tip and 10 µl of the fecal fluid mixture was applied to each well in triplicate. After 22 hours of incubation at 25°C the plates were photographed and the area of the blue halo surrounding each well (a quantitative measure for the absolute amount of enzyme activity) was measured using the software program ImageJ ver. 1.43u for Macintosh (De Fine Licht et al. 2010; Kooij et al. 2011). Enzyme activity measurements were grouped into categories based on the main plant cell wall target of each enzyme (Kacuráková et al. 2000 and Megazyme, Bray, Ireland): amylases (measured with amylose), cellulases (measured with barley ß-glucan and HE-cellulose), hemicellulases (measured with arabinoxylan, curdlan, galactomannan, pachyman, xylan and xyloglucan), pectinases (measured with debranched arabinan and galactan), proteases (measured with casein and collagen), and a rest group (including the substrates chitosan, dextran and pullulan, to measure chitosanases, dextranases and pullulanases respectively). The average halo values were calculated for each triplicate of identical plates after which the data were log transformed and analyzed with R, using Linear Mixed-Effects Models (“lme”) in the package “nlme” (Pinheiro et al. 2013) and posthoc analysis with General Linear Hypotheses (“glht”) in the package “multcomp” (Hothorn et al. 2008). We present these data grouped for the six enzyme categories with colony as random effect.

Protein identification and gene cloning SDS-PAGE and mass spectrometry was performed as described previously (Schiøtt et al. 2010; De Fine Licht et al. 2013; Kooij et al. in press). A mix of fifty fecal droplets was subjected to SDS- PAGE, after which individual protein bands were manually excised from the gel. Proteins were extracted from the gel plugs and digested with trypsin. De novo sequencing of the resulting peptide fragments was performed on the basis of b and y fragment ions present in MS/MS spectra of derivatized and underivatized samples. Amino acid sequences were obtained by manual analysis of the spectra using the software program AminoCalc (Protana A/S, Odense, Denmark) as support. The obtained amino acid sequences were used as queries in a Blast search of the assembled Ac- romyrmex echinatior genome (Nygaard et al. 2011) and a low coverage genome sequence of the Acromyrmex echinatior fungal symbiont (De Fine Licht et al. 2013).

140 RNA extraction and cDNA production was performed as described in Schiøtt et al. (2010). A full- length gene transcript sequence of Lgcel12 was obtained using a RACE (Rapid Amplification of cDNA Ends) strategy. 3’ and 5’-RACE libraries were made from ca. 1 µg of extracted RNA with the SMART RACE cDNA kit (CLONTECH, Mountain View, California, USA), and 3’end and 5’end gene sequences were PCR amplified from these libraries using the gene specific primers SYT-F1 (CAG TCG ACA CTT CCC AGC ACT GTG) or SYT-R1 (GGT ATA CTG ACC AGC AGT GAC GGT G) designed from the BLAST-search-identified sequence reads along with the UPM primer enclosed in the SMART RACE cDNA kit. The following PCR scheme was used in the RACE experiments: one cycle of 95°C for 5 min, 10 cycles of 95°C for 20 sec, 72°C for 30 sec (with a decrease in temperature of 0.5°C in every cycle) and 72°C for 3 min, followed by 35 cycles of 94°C for 20 sec, 67°C for 30 sec and 72°C for 3 min, and ending with one cycle of 72°C for 7 min. PCR products were cloned in pCR4-TOPO using the TOPO-TA cloning procedure (Invitro- gen, Carlsbad, California, USA) and sequenced at Eurofins MWG Operon (Ebersberg, Germany). The full length gene sequence will be deposited in GenBank after publication.

Quantitative real-time PCR Primers for the Lgcel12 gene (Table S1) were designed by matching the obtained cDNA sequences to a partially sequenced genome of the Acromyrmex echinatior symbiont (De Fine Licht et al. 2013) using BLASTn. Matching sequencing reads were assembled and aligned to the cDNA sequence in question to identify intron and exon sequences. Primers were then designed to span an intron to avoid amplification of genomic DNA. The qPCR was run on an Mx3000P QPCR System (Agilent, Santa Clara, CA, USA) in a 20 µl qPCR reaction (0.5 µl cDNA, 10 µl 2x SYBR Premix Ex Taq [TaKaRa Bio Inc., Otsu, Japan] and 0.4 µl of each primer [10µM]) with the following PCR conditions: one cycle of 95°C denaturing for 2 min, followed by 40 cycles of 95°C denaturing for 30 sec, 57°C annealing for 30 sec, and 72°C extension for 30 sec and ending with a melting curve cycle of 95°C for 30 sec, 57°C for 30 sec and 95°C for 30 sec.

We performed qPCR for three genes in total: the target gene Lgcel12 and two housekeeping genes ubiquitin (Ubc, GenBank HQ174771) and glyceraldehyde 3-phosphate dehydrogenase (GAPDH, GenBank HQ174770) with three technical replicates for each sample. All Ct values from the RT-qPCR analyses were analysed using R with packages “ReadqPCR” (Perkins and Kohl 2011) and “NormqPCR” (Kohl and Perkins 2011). Average expression was calculated from the three technical replicates of each sample for two datasets, one with Ct values for the different fungus garden layers including the debris pile and the other with Ct values for the fungus garden mycelium and gongylidia.

141 Based on calculations for housekeeping gene stability with “selectHKs” from the “NormqPCR” package, only GAPDH was used as reference housekeeping gene to determine the normalized expression values for Lgcel12 in the fungus garden layer samples, which produced 2∆Ct values (Schmittgen and Zakrajsek 2000) for each layer. These results were further analysed with R using ANOVA to generate individual p-values with General Linear Hypotheses (“glht”) in the Simulta- neous Inference in General Parametric Models package “multcomp” (Hothorn et al. 2008). For the gene expression comparison between gongylidia and mycelium, both housekeeping genes (Ubc and GAPDH) were used to calculate the relative expression levels of the Lgcel12 gene, resulting in 2-∆∆Ct values (Livak and Schmittgen 2001).

Results Both the Megazyme-assay (Figure 1A) and the CMC-assay (Figure 1B) for endo-1,4-β-glucanase showed a substantial reduction in activity when ants had been feeding on sucrose water for 20 days (W = 23, p = 0.0159; W = 22, p = 0.0278), but low activity remained and was significantly higher than zero for the CMC-assay (V = 15, p = 0.0313) but not for the Megazyme-assay (V = 10, p = 0.0502). Also β-Glucosidase activity in the fecal fluid (Figure 1C) was significantly different between the two diets (W = 25, p = 0.0060), but with activity in the purged ants not being significantly different from zero (V = 6, p = 0.0907).

The cellulase activities were different among the layers of the fungus gardens with a trend of increased activity from top to bottom/debris material (Figure 2). The Megazyme-assay for endo-1,4-β-glucanase (Figure 2A) showed significantly lower activity in the top and middle layer 2 compared to the bottom layer and debris material (overall: χ 3 = 85.979, p < 0.0001, see Figure 2 for post-hoc details). The CMC-assay for endo-1,4-β-glucanase (Figure 2B) showed similar results but only the activity in debris material was significantly higher than the middle layer (overall: 2 χ 3 = 11.480, p = 0.0094, see Figure 2 for post-hoc details). β-Glucosidase activity (Figure 2C) showed larger differences comparable to those for endo-1,4-β-glucanase in the Megazyme essay, with significantly higher activity in the bottom layer and debris material compared to the top layer 2 (overall: χ 3 = 15.789, p = 0.0013, see Figure 2 for post-hoc details).

To better understand the significance of the cellulase activities in the fecal fluid and be able to compare activities to those found in earlier studies of fecal fluid pectinases (Schiøtt et al. 2010) and proteases (Kooij et al. in press) we measured the overall activities of all enzymes important for plant material degradation with AZCL enzyme assays. This showed significant differences among the enzyme groups (Figure 3, F5,70 = 68.108, p < 0.0001), with pectinases having the

142 25 A 20 mU /FD 15 10 (Megazyme assay) 5 Endo-1,4- β -glucanase activity 0 Figure 1 B 800 Enzyme activities for endo-1,4-β-glucanase (from two different

mU /FD assays) and β-glucosidase in fecal fluid of ants provided with

600 either a normal fungus garden diet or with sucrose water for 20 days. Wilcoxon rank sum test showed a significant reduction in

400 enzyme activity for endo-1,4-β-glucanase (A: Megazyme assay, (CMC-assay) W = 23, p = 0.016; B: CMC-assay, W = 22, p = 0.028) and

200 β-glucosidase (C: W = 25, p = 0.006) when ants were deprived

endo-1,4- β -glucanase activity of fungus garden material. Enzyme activity levels are given as milliunits per fecal droplet. 120 C 100 mU /FD 80 60 40 20 β - Glucosidase activity 0

Fungal diet Sucrose water highest activity, followed by amylases and proteases, and cellulases and hemicellulases having lowest activities. Other enzymes targeting the substrates chitosan, dextran and pullulan showed no activity at all.

Mass spectrometry analysis of fecal fluid from Acromyrmex echinatior (Schiøtt et al. 2010; De Fine Licht et al. 2013; Kooij et al. in press) identified a peptide sequence (SYTNLNLNANAQR) that matched the Lgcel12 gene in the genome sequence of Leucoagaricus gongylophorus, which harboured the amino acid sequence SYTNLNLNANLNK. The two sequences deviate in the last three amino acids, but as both combinations gave the exact same peptide mass, this discrepancy can be attributed to difficulties in interpreting the mass spectra. BLAST search allowed us to

143 T A B C op A AB A Middle

A A AB Bottom

B AB B Deb r B B B is

0 5 10 15 20 100 200 300 400 500 600 700 0 100 200 300 400 Endo-1,4-β-glucanase activity U/g Endo-1,4-β-glucanase activity U/g β−Glucosidase activity U/g (Megazyme assay) (CMC-assay)

Figure 2 Enzyme activities for endo-1,4-β-glucanase (from two different assays) and β-glucosidase in three layers of fungus garden (Top, Middle and Bottom) and debris material. (A) Endo-1,4-β-glucanase activity (Megazyme assay) was signifi- 2 cantly higher in the bottom layer of the fungus garden and in the debris material (overall: Kruskal-Wallis χ 3 = 85.976, p < 0.0001). (B) Endo-1,4-β-glucanase activity (CMC-assay) was significantly higher in the debris material compared 2 to the middle layer (overall: Kruskal-Wallis χ 3 = 11.480, p < 0.01). (C) β-Glucosidase activity was significantly higher 2 in the bottom layer and debris material compared to the top layer (overall: Kruskal-Wallis χ 3 = 15.789, p < 0.001). En- zyme activity levels are given as units per gram fungus garden material. Different letters above bars refer to differences in post-hoc tests at P < 0.05.

assign the Lgcel12 gene to glycosyl hydrolase family 12. This protein was the only one in the obtained fecal fluid proteome that could function as a cellulase, so we inferred that this enzyme mediates the cellulose activity measured in the fecal droplets. Relative changes in Lgcel12 gene expression in gongylidia versus normal mycelium (fold-changes: 2-∆∆Ct) showed that the gene was significantly more expressed (fold change = 3.1, min – max SE = 0.33 – 5.82) in the gongylidia than in normal mycelium (Figure 4A, z = -6.358, p < 0.00001), suggesting that the ingestion of this enzyme by the ants for transfer to the fecal fluid represents a mutual adaptation that somehow enhances symbiotic efficiency. In order to correlate the cellulase activity measurements in the fungus garden with the expression of Lgcel12, we measured the normalized expression levels of this gene (Figure 4B) in the same layers of fungus garden that were used for the activity measurements (Figure 2). This showed significantly higher normalized gene expression in the middle layer

(F3,16 = 3.699, p = 0.0339) compared to the other layers and the debris material, consistent with the middle layer having the largest amount of gongylidia (De Fine Licht et al. 2013).

144 3.5 A

3.0 B B Figure 3 2 2.5 Enzyme activity of fecal fluid for six different enzyme classes expressed as the mean area (cm2 ± SE) of coloured halos on 2.0 C C AZCL plates. Differences in activity between the enzyme class-

activity in cm 1.5 es were significant (F5,70 = 68.108, p < 0.0001), with pectinases being the most active, followed by amylases and proteases 1.0 Enzyme and cellulases and hemicellulases being least active (different 0.5 letters indicate differences in post-hoc tests at P < 0.05). The D 0.0 substrates chitosan, dextran and pullulan were grouped together in the class “other”, as no activity was observed for any of these. ylases Other ectinases P Proteases Am Cellulases Hemicellulases

Discussion Since Martin and Weber (1969) found similar cellulose degrading capabilities in the fungal symbiont and fecal fluid of Atta leaf-cutting ants (see also Martin et al. 1975), the joint cellulose degradation capacity of the ant fungus-farming mutualism has remained controversial. Cellulose degrading capabilities were later confirmed for the fungus (Bacci et al. 1995a) and for bacteria in the fungus garden (Bacci et al. 1995b), but the level of cellulose degradation was found to be relatively low (De Siqueira et al. 1998). The debate continued with different techniques giving varying results, some showing that the fungus was incapable of utilizing cellulose as a substrate (Abril and Bucher 2002; Abril and Bucher 2004; Bucher et al. 2004), but most studies indicating mediocre (D’Ettorre et al. 2002; Silva et al. 2006; Erthal et al. 2009; Kooij et al. 2011; Moller et al. 2011) or relatively high cellulose degradation capabilities (Rønhede et al. 2004; Richard et al. 2005; Schiøtt et al. 2008; De Fine Licht et al. 2010; Suen et al. 2010; Nagamoto et al. 2011; Ayl- ward et al. 2013; Grell et al. 2013) without reaching consensus.

The degradation of plant cell wall cellulose is a tedious process, which involves at least three types of cellulases in order to degrade this polysaccharide to single glucose molecules: 1. endo-cellulases, which open up the matrix of the cellulose fibres and cleave the fibres at random places, 2. exo-cellulases, which cut off short cellulose fragments mainly from the ends of the cellulose chain, and 3. β-glucosidases that break down these smaller chains into single glucose molecules (Béguin 1990; Martin et al. 1991). We show here that both endo-1,4-β-glucanase and β-glucosidase activity increase towards the bottom layer of the fungus garden and in the debris

145 A B 1.2 10.0 A ) 5.0 1.0 ) -ΔCt -ΔΔCt 0.8 2.0 old−change (2 0.6 1.0 xpression in gongylidia (2 e 0.5 0.4

Relative expression f B Normalized B 0.2 0.2 0.0 0.1 Top Middle Bottom Debris

Figure 4 Gene expression in gongylidia relative to mycelium and normalized gene expression after scaling for the expression of housekeeping genes in fungus garden layers and debris piles. (A) The Lgcel12 gene is upregulated in gongylidia compared to mycelium with an average 3.1x fold change (0.330 - 5.819 min - max SE, z = -6.358, p < 0.00001). (B) Normalized gene expression of the Lgcel12 gene was significantly higher in the middle layer of the fungus garden compared to any other part of the

fungus garden (F3,16 = 3.699, p = 0.034). Different letters above bars refer to differences in post-hoc tests at P < 0.05. material (Figure 2), which is consistent with earlier findings (Suen et al. 2010; Moller et al. 2011; Aylward et al. 2013; Grell et al. 2013). Although explicit data on exocellulase activity are lacking, this suggest that the fungus garden has the ability to convert cellulose into glucose molecules that can be taken up as nutrients by fungal cells, but that this activity is mostly expressed towards the end of the substrate degradation process when other more easily accessible nutrients have already been decomposed. Although many studies have pointed out that high cellulolytic activity in the bottom of fungus gardens may also be due to bacteria (Bacci et al. 1995b; Suen et al. 2010; Ayl- ward et al. 2012) or yeasts (Carreiro et al. 1997; Middelhoven et al. 2003; Mendes et al. 2012), the present study and recent results by Aylward et al. (2013) and Grell et al. (2013) are consistent in suggesting that a significant fraction of cellulose decomposition in the bottom of gardens is indeed performed by the fungal symbiont. However, this appears to be less likely for the debris material discarded by the ants, as most of the fungal tissue should have died when the ants stop maintaining it, so only yeasts and bacteria are left.

146 In addition to confirming cellulolytic activity towards the end of the decomposition process, our present study shows that early stages of cellulose decomposition are an explicit target in the top of fungus gardens. A single putative endo-cellulase enzyme (LgCel12) of fungal origin apparently escapes degradation in the ant gut to end up in the fecal fluid where it remains active (Figure 1 and 3). Although our earlier mass spectrometry results (Schiøtt et al. 2010) did not indicate the presence of any β-glucosidases we nevertheless found these enzymes to be active in the fecal fluid and to originate from the fungal symbiont as well (Figure 1). The expression of the Lgcel12 gene was also upregulated in the gongylidia relative to the normal mycelium, indicating that this enzyme has been selected to be ingested by the ants to function either in the ant gut or in the fecal fluid, as has also been shown for many other fecal fluid enzymes (Schiøtt et al. 2010; De Fine Licht et al. 2013; Kooij et al. in press). Overall fungal expression of the Lgcel12 gene was highest in the middle section of the gardens (Figure 4), consistent with gardens producing most of the gongylidia when fungal biomass is abundant and substrate not yet depleted (De Fine Licht et al. 2013).

Endo-cellulase enzyme (LgCel12) activity in the top of gardens would likely serve the purpose of loosening up the cell walls in the leaf pulp macerated by the ants to allow hyphal access to the cell interior with proteins and starch (Schiøtt et al. 2010; Moller et al. 2011; Grell et al. 2013; Kooij et al. in press). However, we also obtained evidence for β-glucosidase activity in the fecal fluid, which would produce free glucose from oligosaccharides that could be absorbed by the mycelium. Recently, more advanced mass spectrometry (Schiøtt et al., unpublished) has now also shown exocellulases in the fecal fluid of A. echinatior, indicating that all three steps of cellulose degradation take place in the top of fungus gardens. However, compared to the activity of other fungal enzymes such as proteinases, pectinases and amylase, fecal fluid cellulolytic activity appears to remain relatively low in all three steps (endocellulases, exocellulases, β-glucosidases)(Figure 3).

Overall, our results suggest a scenario in which the fungal symbiont is capable to degrade cellulose but only at a relatively slow rate and not primarily for maximizing glucose production for the symbiosis. As the bottom of fungus gardens hardly produces gongylidia (De Fine Licht et al. 2013), any glucose produced here remains unavailable to the ants and is thus unlikely to benefit the symbiosis. As suggested by Grell et al. (2013), the main function of glucose production at this stage may be to prevent structural fungus-garden collapse and pathogen invasion until the ants discard this otherwise depleted material. At the top of gardens modest endocellulase activity may be required to supplement xylanase breakdown (Schiøtt et al. 2008) to open up plant cells. Proportional production of exocellulases and β-glucosidases at that stage would then merely make sure that complex sugars can benefit fungal growth rather than imposing risks of bacterial growth

147 that may damage symbiotic efficiency. This hypothetical framework would remain consistent with fungus-garden activity of leaf-cutting ants being ultimately focused on maximizing protein acquisition per unit of foraging and processing effort by the ants in spite of this leaving behind considerable amount of non-degraded cellulose material that is discarded by the ants from the bottom layers of fungus gardens (Schiøtt et al. 2008; De Fine Licht et al. 2010; Moller et al. 2011; Grell et al. 2013; Kooij et al. in press).

This interpretation follows the general idea that decomposition of organic matter takes place in a stoichiometric manner, which keeps ratios of liberated carbon, nitrogen and phosphorus relatively constant and releases any surplus (non-limiting) nutrients into the external environment (Man- zoni et al. 2010; Hartman and Richardson 2013; Sinsabaugh et al. 2013). It is well documented that plant diets offer an excess of carbon relative to nitrogen and phosphorus (Manzoni et al. 2010; Bell et al. 2014), so that decomposers (in this case the symbiotic collaboration between leaf-cutting ants and fungal symbiont) should be under selection to invest primarily in enzymes for acquiring the most limiting nutrients (nitrogen and possibly phosphorus), and to adjust carbon decomposition/liberation to the level needed for achieving stoichiometric equilibrium. Indications of the leaf-cutting ant symbiosis to be nitrogen-limited have emanated from the demonstration of nitrogen-fixing bacteria in the fungus gardens of Atta leaf-cutting ants (Pinto-Tomás et al. 2009) and other nitrogen conserving bacteria in the guts of both Atta and Acromyrmex leaf-cutting ants (Sapountzis et al. in review). Depending on the efficiency of these additional nitrogen preserving bacterial symbionts, it may thus turn out that the availability of phosphorus in the plant substrate is ultimately the most limiting factor for the leaf-cutting ant symbiosis, but much further work will be needed to substantiate these hypotheses.

Acknowledgements We thank the Smithsonian Tropical Research Institute (STRI), Panama, for providing logistic help and facilities to work in Gamboa, and the Autoridad Nacional del Ambiente y el Mar (ANAM) for permission to sample ant colonies in Pan- ama, the Danish National Research Foundation (DNRF57) and the ERC (323085) for funding.

References Aanen DK, Eggleton P, Rouland-Lefevre C, et al. (2002) The evolution of fungus-growing termites and their mutualistic fungal symbionts. P Natl Acad Sci USA 99:14887–14892. doi: 10.1073/pnas.222313099 Abril AB, Bucher EH (2002) Evidence that the fungus cultured by leaf-cutting ants does not metabolize cellulose. Ecol Lett 5:325–328. Abril AB, Bucher EH (2004) Nutritional sources of the fungus cultured by leaf-cutting ants. Appl Soil Ecol 26:5–5. doi: 10.1016/j.apsoil.2003.12.008

148 Aylward FO, Burnum-Johnson KE, Scott JJ, et al. (2012) Metagenomic and metaproteomic insights into bacterial communities in leaf-cutter ant fungus gardens. ISME J 6:1688–1701. doi: 10.1038/ismej.2012.10 Aylward FO, Burnum-Johnson KE, Tringe SG, et al. (2013) Leucoagaricus gongylophorus produces diverse enzymes for the degradation of recalcitrant plant polymers in leaf-cutter ant fungus gardens. Appl Environ Microb 79:3770–3778. doi: 10.1128/AEM.03833-12 Bacci M Jr, Anversa M, Pagnocca FC (1995a) Cellulose degradation by Leucocoprinus gongylophorus, the fungus cultured by the leaf-cutting ant Atta sexdens rubropilosa. Anton Leeuw Int J G 67:385–386. Bacci M Jr, Ribeiro SB, Casarotto M, Pagnocca FC (1995b) Biopolymer-degrading bacteria from nests of the leaf- cutting ant Atta sexdens rubropilosa. Braz J Med Biol Res 28:79–82. Bass M, Cherrett JM (1995) Fungal hyphae as a source of nutrients for the leaf-cutting ant Atta sexdens. Physiol Entomol 20:1–6. Bell C, Carrillo Y, Boot CM, et al. (2014) Rhizosphere stoichiometry: are C : N : P ratios of plants, soils, and enzymes conserved at the plant species-level? New Phytol 201:505–517. doi: 10.1111/nph.12531 Béguin P (1990) Molecular biology of cellulose degradation. Annu Rev Microbiol 44:219–248. doi: 10.1146/annurev.mi.44.100190.001251 Biedermann PHW, Klepzig KD, Taborsky M (2009) Fungus cultivation by ambrosia beetles: behavior and laboratory breeding success in three xyleborine species. Environ Entomol 38:1096–1105. doi: 10.1603/022.038.0417 Breznak JA (1982) Intestinal microbiota of termites and other xylophagous insects. Annu Rev Microbiol 36:323–343. doi: 10.1146/annurev.mi.36.100182.001543 Bucher EH, Marchesini V, Abril AB (2004) Herbivory by leaf-cutting ants: Nutrient balance between harvested and refuse material. Biotropica 36:327–332. doi: 10.1111/j.1744-7429.2004.tb00324.x Carreiro SC, Pagnocca FC, Bueno OC, et al. (1997) Yeasts associated with nests of the leaf-cutting ant Atta sexdens rubropilosa Forel, 1908. Anton Leeuw Int J G 71:243–248. D’Ettorre P, Mora P, Dibangou V, et al. (2002) The role of the symbiotic fungus in the digestive metabolism of two species of fungus-growing ants. J Comp Physiol B 172:169–176. doi: 10.1007/s00360-001-0241-0 De Fine Licht HH, Boomsma JJ (2010) Forage collection, substrate preparation, and diet composition in fungus- growing ants. Ecol Entomol 35:259–269. doi: 10.1111/j.1365-2311.2010.01193.x De Fine Licht HH, Schiøtt M, Mueller UG, Boomsma JJ (2010) Evolutionary transitions in enzyme activity of ant fungus gardens. Evolution 64:2055–2069. doi: 10.1111/j.1558-5646.2010.00948.x De Fine Licht HH, Schiøtt M, Rogowska-Wrzesinska A, et al. (2013) Laccase detoxification mediates the nutritional alliance between leaf-cutting ants and fungus-garden symbionts. P Natl Acad Sci USA 110:583–587. doi: 10.1073/pnas.1212709110 De Siqueira CG, Bacci M Jr, Pagnocca FC, et al. (1998) Metabolism of plant polysaccharides by Leucoagaricus gongylophorus, the symbiotic fungus of the leaf-cutting ant Atta sexdens L. Appl Environ Microb 64:4820–4822. Douglas AE (2013) Microbial brokers of insect-plant interactions revisited. J Chem Ecol 39:952–961. doi: 10.1007/s10886-013-0308-x Erthal M Jr, Silva CP, Cooper RM, Samuels RI (2009) Hydrolytic enzymes of leaf-cutting ant fungi. Comp Biochem Phys B 152:54–59. doi: 10.1016/j.cbpb.2008.09.086 Fowler HG, Pagani MI, Da Silva OA, et al. (1989) A pest is a pest is a pest? The dilemma of Neotropical leaf-cutting ants: keystone taxa of natural ecosystems. Environ Manage 13:671–675. Giraudoux P (2013) pgirmess: Data analysis in ecology. Available: http://CRAN.R-project.org/package=pgirmess

149 Grell MN, Nielsen KL, Linde T, et al. (2013) The fungal symbiont of leaf-cutter ants expresses the full enzyme complement to degrade cellulose and other plant polymers. BMC Genomics 14:928. doi: 10.1186/1471-2164-14-928 Grünwald S, Pilhofer M, Höll W (2010) Microbial associations in gut systems of wood- and bark-inhabiting longhorned beetles [Coleoptera: Cerambycidae]. Syst Appl Microbiol 33:25–34. doi: 10.1016/j.syapm.2009.10.002 Hartman WH, Richardson CJ (2013) Differential nutrient limitation of soil microbial biomass and metabolic quotients (qCO2): Is there a biological stoichiometry of soil microbes? PLoS ONE 8:e57127. doi: 10.1371/journal.pone.0057127.s024 Hothorn T, Bretz F, Westfall P (2008) Simultaneous inference in general parametric models. Biometrical J 50:346–363. Kacuráková M, Capek P, Sasinková V, et al. (2000) FT-IR study of plant cell wall model compounds: pectic polysaccharides and hemicelluloses. Carbohyd Polym 43:195–203. doi: 10.1016/S0144-8617(00)00151-X Kohl M, Perkins J (2011) NormqPCR: Functions for normalisation of RT-qPCR data. Available: http://www.bioconductor.org/packages/devel/bioc/html/NormqPCR.html Kooij PW, Aanen DK, Schiøtt M, Boomsma JJ (Chapter 1) Advanced farming ants rear polyploid crop fungi. Kooij PW, Liberti J, Giampoudakis K, et al. (Chapter 3, submitted) Differences in forage-acquisition and fungal enzyme activity contribute to niche segregation in Panamanian leaf-cutting ants. Kooij PW, Rogowska-Wrzesinska A, Hoffmann D, et al. (in press) Leucoagaricus gongylophorus uses leaf-cutting ants to vector proteolytic enzymes towards new plant substrate. ISME J Kooij PW, Schiøtt M, Boomsma JJ, De Fine Licht HH (2011) Rapid shifts in Atta cephalotes fungus-garden enzyme activity after a change in fungal substrate (Attini, Formicidae). Insect Soc 58:145–151. doi: 10.1007/s00040-010-0127-9 Livak KJ, Schmittgen TD (2001) Analysis of relative gene expression data using real-time quantitative PCR and the 2(-Delta Delta C(T)) Method. Methods 25:402–408. doi: 10.1006/meth.2001.1262 Della Lucia TMC, Gandra LC, Guedes RNC (2014) Managing leaf-cutting ants: Peculiarities, trends and challenges. Pest Manag Sci 70:14–23. doi: 10.1002/ps.3660 Manzoni S, Trofymow JA, Jackson RB, Porporato A (2010) Stoichiometric controls on carbon, nitrogen, and phosphorus dynamics in decomposing litter. Ecol Monogr 80:89–106. Martin MM, Boyd ND, Gieselmann MJ, Silver R (1975) Activity of faecal fluid of a leaf-cutting ant toward plant cell wall polysaccharides. J Insect Physiol 21:1887–1892. Martin MM, Jones CG, Bernays EA (1991) The evolution of cellulose digestion in insects [and discussion]. Philos T Roy Soc B 333:281–288. doi: 10.1098/rstb.1991.0078 Martin MM, Weber NA (1969) The cellulose-utilizing capability of the fungus cultured by the Attine ant Atta colombica tonsipes. Ann Entomol Soc Am 62:1386–1387. Mendes TD, Rodrigues A, Dayo-Owoyemi I, et al. (2012) Generation of nutrients and detoxification: possible roles of yeasts in leaf-cutting ant nests. Insects 3:228–245. doi: 10.3390/insects3010228 Middelhoven WJ, Fonseca A, Carreiro SC, et al. (2003) Cryptococcus haglerorum, sp. nov., an anamorphic basidiomycetous yeast isolated from nests of the leaf-cutting ant Atta sexdens. Anton Leeuw Int J G 83:167–174. Moller IE, De Fine Licht HH, Harholt J, et al. (2011) The dynamics of plant cell-wall polysaccharide decomposition in leaf-cutting ant fungus gardens. PLoS ONE 6:e17506. doi: 10.1371/journal.pone.0017506 Mueller UG, Gerardo NM, Aanen DK, et al. (2005) The evolution of agriculture in insects. Annu Rev Ecol Evol S 36:563–595. doi: 10.1146/annurev.ecolsys.36.102003.152626

150 Nagamoto NS, Garcia MG, Forti LC, et al. (2011) Microscopic evidence supports the hypothesis of high cellulose degradation capacity by the symbiotic fungus of leaf-cutting ants. J Biol Res-Thessalon 16:308–312. Noda S, Kitade O, Inoue T, et al. (2007) Cospeciation in the triplex symbiosis of termite gut protists (Pseudotrichonympha spp.), their hosts, and their bacterial endosymbionts. Mol Ecol 16:1257–1266. doi: 10.1111/j.1365-294X.2006.03219.x Nygaard S, Zhang G, Schiøtt M, et al. (2011) The genome of the leaf-cutting ant Acromyrmex echinatior suggests key adaptations to advanced social life and fungus farming. Genome Res 21:1339–1348. doi: 10.1101/gr.121392.111 Ohkuma M (2003) Termite symbiotic systems: efficient bio-recycling of lignocellulose. Appl Microbiol Biot 61:1–9. doi: 10.1007/s00253-002-1189-z Perkins J, Kohl M (2011) ReadqPCR: Read qPCR data. Available: http://www.bioconductor.org/packages/2.14/bioc/html/ReadqPCR.html Pinheiro J, Bates D, DebRoy S, et al. (2013) nlme: Linear and Nonlinear Mixed Effects Models. Available: http://cran.r-project.org/web/packages/nlme/index.html Pinto-Tomás AA, Anderson MA, Suen G, et al. (2009) Symbiotic nitrogen fixation in the fungus gardens of leaf-cutter ants. Science 326:1120–1123. doi: 10.1126/science.1173036 Quinlan R, Cherrett JM (1978) Aspects of the symbiosis of the leaf-cutting ant Acromyrmex octospinosus (Reich) and its food fungus. Ecol Entomol 3:221–230. Quinlan R, Cherrett JM (1979) The role of fungus in the diet of the leaf-cutting ant Atta cephalotes (L.). Ecol Entomol 4:151–160. R Core Team (2013) R: A language and environment for statistical computing. Available: http://www.R-project.org/ Richard F-J, Mora P, Errard C, Rouland-Lefevre C (2005) Digestive capacities of leaf-cutting ants and the contribution of their fungal cultivar to the degradation of plant material. J Comp Physiol B 175:297–303. doi: 10.1007/s00360-005-0485-1 Rodrigues A, Carletti CD, Bueno OC, Pagnocca FC (2008) Leaf-cutting ant faecal fluid and mandibular gland secretion: effects on microfungi spore germination. Braz J Microbiol 39:64–67. Rønhede S, Boomsma JJ, Rosendahl S (2004) Fungal enzymes transferred by leaf-cutting ants in their fungus gardens. Mycol Res 108:101–106. doi: 10.1017/S0953756203008931 Sapountzis P, Hansen LH, Sørensen SJ, et al. (in review) Phylogenetic distribution, stability and function of gut bacterial endosymbionts in fungus-growing ants. Schiøtt M, De Fine Licht HH, Lange L, Boomsma JJ (2008) Towards a molecular understanding of symbiont function: Identification of a fungal gene for the degradation of xylan in the fungus gardens of leaf-cutting ants. BMC Microbiol 8:40. doi: 10.1186/1471-2180-8-40 Schiøtt M, Rogowska-Wrzesinska A, Roepstorff P, Boomsma JJ (2010) Leaf-cutting ant fungi produce cell wall degrading pectinase complexes reminiscent of phytopathogenic fungi. BMC Biol 8:156. doi: 10.1186/1741-7007-8-156 Schmittgen TD, Zakrajsek BA (2000) Effect of experimental treatment on housekeeping gene expression: validation by real-time, quantitative RT-PCR. J Biochem Bioph Meth 46:69–81. Silva A, Bacci M Jr, Pagnocca FC, et al. (2006) Production of polysaccharidases in different carbon sources by Leucoagaricus gongylophorus Moller (Singer), the symbiotic fungus of the leaf-cutting ant Atta sexdens Linnaeus. Curr Microbiol 53:68–71. doi: 10.1007/s00284-005-0431-1

151 Sinsabaugh RL, Manzoni S, Moorhead DL, Richter A (2013) Carbon use efficiency of microbial communities: stoichiometry, methodology and modelling. Ecol Lett 16:930–939. doi: 10.1111/ele.12113 Suen G, Scott JJ, Aylward FO, et al. (2010) An insect herbivore microbiome with high plant biomass-degrading capacity. PLoS Genet. doi: 10.1371/journal.pgen.1001129 Watanabe H, Tokuda G (2001) Animal cellulases. Cell Mol Life Sci 58:1167–1178. Weber NA (1966) Fungus-growing ants. Science 153:587–604. Zhang N, Suh S-O, Blackwell M (2003) Microorganisms in the gut of beetles: evidence from molecular cloning. J Invertebr Pathol 84:226–233. doi: 10.1016/j.jip.2003.10.002 Zhang YHP, Hong J, Ye X (2009) Chapter 14: Cellulase Assays. In: Mielenz JR (ed) Biofuels: methods and protocols. Humana PressInc, pp 213–231

152 Acknowledments 153 First of all I want thank the appr. 7000 (six-legged) girls who’s butt I was allowed to pinch, without them this thesis would not exist.

Special thanks go out for my two supervisors, Koos Boomsma and Morten Schiøtt, for all their support during the last three and a half years. You both have been great teachers and showed me how to see results in a bigger picture. Our discussions on various subjects have inspired me to think beyond the results as they were and reflect them to the greater aspects of the science of mutualisms.

Duur Aanen, without whom I would not have found the Centre of Social Evolution. My Masters project with you on fungus-growing termites led me to a great interest in fungus-growing insect systems and mutualisms in general. Also thank you for our work together on entangling the won- ders of the multiple nuclei in the ant fungi.

Michael Poulsen, for your expert help on incompatibility in ant-fungi, the long hours of squeezing ants, staring at fungi, and analyzing data. And off course thank you for the interesting discussions we had about fungus-growing ants, fungus-growing termites, and other topics, scientific or not.

My other co-authors, Adelina Rogowska-Wrzesinska, Peter Roepstorff and Daniel Hoffmann, for your help and expertise on various aspects of the proteases produced by the fungus and transport to the right place by the ants.

The Smithsonian Tropical Research Institute, Panama, for allowing me to do field work at the Gamboa field station and the scientific support there.

The Department of Biology, University of Copenhagen, for giving me this opportunity to perform and defend my PhD.

The Danish National Research Foundation for providing part of the money for this PhD.

I want to thank Henrik, for his great supervision during my internship at the Centre, which insti- gated me to apply for a PhD position here. Also thanks for the various discussions we had on the multitude of topics of fungus-growing ants.

My office mate, Luigi. I had a great time sharing my office with you these three and a half years,

154 from interesting discussions on politics, our both research topics or when to put up our Christmas decorations.

Bernhardt and Luigi, for the long evenings of shouting at each other while playing FIFA, and still find time together to discuss science.

Panos and Saria, for the after work beers at the end of these three years as a nice break in between the writing, and reminding me there are other topics in science then the ones described in this thesis.

And of course, Sylvia and Louise, the best lab-techs in the world! You both deserve a special place here. You help has been fantastic! Especially to Sylvia, for listening to my whining about dirty labs over and over and for the delicious banana-chocolate cakes.

My parents, Jos and José, for making it possible for me to study and make it possible for me to stand where I stand today. And my parents as well as my brothers and sisters, Taco, Eelke, Sanne and Jojanneke (and Jalke and Avine) for their support during these years.

My new family in Colombia, Edgar, Gladys, Paula and Juan, for welcoming me in their midst and their support.

My friends who I left in the Netherlands or who moved away themselves, Remco&Dorine, Roos&Freek, Silvan&Zuuz, Marijke, Jan-Willem, Ezra, Fam&Chris, Liset and others I might have forgot. As well as all the new friends I found, Kostis&Laetitia, Claudia&Francesco, Danka, Lilia&Martin, Rasmus, Marlene, Jesper, Caya, Vale, Lina&Lina, Alex, Vicky, and everyone else.

And everyone else at the Centre for Social Evolution, and the Section for Ecology and Evolution in Copenhagen, Denmark, and the Laboratory of Genetics in Wageningen, The Netherlands.

Poes, welcoming me every morning with a happy meow and for keeping my lap warm while writing.

And last but not least, Andrea, mi amor. Without this PhD we would probably never have met, and without you this PhD would have been the most boring ever. You bring the light in my life, and I will always be grateful for all your support and love during this years! I love you.

155