Honors Thesis

THE PENNSYLVANIA STATE UNIVERSITY SCHREYER HONORS COLLEGE

DEPARTMENT OF VETERINARY AND BIOMEDICAL SCIENCES

UNDERSTANDING OPHIOCORDYCEPS, THE ZOMBIE ANT FUNGUS: A CASE STUDY IN HOST BEHAVIORAL MANIPULATION

BENJAMIN FOWLER SPRING 2016

A thesis submitted in partial fulfillment of the requirements for baccalaureate degrees in Immunology and Infectious Disease & Toxicology with honors in Immunology and Infectious Disease

Reviewed and approved* by the following:

David P. Hughes Assistant Professor of Entomology and Biology Thesis Supervisor

Pamela A. Hankey Giblin Professor of Immunology Honors Adviser

* Signatures are on file in the Schreyer Honors College. i

ABSTRACT

Millions of years of evolution have led many pathogens to develop unique strategies to maximize their transmission. Behavioral manipulation is one of these strategies and has captured wide audiences with the idea of zombie-like mind control. Ophiocordyceps is a genus of fungal parasites that infect insects, leading to the zombie-ant phenomenon, in which an infected ant leaves its home colony, climbs onto foliage above the forest floor, and subsequently dies. The species complex O. unilateralis even causes infected ants to bite into the plant’s flesh prior to death. Previous work has shown that this manipulation is essential for successful maturation of the fungus, but studies have not evaluated the role this manipulation may play in transmission of the fungus to susceptible host ants. Chapter 2 of this thesis introduces a novel strategy to evaluate the transmission potential of a particular species of the zombie-ant fungus, O. camponoti-atricipis. Experiments using spore-clocks and gravitational grids indicated that the fungus tightly concentrates infectious ascospore release during the early morning hours with a clear peak between 0500 and 0600 hours and that the release of spores is focused downwards by gravitropic growth. Chapter 3 of this thesis covers efforts to use phylogenetic analyses to assess the host specificity of O. unilateralis sensu lato. Together, these studies of transmission and host selectivity provide insights into the evolution of both behavioral manipulation and host specificity that help understand and model potential zoonotic diseases of humans.

TABLE OF CONTENTS

LIST OF FIGURES ...... iii

ACKNOWLEDGEMENTS ...... v

Chapter 1 Introduction ...... 1

References ...... 6

Chapter 2 Transmission Potential ...... 8

Background ...... 8 Methods ...... 12 Sample Collection ...... 12 Physical Characterization ...... 13 Temporal Release Monitoring ...... 13 Results ...... 14 Fecundity and morphological characteristics of spore producing body ...... 14 Timing of spore release ...... 17 Discussion ...... 18 Conclusions ...... 22 References ...... 23

Chapter 3 Host Specificity and Fungal Diversity ...... 27

Background ...... 27 Methods ...... 30 Sample Collection ...... 30 DNA Extraction, Cleaning, PCR, Sequencing, and Phylogenetic Analysis...... 30 Discussion ...... 31 References ...... 33

Chapter 4 Conclusions ...... 37

Appendix A Attachment to Leaves ...... 40

Ophiocordyceps lloydii var. binata ...... 42 Ophiocordyceps unilateralis ...... 50 References ...... 53

Academic Vita ...... 54

iii

LIST OF FIGURES

Figure 1: A dead ant attached to a palm leaf in the Amazon with the fungus Ophiocordyceps camponoti-atricipis growing from its head ...... 4

Figure 2: Morphology of O. camponoti-atricipis. A) O. camponoti-atricipis specimen, the red arrow indicates the ascoma and the black brace indicates measured length to ascoma. B) Composite image of the internal and external morphology of the ascoma. An ostiole is indicated with a black arrow and a perithecium is indicated with a red arrow. C) A perithecium removed from an ascoma on a 25G needle (bar = 100 microns). D) Cluster of asci as they appear in a smashed perithecium, 100x. E) Ascus consisting of eight ascospores, 100x. F) Individual ascospore, 100x...... 11

Figure 3: Morphological and reproductive quantification of O. camponoti-atricipis. A) The distance of the fungal stalk (stroma) from the ant cadaver to the fruiting body (ascoma). The mean length to ascoma is 3.97 mm (n=40, s.d. 2.52 mm, range 1.03-11.5 mm). B) Inner circle (red): Growth of the fungal stalk. Stromal growth clustered around directly downwards. The mean distance from vertical is 25.8° (n=19, s.d. 21.1°, range 1.5-76.6°). Outer circle (blue): Direction of ant. The direction the ant faced in the same plane as the fungus was random and had no association with the angle of stromal growth. C) The number of ostioles present on a single ascoma. The mean ostiole count is 185 (n=10, s.d. 109, range 66-413). D) The mean count of the ascospores contained in a single perithecium is 2,212 (n=34, s.d. 761, range 719-4282)...... 16

Figure 4: Hourly quantification of ascospores released by O. camponoti-atricipis. After either 12 hours (BD15_5, BD15_6, & BD15_7) or 24 hours (BD15_1, BD15_23, BD15_19, BD15_10, BD15_28, the number of ascospores released each hour was counted. Estimated values for specimens with indistinct hourly patches or such dense spore release that individual ascospores were indistinguishable are designated with dotted lines. The mean count of released ascospores is noted with a dashed line...... 17

Figure 5: Ophiocordyceps lloydii var. binata. Note the light tan mycelial growth covering the ant's body and attaching it to the leaf (red circle)...... 41

Figure 6: Higher magnification O. lloydii var. binata. Note that the ant's mandibles (red arrow) are not biting leaf tissue and that fungal growth is extensive around the base of the legs (red circle)...... 41

Figure 7: Ophiocordyceps lloydii var. binata mandibles. Note the clustering of fungal growth around the mouthparts (red circle)...... 43

Figure 8: O. lloydii var. binata mycelium growing from the base of the legs viewed from the basal surface after removal of the leaf. Note how the growth radiates outwards...... 44

Figure 9: Higher magnification of the basal side of O. lloydii var. binata. Note the tubular mesh with a membrane (red arrow) between the strands...... 45 iv

Figure 10: Even higher magnification of the basal side of O. lloydii var. binata. Note the mycelium on the right and the tear in the membrane (red arrow) which reveals empty space below the membrane...... 46

Figure 11: Ophiocordyceps lloydii var. binata growth on a leaf surface. Note the smooth tubular growths...... 47

Figure 12: The interface between the surface of a leaf (left, rough ~1 micron patches) and Ophiocordyceps lloydii var. binata growth (right, smooth tubular strands)...... 48

Figure 13: Leaf surface without O. lloydii var. binata growth. Note the rough appearance, particularly in comparison to Figure 11...... 49

Figure 14: O. unilateralis basal surface after leaf was removed at magnification roughly equivalent to Figure 8...... 51

Figure 15: O. unilateralis basal surface at higher magnification (similar to Figure 9). Note the lack of a membrane between mycelial strands...... 52

ACKNOWLEDGEMENTS

First and foremost, I would like to acknowledge David, my PI and thesis supervisor. I joined the Hughes Lab as he was finishing a major grant proposal, and so worked in the lab for over a month before I actually met the mystical man for whom I worked. After that, David was with me at every step, from my first undergraduate grant proposal to my sixth, from the evolution of my project creating a fungal database to microscopic evaluations to sequencing fungal genes. His rephrasing of my naïve questions into actual research questions was invaluable and I could not, nor would I want to, produce this thesis without him. By the way, I need your signature again.

Second, I would like to acknowledge João. When I volunteered to help him with a project in the spring of 2014, I never dreamed that less than a year later we would mark my 21st birthday in the Brazilian Amazon with a 24-hour observation-based experiment. João, I learned more about fungi and fieldwork from you than I ever thought possible, and even though I still cannot say your name right, I am so glad you asked for help two years ago. Thank you.

To the rest of the lab, at some point I pestered each and every one of you with questions or random brain droppings. Thank you for helping me refine my work, teaching me new techniques, and for letting me give input into your work. The opportunity to collaborate with so many people on so many different projects has been an invaluable experience.

Finally, I would like to acknowledge John and Missy of the Penn State Microscopy and

Cytometry Facility – University Park, PA for training me on instruments and helping me gather the data I needed to make this project a reality. Thank you very much.

Chapter 1

Introduction

“The chicken is only an egg’s way of making another egg.” – Richard Dawkins

Evolution is the key to biology; reproduction is the key to evolution. The perfect organism, if it produces no offspring, will die and disappear, forgotten by time. However, if conditions dictate that an individual survives to produce more offspring than a rival does (e.g. has greater fitness), the characteristics that endow that selective advantage will persist in the population. In the most basic biological sense, the sole purpose of life is to reproduce (Darwin,

1859). Nowhere is this more evident than in the biology of pathogens. Parasites exist to make more parasites, and, with reproduction as the impetus, parasites have evolved fascinating strategies to maximize their proliferation. One of the most amazing and intricate strategies is the manipulation of host behavior. Some pathogens, upon infecting a susceptible host, are able to induce a behavioral change that somehow increases the parasite’s fitness. This alteration is generally understood to increase the success of transmission (Poulin, 1998) thus increasing parasite fitness.

In the case of manipulation by a parasite, host behavior changes are not the reaction of an infected organism to disease, rather, they are the expression of an extended phenotype by the pathogen. The extended phenotype is the concept that genes, the entity causing any phenotype, have an impact that extends outside the organism possessing and expressing the gene. In other 2 words, genes can affect the environment around them (Dawkins, 1982). For example, mice from the genus Peromyscus dig burrows with an entrance tunnel and sometimes an exit tunnel. The length and presence or absence of these tunnels varies based upon the genes present in four distinct genetic loci (Weber et. al, 2013). Similarly, genes of manipulating pathogens are responsible for the occurrence and severity of host behavioral changes. Numerous pathogens are able to control host behavior.

Toxoplasma gondii is a protozoan parasite that causes toxoplasmosis in many mammals, including humans; however, the target hosts of T. gondii are members of the feline family, from domestic cats to lions (Cheadle et. al, 1999). There is an abundance of evidence showing that T. gondii has a significant impact on the behavior of its non-felid, non-target hosts. These behaviors are best studied in rats such as Rattus norvegicus. Healthy rats display marked aversions to the scent of cats, but rats infected with this parasite do not display these avoidance behaviors, and in fact are bolder and spend more time in new environments with unknown dangers (Berdoy et. al,

2000). This suggests that T. gondii is manipulating non-target host behavior to make predation by, and thus transmission to, its target host, felines, more likely. This manipulation has been documented in T. gondii-infected chimpanzees, which showed a correlative attraction to leopard urine in a recent study (Poirette et. al, 2016) and some studies even show correlations between infection with T. gondii and schizophrenia in humans (Torrey & Yolken, 2003). Other manipulated behaviors are also a result of infection, such as the altered web weaving of

Plesiometa argyra when parasitized by the larval wasp Hymenoepimecis sp. Female

Hymenoepimecis wasps attack and inject a single egg into the body of the orb-weaving spider, P. argyra. When the parasitic larva is ready to mature, it somehow forces the spider to change its web-spinning pattern from a circular orb to a much smaller trapezoidal cocoon (Eberhard, 2000). 3 This altered web-spinning behavior creates a structure to which the larva can attach its cocoon while it matures.

Fungi in the genus Ophiocordyceps also manipulate the behavior of their hosts, ants of various genera (Andersen et. al, 2009). Members of the species complex Ophiocordyceps unilateralis are the most common and they infect ants of the genus Camponotus, or carpenter ants (Evans et. al, 2011a). Carpenter ants are found worldwide and are large ants, some reaching nearly 1.5 cm in length. They feed upon sugar and protein and live in nests carved out of wood

(Cranshaw & Redek, 2013). O. unilateralis is also widely dispersed throughout the tropical regions of the planet, although members of the complex are also found less frequently in temperate ecosystems such as the U.S. states South Carolina and Florida (de Bekker et. al, 2014).

For the first days to several weeks of infection with O. unilateralis, it is thought that infected ants show neither outward signs of infection nor any behavioral changes. Behavioral manipulation only occurs later, between 14 and 22 days post infection under laboratory conditions (de Bekker et. al, 2014). During manipulation, ants behave “drunkenly”, appearing to be unable to remain steady on their legs and occasionally falling from the surface upon which they are walking. During this erraticism, infected ants venture onto elevated leaves or twigs, where healthy ants are not normally found. Upon yet undetermined stimuli, the ant bites into the leaf or twig and remains there (for more information about this biting behavior and other attachment strategies see Appendix 1). The mandibular muscles undergo extensive atrophy and this degradation makes the attachment long lasting (Hughes, et. al, 2011a). The biting behavior of so-called zombie ants is extremely forceful and leaves distinctive impressions on leaf tissue, which are present on 48 million year old leaf fossils (Hughes, et. al, 2011b). The ability to manipulate ant behavior is a vital component of the O. unilateralis life cycle. 4

Figure 1: A dead ant attached to a palm leaf in the Amazon with the fungus Ophiocordyceps camponoti-atricipis growing from its head After guiding the ant to its death on an elevated platform, O. unilateralis proceeds to grow from the dorsal pronotum of the ant. It forms a long narrow stalk, called a stroma, upon which a sexual reproductive body, called an ascoma, forms (Figure 1). Within the ascoma infectious particles called ascospores form in clusters of eight, called asci. Ascospores are not resistant to degradation upon exposure to the environment outside of the ascoma and likely persist for less than 72 hours (Evans & Samson, 1984). As discussed earlier, reproduction is essential to the success of a parasite. Somehow, these infectious ascospores must bridge the gap from their production site in the ascoma elevated above the forest floor to the susceptible

Camponotus sp. traveling on trails below. Chapter 2 of this thesis explains the surprising conclusions of a series of experiments that evaluated the transmission of O. camponoti-atricipis, 5 a particular species within the O. unilateralis complex. We found that spores are dispersed into the environment persistently over time and we suggest that susceptible ants are infected when they pass ascospores that have fallen to the ground below an ant cadaver. There is a caveat though, spreading spores and infecting a new host is likely not as simple as a spore merely reaching any ant of the genus Camponotus.

Researchers have long suspected and recently begun to present evidence that infection by

Ophiocordyceps sp. is highly host specific. In fact, some propose that each species of

Ophiocordyceps can only infect one species of ant. Initially these studies focused upon morphological characteristics. There are marked differences in ascospore size and development between the species documented (Evans et. al, 2011b). One study found that in laboratory conditions O. unilateralis sensu lato isolated from Camponotus castaneus can successfully infect

C. castaneus, C. americanus, and C. pennsylvanicus, but can only induce manipulation of C. castaneus and C. americanus, which are closely related (de Bekker et. al, 2014). This study did not evaluate the morphology or phylogenetics of the O. unilateralis species used for infection.

More recently, phylogenetic analysis of the O. unilateralis species complex has shown additional genetic differences between members infecting unique species of Camponotus (Araújo et. al,

2015). These studies are all suggestive that infection with Ophiocordyceps unilateralis species is host specific. Chapter 3 of this thesis details our work investigating the genetic diversity of O. unilateralis sensu lato infecting a single species of ant, Camponotus atriceps, from the Brazilian

Amazon. Finally, Chapter 4 of this thesis concludes by summarizing the information presented and highlighting potential questions for further exploration. 6 References

Darwin C. On the Origin of Species. 1859.

Poulin R. The evolution of parasite manipulation of host behaviour: A theoretical analysis.

Parasitology. 1994; 109:109-18. doi:10.1017/S0031182000085127

Dawkins R. The Extended Phenotype. Oxford: Oxford University Press; 1982.

Weber JN, Peterson BK, Hoekstra HE. Discrete genetic modules are responsible for complex

burrow evolution in Peromyscus mice. Nature. 17 Jan 2013; 493:402-405.

Cheadle MA, Spencer JA, Blagburn BL. Seroprevalences of Neospora caninum and Toxoplasma

gondii in nondomestic felids from southern Africa. J Zoo Wildlife Med. 1999; 30(2):248-

251.

Berdoy M, Webster JP, Macdonald DW. Fatal attraction in rats infected with Toxoplasma gondii.

Proc R Soc Lond. 7 April 2000; 267:1591-1594.

Poirette C, Kappeler PM, Ngoubangoye B, Bourgeois S, Moussodji M, Charpentier MJE.

Morbid attraction to leopard urine in Toxoplasma-infected chimpanzees. Current

Biology. 8 February 2016; 26(3):R98-R99. doi: 10.1016/j.cub.2015.12.020

Torrey EF, Yolken RH. Toxoplasma gondii and schizophrenia. Emerging Infectious Diseases.

November 2003; 9(11).

Eberhard WG. Spider manipulation by a wasp larva. Nature. 20 July 2000; 406:255.

http://www.nature.com/nature/journal/v406/n6793/pdf/406255a0.pdf

Andersen SB, Gerritsma S, Yusah KM, Mayntz D, Hywel-Jones NL, Billen J, Boomsma JJ,

Hughes DP. The life of a dead ant: The expression of an adaptive extended phenotype.

Am Nat. 2009; 174:424-433 doi:10.1086/603640 7 Evans HC, Elliot SL, Hughes DP. Ophiocordyceps unilateralis: A keystone species for

unraveling ecosystem functioning and biodiversity of fungi in tropical forests? Commun

Integr Biol. 2011a; 4(5):598-602. doi: 10.4161/cib.4.5.16721

Cranshaw W, Redak R. Bugs Rule!. Princeton: Princeton University Press; 2013. de Bekker, C, Quevillon LE, Smith PB, Fleming KR, Ghosh D, Patterson AD, Hughes DP.

Species-specific ant brain manipulation by a specialized fungal parasite. BMC

Evolutionary Biology. 2014; 14:166. doi: 10.1186/s12862-014-0166-3

Hughes DP, Andersen SB, Hywel-Jones NL, Himaman W, Billen J, Boomsma JJ. Behavioral

mechanisms and morphological symptoms of zombie ants dying from fungal infection.

BMC Ecology. 2011a; (11):13. doi: 10.1186/1472-6785-11-13

Hughes DP, Wappler T, Labandeira CC. Ancient death-grip leaf scars reveal ant-fungal

parasitism. Biology Letters. 2011b; 7(1):67-70. doi: 10.1098/rsbl.2010.0521

Evans HC, Samson RA. Cordyceps species and their anamorphs pathogenic on ants (Formicidae)

in tropical forest ecosystems: II: The Camponotus (Formicinae) complex. Trans Br

Mycol Soc. 1984; 82:127-150.

Evans HC, Elliot SL, Hughes DP. Hidden diversity behind the zombie-ant fungus

Ophiocordyceps unilateralis: Four new species described from carpenter ants in Minas

Gerais, Brazil. PLOS One. 2011b; 6(3):e17024. doi: 10.1371/journal.pone.0017024

Araújo JPM, Evans HC, Geiser DM, Mackay WP, Hughes DP. Unravelling the diversity behind

the Ophiocordyceps unilateralis (Ophiocordycipitaceae) complex: Three new species of

zombie-ant fungi from the Brazilian Amazon. Phytotaxa. 2015; 220:224-238

doi:10.11646/phytotaxa.220.3.2

Chapter 2

Transmission Potential

“Because pandemics almost always begin with the transmission of an animal microbe to a

human…”

– Nathan Wolfe

This chapter has been adapted from a manuscript submitted to BMC Ecology for publication with the title “Quantifying the transmission potential of an entomopathogenic fungus that controls the behavior of ants.”

Background

Some parasites have evolved the ability to alter the behavior of their hosts in ways that increase host transmission. This is the concept of the extended phenotype, a phenomenon in which altered host behavior is the result of natural selection acting on parasite genomes to modify animal behavior (Dawkins, 1982). Such control is assumed to be expensive and the majority of parasite life on Earth transmits effectively without such elaborate manipulative strategies (Poulin, 1994). However, a diverse group of pathogens exhibit extended phenotypes and it is generally assumed, though rarely demonstrated, that the control of host behavior increases transmission. 9 One category where transmission is clearly optimized by parasite control of host behavior is trophic transmission. Here parasites manipulate the behavior of one host, which increases the likelihood that the host is eaten by a secondary host in which the parasite completes development. This is an indirect lifecycle. For example, the parasite Diplostomum spathaceum, which infects fish as an intermediary host, alters the behavior of infected fish in such a way as to make them more susceptible to predation by the target host, fish-eating birds (Seppälä et. al,

2004). D. spathaceum and other parasites, such as Polymorphus minutus and Microphallus sp., respectively infecting crustaceans and freshwater snails, manipulate their host leading to increased predation by a specific target host (Seppälä et. al, 2004; Médoc et. al, 2006; Levri et. al, 1996). In one elegant study, intermediate hosts were 30 times more likely to be preyed upon than uninfected animals (Lafferty et. al, 1996). Still other parasites such as Curtuteria australis, a parasite of the New Zealand cockle (Austrovenus stutchburyi), induce manipulation that increases predation by non-target species in addition to target species (Mouritsen & Poulin,

2003). In all of these cases, the parasite-induced host manipulation increases the parasite’s transmission potential by facilitating trophic transmission.

However, not all manipulation results in the host becoming easy-to-catch prey for a predator that is also part of the parasite’s lifecycle. In some cases, manipulation functions to disperse propagules from the current host to a susceptible host of the same species. This is a direct lifecycle. Prominent examples of behavioral manipulation as part of a direct lifecycle are baculoviruses that infect insects. After infecting a lepidopteran host, baculoviruses proliferate and eventually produce a series of proteins that induce the caterpillar to climb to the top of a food plant prior to death (Kamita et. al, 2005; Goulson, 1997; Rebolledo et. al, 2015). This manipulation is assumed to function because it increases the infective potential of the virus by 10 spreading infectious virions over a greater area which, it is assumed, increases the likelihood that a healthy foraging host will encounter the infectious cadaver (Goulson, 1997; Rebolledo et. al,

2015). This seems a reasonable assumption as caterpillars of the same species (the target future host) eat the leaves the virions have been released onto.

Another example of well-documented host manipulation in a direct lifecycle system is the zombie-ant fungus, Ophiocordyceps unilateralis sensu lato. The O. unilateralis complex of species (see Evans et. al, 2011; Araújo et. al, 2015) infects ants of the tribe Camponotini and causes ants to bite onto leaves and twigs above the forest floor (Andersen et. al, 2009). Soon afterwards, the ant is killed and the fungus grows a stalk (stroma) from the dorsal pronotum

(Figure 2A) where, subsequently, the fungus develops a sexual spore-producing body (ascoma,

Figure 2A red arrow, Figure 2B) (Andersen et. al, 2012). The ascoma contains numerous spore sacks, called perithecia (Figure 2B red arrow, Figure 2C) which are individually capped by an ostiole (Figure 2B black arrow). Within each perithecia, clusters of asci (Figure 2D) form. Each individual ascus (Figure 2E) is made of 8 ascospores (Figure 2F). One of the most common species within the O. unilateralis complex is O. camponoti-atricipis, which is endemic to the

Brazilian Amazon (Araújo et. al, 2015). The presence of this extended phenotype would seem to indicate that manipulation to the phyllosphere increases the transmissive success of this group of pathogens. Previous work showed that positioning in the phyllosphere was necessary for the development of the stalk that precedes transmission (Andersen et. al, 2009). However, previous studies did not document the transmissive phase of the Ophiocordyceps unilateralis species’ life cycle. 11 A B

C D E F

In order to understand the parasite’s transmission, we first sought to understand the characteristics of the fungus that are relevant to infectivity and spore dispersion. Using samples collected freshly in the field, we documented morphological characteristics that could impact reproduction. In particular, we sought to determine whether there was directionality of stromatal growth and to quantify the number of ascospores produced in mature ascomata. We also conducted field studies to determine the timing of O. camponoti-atricipis ascospore release. 12 Combining these data, we estimate the period of time over which O. camponoti-atricipis is capable of infecting susceptible ants. The results support the previously suggested hypothesis that Ophiocordyceps unilateralis species exhibit iteroparous reproduction (Andersen et. al,

2012).

Methods

Sample Collection

Fieldwork was conducted in Reserva Florestal Adolpho Ducke (02°55’S, 59°59W) near

Manaus, Amazonas, Brazil in January 2014 and January 2015. O. camponoti-atricipis specimens were collected between 0900 and 1600 hrs. Prior to specimen collection, cardinal orientation of the substrate, ant, and primary face of the fruiting body were determined as well as an approximate height from the leaf litter. Additionally, using a small grid harnessed and suspended so that a series of vertical lines remained parallel to the gravity vector, photographs were taken of 19 specimens to determine the spatial orientation of the fungus. The angle of stromatal growth was measured using CellSens software. Samples were collected and maintained in humid, air- temperature tubes until preparation for further analysis. 13 Physical Characterization

Samples were photographed and then desiccated within 24 hours of collection. O. camponoti-atricipis stromata and ascomata were measured and photographed using an Olympus

SZX16 dissecting microscope and CellSens software (Figure 2A). O. camponoti-atricipis ostioles were counted manually using 2-3 images (number determined by shape of individual ascomata) taken 120° apart using an Olympus SZX16. Subsequently, individual O. camponoti- atricipis perithecia were dissected from each ascomata under a stereoscopic microscope. Intact isolated perithecia were mounted individually in 1.7 mM lacto-fuchsin and the asci and ascospores were spread into a single plane for brightfield microscopy using a Keyence BZ-9000.

Ascospores were counted from composite images at X40 magnification using BZII Analyzer’s detail-picked brightness-based hybrid cell count while asci were excluded from the hybrid cell count and counted manually.

Temporal Release Monitoring

Petri dishes containing unsupplemented agar were divided into 12 or 24 sections, creating a “collection clock”, and individual O. camponoti-atricipis samples were attached on the petri dish lid with the primary face of the ascoma oriented directly downwards. Samples for temporal monitoring of ascospore release were placed on "collection clocks” within 4 hours of collection and clocks were prepared between 1600 and 1800 hrs and spore collection began at 1800 hrs.

Each hour for the next 12 or 24 hours, the samples were moved to the next section of the petri dish. The spore collection was conducted in two rounds. The first round of five specimens was 14 collected for 12 hours and the second round of six specimens was collected for 24 hours. The 12- hour cycle contained five specimens and was conducted in an area with distant low-intensity artificial light throughout the night. The 24-hour cycle contained six specimens and was conducted in an area without any artificial light. After a complete cycle, at either 0600 hrs or

1800 hrs, the petri dishes were examined under a microscope and the ascospores present in each section were counted. Samples were then allowed to dry for preservation and photographed using a Canon 7D equipped with a MP-E 65mm Macro Lens, fitted with a MT-24EX Macro Elite

Flash.

Results

Fecundity and morphological characteristics of spore producing body

As with all species in the complex Ophiocordyceps unilateralis s.l. the species O. camponoti-atricipis grows a stalk (stroma) from the cadaver of the host ant (Camponotus atriceps) that the fungus previously manipulated to bite into a leaf before it killed the ant. The observation that the ant cadaver’s mandibles are embedded into plant tissue is evidence of the manipulation that occurred pre-mortem. These ants (i.e. Camponotus or carpenter ants) do not, as part of their normal behavior, bite into leaf tissue. The stalk that grows from the ant cadaver develops an ascoma, or sexual spore-producing body, which forms on one side (Figure 2A). The position of this ascoma is relevant to transmission. We determined that the mean stromatal length from the host body to the ascoma was 3.97 mm (n=40, s.d. 2.52, range 1.03-11.5, Figure 3A) with a mean width of 0.22 mm (n=42, s.d. 0.09, range 0.05-0.53). The mean length of the stroma 15 (from base to tip) was 15.20 mm (n=35, s.d. 5.55, range 1.81-26.49). The mean ascoma was 0.84 mm wide (n=36, s.d. 0.27, range 0.36-1.68,) by 1.18 mm long (n=36, s.d. 0.47, range 0.12-2.84).

An ascoma has a mean of 185 ostioles (n=10, s.d. 109, range 66-413, Figure 3C). The ostioles are the openings from which spores are released. The stroma initially grows perpendicular to the ant body, but then curves so that it grows towards the ground. The mean angle from vertical was

25.8° (n=19, s.d. 21.1, range 1.5-76.6°, Figure 3B) but the data are unevenly distributed and are clustered around 0°. The angle the ant faces in space is random (Figure 3B). To measure potential reproductive output we removed 34 intact perithecia from 5 ascomata that grew from 3 cadavers. We counted the number of ascospores in each perithecium. The mean number of ascospores in an O. camponoti-atricipis perithecium was 2,212 (s.d. 761, range 719-4,282,

Figure 3D). 16 A B

C D

17 Timing of spore release

In addition to fecundity, we measured the timing of spore release. We collected 11 cadavers with mature ascomata to gather data on the timing of spore release. Of these 11 specimens, a total of 8 released spores. Although all eight released spores, for three of the ascomata the quantity of the spores could not be counted, as the spores clustered densely on the agar they were shot onto. For these three specimens we conservatively estimated spore number

(Figure 4 dotted lines, red area). The five quantifiable samples were compared (Figure 4 solid lines) and then the time when we observed maximum spore release was recorded to allow comparison with the remaining three samples.

0600 hrs. Two specimens released very low numbers of spores throughout the day but still peaked in the range of 0400-0600 hrs. The mean spore release rapidly increased between 0200-

0500 hrs, beginning at 43 spores between 0200 and 0300 and then increasing quickly to 969 during the 0300-0400 hrs measurement window. The increase continued for the next two hours reaching a mean of 1,745 spores (s.d. 1,808, range 41-5,061) and 2,115 spores (s.d. 1,792, range

322-5,243) at 0400 hrs and 0500 hrs respectively. At 0600 hrs the mean release was reduced to

1,478 spores (s.d. 1503, range 503-4,466) and release effectively ceased at 0700 hrs when an average of 92 spores (s.d. 84, range 0-332) were released (Figure 4). When spore release was at its maximum rate we counted 5,243 spores in one hour, although there were two specimens that released such large numbers of spores that it was impossible to distinguish the individual ascospores, so the peaks were conservatively estimated (Figure 4 red area). Our data also show that each day the fungus releases a mean of 5,296 ascospores (n=7, s.d. 5058, range 1,138-

16,603).

Discussion

In this study, our goal was to quantify the potential transmissive capabilities of a specialized fungal parasite of ants, O. camponoti-atricipis. We found that each perithecium contained about 2,200 ascospores (Figure 3D). Spores are shot out through ostioles and we counted a mean of 185 ostioles per ascomata (Figure 3C). Every ostiole caps a perithecium, 19 which indicates that the mean O. camponoti-atricipis specimen produces about 400,000 ascospores (number of ascospores per perithecium x number of ostioles). This amount is higher than expected. Studies of Strepsiptera, another parasite of insects, indicate that species infecting nest-building insects that are predictable in time and space produce fewer propagules than species that disperse propagules into the environment that must then infect widely dispersed hosts that are not predictable (Maeta et. al, 1998). Previous work suggested that Ophiocordyceps camponoti-rufipedis (a member of the O. unilateralis species complex) infecting Camponotus rufipes in the Atlantic Rainforest of Brazil manipulated its host to die near trails (Loreto et. al,

2014). This implied that spore production was targeted on the ant trail, implying that lower numbers of spores would be produced. Although we did not measure the trail activity of the target host in this species, Camponotus atriceps, it is similar to Camponotus rufipes.

Additionally, we know the ascospore size (80-85 x 3 µm) and that the ascospores fall down to the ground quickly after being released (Araújo et. al, 2015). Taken together, the predictable nature of the host in time and space implied that O. camponoti-atricipis would produce a smaller cohort of infectious spores, which was not the case. However, since this is the first study to quantify the fecundity of a specialized fungal parasite of an insect we do not have a benchmark against which to compare our findings.

Our measurements show that there is a stereotyped pattern to the development of O. camponoti-atricipis. While morphological development likely impacts how the parasite transmits between hosts we were unable to show a definitive link between these significant morphological traits and reproductive potential. A principal morphological feature is the distance of the spore producing body from the cadaver of the ant. Total stromatal length varied from 1.81 mm to 26.49 mm but the distance from the ant to where the fruiting body formed was more consistent and 20 clustered around the mean of 3.97 mm (s.d. 2.52, Figure 3A). There was no association between the total length and the length to ascoma (R2 = 0.035), which suggests that they are controlled by distinct, independent factors. Fungal growth, particularly the formation of a fruiting body, is controlled by numerous factors. One of the most important factors is nutrient availability. In

Saccharomyces macrospora fruiting body formation requires biotin while for both S. macrospora and Aspergillus nidulans an arginine source is essential for the growth of a fruiting body (Pöggler et. al, 2006). Light, temperature, pH, and host or self-production of chemical compounds have all been linked to fruiting body formation in various fungal species and any of these could be the driving factors of O. camponoti-atricipis maturation (Pöggler et. al, 2006).

Some fungi have also been shown to exhibit gravitropism, which is the development of particular fungal structures as a result of gravity (Corrochano & Galland, 2006). To explore the potential for gravitropism in O. camponoti-atricipis, we measured the angle of stromatal growth with respect to the gravity vector. Our data show that the fungal stalk exhibits growth that is parallel to the gravity vector (Figure 3B). To assess the possibility that this directionality of growth was a result of ant orientation rather than fungal gravitropism, we also measured the direction the host ants faced in the same plane as the fungal stroma. We found that there was no association between the direction of the host ant and the spatial orientation of the fungal growth

(T-Test p=0.34). The direction the ant faced in a vertical plane was random, while mean stromatal growth was almost directly downwards. This strongly suggests that the growth of O. camponoti-atricipis is directed by gravitational force.

The adaptation of gravitropic growth by O. camponoti-atricipis is likely a result of the gap between the location of susceptible hosts and the environment in which the fungus can mature. O. camponoti-rufipedis, a species closely related to O. camponoti-atricipis, cannot 21 complete its life cycle within the ant colony and other Ophiocordyceps species fail to mature elsewhere in the forest (Andersen et. al, 2009; Loreto et. al, 2015). This suggests that the extended phenotype of O. unilateralis s.l. whereby manipulation leads the ants to die attached to leaves is essential to parasite development and growth. While manipulation of the ant host to plant surfaces enables fungal growth and maturation, the manipulation also distances the fungus from uninfected hosts. Gravitropism is likely an adaptation that, combined with the highly directional spore release of O. unilateralis s.l., enables the fungus to target ascospore release towards the forest floor (Evans & Samson, 1984). Our data indicate that the stromatal growth angle is a controlled parameter of Ophiocordyceps development that likely affects the dispersion of infectious spores.

Additionally, our fieldwork determined that each morning the fungus releases a mean of

5,058 ascospores. If we assume this release is constant, every O. camponoti-atricipis specimen

(producing 400,000 total ascospores each) could release ascospores daily for over two months. In line with this, previous work has documented O. unilateralis s.l. specimens persisting in the environment for months (Andersen et. al, 2009; Pontoppidan et. al, 2009; Hughes et. al, 2011).

However, as shown in previous studies, this fungus is susceptible to hyperparasitism resulting in castration (Andersen et. al, 2009; Andersen et. al, 2012). These data suggest that to overcome the gap between the environment amenable to fungal maturation (elevated on plant growth) and the location frequented by susceptible hosts (the forest floor), O. camponoti-atricipis has adapted a bet hedging strategy that creates a persistent minefield of short-lived spores on the forest floor.

This adaptation allows the fungus to continue its life cycle by infecting new hosts despite its need for particular growth conditions found in the lower phyllosphere. 22 We found that during a single day ascospore release is temporally focused and that ascospore release exclusively occurs in significant amount between 0200 hrs and 0700 hrs with a clear peak at 0500 hrs (Figure 4). Light has been documented as a potent inducer of fungal spore release (Corrochano & Galland, 2006). On the days of sample collection sunrise occurred around

0602 hrs and our studies were conducted in two disparate areas, one with artificial lighting the other with only natural lighting. Despite the difference in lighting between our two experimental sites, there was no difference in ascospore release temporality, indicating that for O. camponoti- atricipis light does not play a role in ascospore release. Relative humidity has also been linked with spore release in fungi (Gottwald et. al, 1997). In the Amazon, temperature is at a minimum and relative humidity at a maximum in the hour preceding sunrise. We propose that these factors are the driving force behind O. camponoti-atricipis spore release. We did not measure the circadian rhythms in ant foraging but this is clearly data that futures studies should seek to gather.

Conclusions

In this study we document the fecundity, taxonomic and functionally-relevant morphologies, and temporality of O. camponoti-atricipis with regards to transmission potential.

We show that each fungal perithecium contains a mean of 2,200 ascospores and that every fungal specimen potentially harbors a mean of 400,000 ascospores that could be released over two months or more. We also reveal differential growth characteristics of O. camponoti-atricipis that could play a role in transmission. In particular, we document apparent gravitropic growth of the 23 fungal stalk which directs spore release towards the ground, the ultimate destination of the spores which are designed to attach to foraging ants. Finally, we discovered that O. camponoti-atricipis temporally focuses spore release, which we attribute to possible environmental factors such as relative humidity. Future work should examine the forest floor for the presence of ascospore minefields and should experimentally manipulate potential growth factors to further elucidate the components affecting Ophiocordyceps development. In addition, it is important to quantify how changes in ant foraging behavior affect transmission.

References

Dawkins R. The Extended Phenotype. Oxford: Oxford University Press; 1982.

Poulin R. The evolution of parasite manipulation of host behaviour: A theoretical analysis.

Parasitology. 1994; 109:109-18. doi:10.1017/S0031182000085127

Seppälä O, Karvonen A, Valtonen T. Parasite-induced change in host behaviour and

susceptibility to predation in an eye fluke-fish interaction. Anim Behav. 2004; 68:257-

263. doi:10.1016/j.anbehav.2003.10.021

Médoc V, Bollache L, Beisel J. Host manipulation of a freshwater crustacean (Gammarus

roeseli) by an acanthocephalan parasite (Polymorphus minutus) in a biological invasion

context. Int J Parasitol. 2006; 36:1351-1358. doi:10.1016/j.ijpara.2006.07.001

Levri EP, Lively CM. The effects of size, reproductive condition, and parasitism on foraging

behaviour in a freshwater snail, Potamopyrgus antipodarum. Anim Behav. 1996; 51:891-

901. doi:10.1006/anbe.1996.0093 24 Lafferty KD, Morris AK. Altered behavior of parasitized killifish increases susceptibility to

predation by bird final hosts. Ecology. 1996; 77:1390-1397 doi:10.2307/2265536

Mouritsen KN, Poulin R. Parasite-induced trophic facilitation exploited by a non-host predator:

A manipulator’s nightmare. Int J Parasitol. 2003; 33:1043-1050 doi:10.1016/S0020-

7519(03)00178-4

Kamita SG, Nagasaka K, Chua JW, Shimada T, Mita K, Kobayashi M, Maeda S, Hammock BD.

A baculovirus-encoded protein tryosine phosphatase gene induces enhanced locomotory

activity in a lepidopteran host. Proc Natl Acad Sci USA. 2005; 102:2584-2589.

doi:10.1073/pnas.0409457102

Goulson D Wipfelkrankheit: modification of host behaviour during baculoviral infection.

Oecologia. 1997; 109:219-228. doi:10.1007/s004420050076

Rebolledo D, Lasa R, Guevara R, Murillo R, Williams T. Baculovirus-induced climbing

behavior favors intraspecific necrophagy and efficient disease transmission in Spodoptera

exigua. PLoS ONE. 2015; 10:e0136742 doi:10.1371/journal.pone.0136742

Evans HC, Elliot SL, Hughes DP. Hidden diversity behind the zombie-ant fungus

Ophiocordyceps unilateralis: Four new species described from carpenter ants in Minas

Gerais, Brazil. PLoS One. 2011; 6:e17024 doi:10.1371/journal.pone.0017024

Araújo JPM, Evans HC, Geiser DM, Mackay WP, Hughes DP. Unravelling the diversity behind

the Ophiocordyceps unilateralis (Ophiocordycipitaceae) complex: Three new species of

zombie-ant fungi from the Brazilian Amazon. Phytotaxa. 2015; 220:224-238

doi:10.11646/phytotaxa.220.3.2 25 Andersen SB, Gerritsma S, Yusah KM, Mayntz D, Hywel-Jones NL, Billen J, Boomsma JJ,

Hughes DP. The life of a dead ant: The expression of an adaptive extended phenotype.

Am Nat. 2009; 174:424-433 doi:10.1086/603640

Andersen SB, Ferrari M, Evans HC, Elliot SL, Boomsma JJ, Hughes DP. Disease dynamics in a

specialized parasite of ant societies. PLoS ONE. 2012; 7:e36352

doi:10.1371/journal.pone.0036352

Maeta Y, Takahashi K, Shimada N. Host body size as a factor determining the egg complement

of Strepsiptera, an insect parasite. Int J Insect Morphol. 1998; 27:27-37

doi:10.1016/S0020-7322(97)00033-0

Loreto RG, Elliot SL, Freitas MLR, Pereira TM, Hughes DP. Long-term disease dynamics for a

specialized parasite of ant societies: A field study. PLoS ONE. 2014; 9:e103516

doi:10.1371/journal.pone.0103516

Pöggler S, Nowrousian M, Kϋck U. Fruiting-body development in ascomycetes. In Kϋes U,

Fischer R, editors. The Mycota: Growth, differentiation and sexuality. Berlin: Springer;

2006. p. 325-355.

Corrochano LM, Galland P. Photomorphogenesis and gravitropism in fungi. In Kϋes U, Fischer

R, editors. The Mycota: Growth, differentiation and sexuality. Berlin: Springer; 2006. p.

233-259.

Evans HC, Samson RA. Cordyceps species and their anamorphs pathogenic on ants (Formicidae)

in tropical forest ecosystems II. The Camponotus (Formicinae) complex. Trans Br Mycol

Soc. 1984; 82:127-150 26 Pontoppidan M-B, Himaman W, Hywel-Jones NL, Boomsma JJ, Hughes DP. Graveyards on the

move: The spatio-temporal distribution of dead Ophiocordyceps-infected ants. PLoS

ONE. 2009; 4:e4835 doi:10.1371/journal.pone.0004835

Hughes DP, Andersen SB, Hywel-Jones NL, Himaman W, Billen J, Boomsma JJ. Behavioral

mechanisms and morphological symptoms of zombie ants dying from fungal infection.

BMC Ecology. 2011; 11 doi:10.1186/1472-6785-11-13

Gottwald TR, Trocine TM, Timmer LW. A computer-controlled environmental chamber for the

study of aerial fungal spore release. Phytopathology. 1997; 87:1078-1084.

doi:10.1094/PHYTO.1997.87.10.1078

27 Chapter 3

Host Specificity and Fungal Diversity

“When you have seen one ant, one bird, one tree, you have not seen them all.” – E.O. Wilson

Background

One of the most remarkable characteristics of many pathogens is the specificity with which the disease targets particular host species. Pathogen specificity determines who can be afflicted with any particular disease and this selectivity can control the spread of pathogens throughout diverse populations. Pathogen specificity is a key factor affecting the ability of diseases to spillover from one population to another. Understanding the underlying mechanisms of host specificity will enable more accurate predictions of diseases of global concern, perhaps even allowing prescient researchers to predict the “next HIV”.

To understand the importance of host specificity, consider the Poxviridae, a viral family encompassing members that exhibit both broad and narrow species specificity (McFadden,

2005). Perhaps the best example of our understanding of specificity enabling control of a deadly disease is variola virus, the causative agent of smallpox. Variola virus only infects humans and as such has no animal reservoir. This specificity allowed for the first coordinated eradication of a disease, which in 1977 eliminated variola virus from circulation in the wild saving untold millions from death and disfigurement (Fenner et. al, 1988). Shortly afterwards all laboratory stocks, save two in secure facilities in the United States and Russia, were destroyed (Breman & 28 Arita, 1980). Despite this successful eradication campaign, little is known about what makes variola human-specific (McFadden, 2005). Other Poxviridae such as monkeypox, cowpox, and vaccinia virus are not so highly selective of hosts; monkeypox in particular regularly emerges from a yet unidentified reservoir to human and other primate populations in Central Africa

(Lewis-Jones, 2004). Understanding the drivers of pathogen specificity is vital to enable efforts to predict future spillovers and epidemics.

While viruses are likely the best example of species specificity, fungi also exhibit high levels of host selectivity. The rust fungi (order Pucciniales) are also extremely species specific, with many species only infecting a single species, or even a single cultivar of plant (Parker et. al,

1994; Health, 1981). In one study, twenty-nine related plant species were inoculated with the rust fungus Puccinia eupatorii and monitored for signs of infection. Only one, Campuloclinium macrocephalum, exhibited signs of infection, indicating that P. eupatorii is very host specific

(Retief et. al, 2016). Numerous other species of rust fungi have exhibited similar abilities to infect only a single species of plant. These fungi are being explored as potential biocontrol agents for invasive plant species (Retief et. al, 2016; Ellison et. al, 2008). Similar to the

Poxviridae, while the host-specificity of rust fungi is well documented, the exact mechanism for this trait is poorly understood. Evidence from Arabidopsis thaliana suggests that there is a role for defense signaling through the jasmonic acid pathway and salicylic acid signaling, two signals that are linked to the expression of plant defense genes (Mellersh et. al, 2003; Pozo et. al 2005).

Using a comparative model of Hemileia vastatrix infection in susceptible and resistant cultivars of the natural host Coffee arabica and Coffee congensis, another study found that numerous other plant immune pathways, including cell death, lignification, and production of phenol-like 29 compounds, are involved in resistance to fungal infections (Silva et. al, 2002). Overall, the determinant factors of host specificity for rust fungi appear to be highly diverse.

Members of the species complex Ophiocordyceps unilateralis are also thought to be highly host specific (Evans et al. 2011; Araújo et al. 2015). Typically, this is described as one species of fungus infecting one species of ant (Hirata, 2014); however, laboratory infections have shown that some O. unilateralis sensu lato isolates are capable of infecting at least two related ant species (de Bekker et. al, 2014) but it is not known if this happens in nature. Additional phylogenetic studies of specimens collected from the Brazilian Amazon have shown that the species of fungi infecting 12 different hosts have 12 distinct genetic profiles (Araújo et. al,

2015). This study, and more recent, yet to be published, work by Araújo et. al firmly supports the idea, based on worldwide collections, that each species of O. unilateralis infects a single species of host. Thus far, the genetic analyses of O. unilateralis s.l. have focused on identification of novel species from new hosts. Speciation has not been examined over a large sample of ants of a single species to determine if multiple strains or species of O. unilateralis s.l. can infect a single species of ants. To answer that question, we selected a previously identified species in the O. unilateralis complex, O. camponoti-atricipis, and sought to sequence four genes from fungi extracted from 58 specimens. O. camponoti-atricipis only infects the ant Camponotus atriceps in the Brazilian Amazon (Araújo et. al, 2015) and is one of the most common O. unilateralis s.l. species in Reserva Florestal Adolpho Ducke, where we conducted fieldwork. Using specimens of this single host species, we aim to determine the genetic variation across individuals, using four genes (LSU, TEF, SSU, and RPB1) that are commonly used to determine the relatedness and speciation of the kingdom Fungi. Although we have not received sequence results yet, based upon the literature, we expect to find that the samples will present highly homogenous sequences 30 indicating that O. unilateralis sensu lato is species specific and that each species of ant is most likely only attacked by a single species from the O. unilateralis species complex.

Methods

Sample Collection

During January 2014, fieldwork was conducted in Reserva Florestal Adolpho Ducke

(02°55’S, 59°59W) 26 km from Manaus, Amazonas, Brazil. Over the course of several weeks,

70-80 O. camponoti-atricipis specimens were located and photographed individually. Ant cadavers and the leaves, twigs, or palm spines to which they are attached were collected and desiccated upon return to the field laboratory. Samples were maintained in a desiccated state until January 2016.

DNA Extraction, Cleaning, PCR, Sequencing, and Phylogenetic Analysis

To extract DNA, the stroma and ascoma of each specimen was ground by hand in 50 µl

CTAB Extraction Solution (GBiosciences). An additional 400 µL CTAB Extraction Solution was then added and samples were incubated at 60°C for 20 minutes. Samples were centrifuged for 10 minutes at 14,000 RPM and the supernatant was transferred to a new tube, to which 500

µL chloroform:isoamyl alcohol (24:1) was added. Samples were then centrifuged for 20 minutes at 14,000 RPM, the supernatant transferred to a new tube and then cleaned using GeneClean III

Kit (MP Bio). PCR was done using reaction mixtures with 4.5 µl Buffer E (ThermoFisher), 0.5

µl forward and reverse primers, 1.0 µl template, 18.4 µl MiliQ water, and 0.1 µl Platinum Taq 31 Polymerase (Invitrogen). PCR products were run on 1.5% agarose gels prior to Sanger

Sequencing at Penn State’s Genomics Core Facility. Phylogenetic analysis will be performed using the procedures described by Kepler et, al (2011), with potential modifications to fit the data set acquired.

Discussion

This study has not yet produced any results. We have successfully amplified 44 samples of LSU and 23 samples of TEF and are continuing to amplify DNA from 58 samples of O. camponoti-atricipis. Since we are using desiccated samples stored at room temperature for two years, successful amplification of the target genes is challenging, thus progress is slow.

Therefore, although incomplete at the time of this writing, we do expect to have results in the near future. The remainder of this chapter will briefly discuss the potential outcomes and the significance of several possible outcomes.

The preponderance of literature posits that Ophiocordyceps species infecting ants each only infect a single host and suggests that most ants are infected by only one species of

Ophiocordyceps (Evans et. al, 2011; Kobmoo et. al, 2012; Araújo et. al, 2015). We may find that there is minimal genetic variation across our sample of 58 O. camponoti-atricipis specimens infecting only Camponotus atriceps. This, if confirmed, would indicate that O. camponoti- atricipis is the only species capable of infecting C. atriceps. One possible explanation for this involves the ant immune system. Insects have both cellular and humoral immune systems and both are involved in defense against parasitic fungi (Fargues & Remaudiere, 1977; Vilcinskas &

Götz, 1999). Insects are highly adept at recognizing fungal cell walls through the β-1,3-glucan 32 receptor, which triggers innate immune pathways (Ma & Kanost, 2000). Interactions between O. unilateralis s.l. cell wall proteins and the particular β-1,3-glucan receptor of each host could contribute to host specificity. Further molecular studies of the interaction of O. unilateralis s.l. with its target host and with non-target hosts will help elucidate the exact mechanism governing specificity of the fungal complex.

Another possibility is that we will find small sequence variations that are not indicative of cryptic species but do correlate with geographic location. The specimens analyzed were collected from four different locations across the Brazilian states of Amazonas and Roraima, including several that were collected from the archipelago of Anavilhanas in the Rio Negro. This segregation and isolation may result in small polymorphisms between O. camponoti-atricipis from each distinct collection site. It is interesting to note that previous research has shown that O. unilateralis s.l. infecting C. leonardi in Thailand creates graveyards of manipulated ants

(Pontoppidan et. al, 2009). These graveyards are composed of several cryptic O. unilateralis s.l. species that each infect a single ant species (Kobmoo et. al, 2012). Thus far, though, no study has looked for genetic variation between graveyards nor have any used genetics to investigate whether graveyards are colony specific. One impressive study monitored infected cadaver location for 20 months and noted that cadavers are concentrated around individual colonies, an observation that tentatively suggests that these graveyards may be colony specific (Loreto et. al,

2014). If our sequences of O. camponoti-atricipis show small geographic variation, then further studies looking at the inter-graveyard variation should be conducted.

A third possibility is that our sequences of O. unilateralis s.l. isolated from C. atriceps exhibit high levels of variation suggestive of multiple cryptic species classified as O. camponoti- atricipis. Two possibilities exist within this category. First, the variant species may be species 33 previously thought to infect only other species of Camponotus, such as O. camponoti-bispinosi,

O. camponoti-indiani (Araújo et. al, 2015), O. camponoti-rufipedis, O. camponoti-balzani, O. camponoti-melanotici or O. camponoti-novogranadensis. (Evans et. al, 2011a). Second, the variant species may be uncharacterized examples of increasing diversity in the O. unilateralis species complex (Evans et. al, 2011b). Either one of these would suggest that the division of the species complex into its component cryptic species must proceed with larger sample sizes to account for this single host diversity. It also would appear to refute the hypothesis that each species of O. unilateralis only infects one species and that each species of ant is only infected by one species within the O. unilateralis complex.

We cannot make any conclusions about the results of this experiment, however, we can conclude that whatever the outcome, it is important that studies evaluating the genetic diversity, or lack thereof, of Ophiocordyceps species must look at the diversity in single-host populations as well as across multiple hosts.

References

McFadden G. Poxvirus tropism. Nature Reviews Microbiology. 2005; 3:201-213. doi:

10.1038/nrmicro1099

Fenner F, Henderson DA, Arita I, Ježek Z, Ladnyi ID. Smallpox and its eradication. Geneva:

World Health Organization. 1988.

Breman JG, Arita A. The confirmation and maintenance of smallpox eradication. New Engl J

Med. 1980; 303(22):1263-1273. doi: 10.1056/NEJM198011273032204

Lewis-Jones, S. Zoonotic poxvirus infections in humans. Curr Opin Inf Dis. 2004; 17(2):81-89. 34 Parker A, Holden ANG, Tomley AJ. Host specificity testing and assessment of the pathogenicity

of the rust, Puccinia abrupta var. partheniicola, as a biological control agent of

Parthenium weed (Parthenium hysterophorus). Plant Pathology. 1994; 3:1-16.

Health MC. A generalized concept of host-parasite specificity. Phytopathology. 1981;

71(11):1121-1123.

Retief E, van Rooi C, Breeyen AD. Environmental requirements and host-specificity of Puccinia

eupatorii, a potential biocontrol agent of Campuloclinium macrocephalum in South

Africa. Australasian Plant Pathology. 2016; doi: 10.1007/s13313-016-0401-z

Ellison CA, Evans HC, Djeddour DH, Thomas SE. Biology and host range of the rust fungus

Puccinia spegazzinii: A new classical biological control agent for the invasive, alien

weed Mikania micrantha in Asia. Biological Contol. 2008; 45(1):133-145. doi:

10.1016/j.biocontrol.2007.12.001

Mellersh DG, Heath MC. An investigation into the involvement of defense signaling pathways in

components of the nonhost resistance of Arabidopsis thaliana to rust fungi also reveals a

model system for studying rust fungal compatibility. Phytopathology. 2003; 16(5):398-

404. doi: 10.1094/MPMI.2003.16.5.398

Pozo MJ, Van Loon LC, Pieterse CMJ. Jasmonates – Signals in plant-microbe interactions. J

Plant Growth Regulation. 2005; 23(3):211-222. doi: 10.1007/s00344-004-0031-5

Silva MC, Nicole M, Guerra-Guimaràes L, Rodrigues Jr. CJ. Hypersensitive cell death and post-

haustorial defence responses arrest the orange rust (Hemileia vastatrix) growth in

resistant coffee leaves. Physiological and Molecular Plant Pathology. 2002; 60:169-183.

doi: 10.1006/pmmp.2002.0389 35 Evans HC, Elliot SL, Hughes DP. Hidden diversity behind the zombie-ant fungus

Ophiocordyceps unilateralis: Four new species described from Carpenter ants in Minas

Gerais, Brazil. PLOS One. 2011; 6(3):e17024. doi: 10.1371/journal.pone.0017024

Araújo JPM, Evans HC, Geiser DM, Mackay WP, Hughes DP. Unravelling the diversity behind

the Ophiocordyceps unilateralis (Ophiocordycipitaceae) complex: Three new species of

zombie-ant fungi from the Brazilian Amazon. Phytotaxa. 2015; 220:224-238

doi:10.11646/phytotaxa.220.3.2

Hirata K. The zombie ant: Parasitic fungi and behavior manipulation. 2014. Web. www.reed.edu/

biology/courses/BIO342/2014_syllabus/2014_WEBSITES/khsite/phylogeny.html de Bekker, C, Quevillon LE, Smith PB, Fleming KR, Ghosh D, Patterson AD, Hughes DP.

Species-specific ant brain manipulation by a specialized fungal parasite. BMC

Evolutionary Biology. 2014; 14:166. doi: 10.1186/s12862-014-0166-3

Kepler RM, Kaitsu Y, Tanaka E, Shimano S, Spatafora JW. Ophiocordyceps pulvinata sp. nov.,

a pathogen of ants with a reduced stroma. Mycoscience. 2011; 52: 39–47.

Kobmoo N, Mongkolsamrit S, Tasanathai K, Thanakitpipattana D, Luangsa-Ard JJ. Molecular

phylogenies reveal host-specific divergence of Ophiocordyceps unilateralis sensu lato

following its host ants. Molecular Ecology. 2012; 21(12):3022-3031. doi:

10.1111/j.1365-294X.2012.05574.x

Fargues J, Remaudiere G. Considerations on the specificity of entomopathogenic fungi.

Mycopathologia. 1977; 62(1):31-37.

Vilcinskas A, Götz P. Parasitic fungi and their itneractions with the insect immune system.

Advances in Parasitology. 1999; 43:297-313. doi: 10.1016/S0065-308X(08)60244-4 36 Ma C, Kanost MR. A beta1,3-glucan recognition protein from an insect, Manduca sexta,

agglutinates microorganisms and activates the phenoloxidase cascade. J Biol Chem.

2000; 275(11):7505-7514.

Pontippidan M, Himaman W, Hywel-Jones NL, Boomsma JJ, Hughes DP. Graveyards on the

move: The spatio-temporal distribution of dead Ophiocordyceps-infected ants. PLOS

One. 2009; 4(3):e4835. doi: 10.1371/journal.pone.0004835

Loreto RG, Elliot SL, Freitas MLR, Pereira TM, Hughes DP. Long-term disease dynamics for a

specialized parasite of ant societies: A field study. PLOS One. 2014; 9(8):e103516. doi:

10.1371/journal.pone.0103516.

Evans HC, Elliot SL, Hughes DP. Ophiocordyceps unilateralis: A keystone species for

unraveling ecosystem functioning and biodiversity of fungi in tropical forests? Commun

Integr Biol. 2011; 4(5):598-602. doi: 10.4161/cib.4.5.16721

37 Chapter 4

Conclusions

“I am turned into a sort of machine for observing facts and grinding out conclusions.”

– Charles Darwin

Ophiocordyceps, a genus of parasitic fungi that infects ants, is most famous for its ability to induce the zombie-ant phenomenon. This behavioral manipulation is an extended phenotype of the fungus. In Chapter 2, we investigated the transmission potential of a model species,

Ophiocordyceps camponoti-atricipis in its natural habitat, the Brazilian Amazon. Measurements of fungal orientation showed that the growth of O. camponoti-atricipis occurs parallel to gravity, indicating that development is gravitropic. Additionally, observations showed a clear concentration of infectious ascospore release around the early morning hours that peaked between 0500 and 0600 hours and released only about 1 percent of the total ascospore production. These findings suggest that humidity, which peaks in the early morning hours, is the driving factor of O. camponoti-atricipis ascospore release and also suggest that ascospores are released daily over two months. This repetitive release creates a persistent patch of infectious ascospores on the forest floor that are encountered by ants that pass through the area. Without the elevated positions of cadavers that result from behavioral manipulation of the host ant by the parasitic fungus, this minefield could not form, indicating that the O. camponoti-atricipis is highly adapted to its niche. In Chapter 3, we discuss our efforts to elucidate the potential diversity of O. unilateralis infecting Camponotus atriceps to shed light on the host specificity of 38 the parasitic fungal genus. Although this experiment has not yielded results, no matter what the findings the conclusions will be guide future studies of Ophiocordyceps.

Both of these projects produced a myriad of questions for future studies. In Chapter 2, we proposed a novel model of transmission for Ophiocordyceps. An important question stemming from this model is whether the proposed persistent ascospore patches, or minefields, can be shown to exist in the forest. This study will require high magnification field microscopy of the substrate below cadavers, a technique for which the resources were not in place in January 2014,

2015, or 2016. Additional studies of spore release timing should be done to increase the sample size; moreover, some of these experiments should be done under conditions of controlled humidity and temperature to test the hypothesis that relative humidity is the driving factor for

Ophiocordyceps ascospore release.

In Chapter 2, we also looked at the growth patterns of O. camponoti-atricipis. Ideally, to support our field observations we would conduct growth experiments under laboratory conditions. Unfortunately, the culture techniques to successfully grow mature O. unilateralis stromata and ascomata from an ant cadaver have not been developed. Once those techniques are discovered, infected cadavers should be placed with different orientations as well as with different light sources and wind directions to evaluate whether gravitropism or another tropism is the primary driving factor of O. camponoti-atricipis growth direction. The work of Chapter 2 also developed the techniques needed to evaluate the transmission potential of more species of

Ophiocordyceps.

In Chapter 3, we sought to evaluate the potential for cryptic speciation of the fungus currently classified as O. camponoti-atricipis. A persistent question throughout this project was what could be potential drivers of the species specificity of the many Ophiocordyceps 39 unilateralis species. In-depth molecular studies of natural and non-natural host infections will be required to answer that question. Chapter 3 also raises the question of how broad or specific future studies of the biodiversity of parasitic fungi should be. For decades, these pathogens were classified merely on the macroscopic characteristics. More recently, studies have used microscopic evaluation of fungal morphology to assess the speciation of Ophiocordyceps. Now, genetic tools may allow even these classifications to be parsed into a series of cryptic species.

The questions answered and raised by the experiment discussed in Chapter 3 are necessary to answer to introduce guidance and clarity to the documentation of novel parasitic fungi.

Overall, the data presented and interpreted in this thesis will provide the foundation for many future studies of Ophiocordyceps. Additionally, in combination with work from other sources, they will increase the understanding of the evolution, transmission, and host-specificity of pathogens of human concern. Perhaps derivations of this work will even contribute to the identification, characterization, treatment, and eradication of numerous diseases that threaten humanity. 40

Appendix A

Attachment to Leaves

Ophiocordyceps is a genus with over 150 species among which there are several different attachment strategies and preferred attachment substrates (Sung et. al, 2007). Ophiocordyceps unilateralis (Figure 1), the species complex studied in Chapter 2 and 3, manipulates ants to bite their mandibles firmly into twigs (or similar ligneous growths such as spines) and leaves

(Andersen et. al, 2009). Another species, Ophiocordyceps lloydii var. binata induces a behavioral change that brings infected ants to the abaxial side (i.e. underside) of leaves, but does not induce biting behavior (Evans et. al, 1984). Instead, these ants are attached to leaves by extensive growth of fungal mycelium (Figure 5 and 6). To document this, samples O. lloydii var. binata were photographed and assessed for biting and attachment by mycelia at the base of the legs and the mouthparts. Of 76 specimens, none evidenced biting, 53 were attached by mycelial growth from the mandibles, and all were attached by mycelial growth from the base of the legs.

To further evaluate the differential attachment to leaves by these related species, several specimens of both O. unilateralis and O. lloydii var. binata were carefully separated from their leaves. After sputter coating with 5 nm of gold-palladium, scanning electron micrographs were taken of the both the ant and the leaves. 41

Figure 5: Ophiocordyceps lloydii var. binata. Note the light tan mycelial growth covering the ant's body and attaching it to the leaf (red circle).

Upon examination of O. lloydii var. binata with the SEM, we quickly noticed the sparse fungal growth around the mandibles (Figure 7) and the dense fungal growth from the base of the legs (Figure 8). As we increased the magnification, we made a surprising discovery. Covering the space between the fungal mycelia growing from the base of the cadaver’s legs was something that looked very much like a membrane (Figure 9 and 10). Closer examination revealed that this was indeed a thin layer without any supporting structure underneath (Figure 10). Looking then at the leaf to which the cadaver was previously attached, we noted stark differences between the bare leaf surface and the leaf surface covered in fungal growth (Figure 11-13). Areas close to mycelia appeared smooth (Figure 11) while the leaf surface without fungal growth appeared rough (Figure 13). The smooth patches surrounding the fungus had clearly defined edges and appeared to radiate from fungal tissue. On the basis of these images, we hypothesize that O. lloydii var. binata secretes a glue-like substance to facilitate a stronger attachment of ant cadavers to leaves.

Figure 7: Ophiocordyceps lloydii var. binata mandibles. Note the clustering of fungal growth around the mouthparts (red circle).

Figure 8: O. lloydii var. binata mycelium growing from the base of the legs viewed from the basal surface after removal of the leaf. Note how the growth radiates outwards.

Figure 9: Higher magnification of the basal side of O. lloydii var. binata. Note the tubular mesh with a membrane (red arrow) between the strands.

Figure 11: Ophiocordyceps lloydii var. binata growth on a leaf surface. Note the smooth tubular growths.

Figure 12: The interface between the surface of a leaf (left, rough ~1 micron patches) and Ophiocordyceps lloydii var. binata growth (right, smooth tubular strands).

Figure 13: Leaf surface without O. lloydii var. binata growth. Note the rough appearance, particularly in comparison to Figure 11.

50 Ophiocordyceps unilateralis

After discovering what we believe to be a glue-like secretion on the leaf-fungus interface of O. lloydii var. binata, we examined O. unilateralis sensu lato to determine whether it also produced a similar membranous covering between mycelia. SEM examination of the basal surface of the cadaver showed that O. unilateralis does produce a fungal patch similar to that produced by O. lloydii var. binata between the base of the legs and the leaf (Figure 14). This growth does not, however, show any evidence of a membrane or secretion that could assist with attachment to leaves (Figure 15). We hypothesize that secreting a glue-like substance is energy intensive and that because O. unilateralis evolved the ability to induce ant biting, which results in a much stronger connection between the ant and leaf, the ability to secrete this substance was lost. Further studies to elucidate the composition of the membranous structure, to measure the energy requirements of secretion, and to quantify the difference in attachment strength between

O. lloydii var. binata and O. unilateralis sensu lato are necessary to further support or refute this hypothesis.

Figure 14: O. unilateralis basal surface after leaf was removed at magnification roughly equivalent to Figure 8.

Figure 15: O. unilateralis basal surface at higher magnification (similar to Figure 9). Note the lack of a membrane between mycelial strands.

53 References

Sung G, Hywel-Jones NL, Sung J, Luangsa-ard JJ, Shrestha B, Spatafora JW. Phylogenetic

classification of Cordyceps and the calvicipitaceous fungi. Stud Mycol. 2007; 57:5-59.

doi: 10.3114/sim.2007.57.01

Andersen SB, Gerritsma S, Yusah KM, Mayntz D, Hywel-Jones NL, Billen J, Boomsma JJ,

Hughes DP. The life of a dead ant: The expression of an adaptive extended phenotype.

Am Nat. 2009; 174:424-433 doi: 10.1086/603640

Evans HC, Samson RA. Cordyceps species and their anamorphs pathogenic on ants (Formicidae)

in tropical forest ecosystems II. The Camponotus (Formicinae) complex. Trans Br Mycol

Soc. 1984; 82:127-150

Academic Vita

Benjamin Fowler 717-380-4778 [email protected] Education The Pennsylvania State University, University Park, PA Schreyer Honors College College of Agricultural Sciences Immunology and Infectious Disease B.S. & Toxicology B.S. Global Health minor

Research, Clinical, and Work Experience “Elucidating the reproductive evolution and transmission mechanism of Ophiocordyceps” The Pennsylvania State University, University Park, PA January 2014 – Present David Hughes, Assistant Professor of Entomology and Biology

Honors Thesis: “Understanding Ophiocordyceps, the zombie ant fungus: A case study in host behavioral manipulation” Schreyer Honors College, University Park, PA David Hughes, Assistant Professor of Entomology and Biology

“Characterizing the comorbid disease burden and treatment outcomes of patients with drug-resistant HIV in Mbour, Senegal” The Pennsylvania State University, University Park, PA May 2015 - Present Rhonda BeLue, Associate Professor of Health Policy and Administration

“Assessing tropical ecosystems as indicators of global climate change” The Pennsylvania State University, Center Valley, PA April 2013 - July 2013 Jacqueline McLaughlin, Associate Professor of Biology

Fish Room Sales Associate That Fish Place – That Pet Place, Lancaster, PA May 2013 - August 2013

Extracurricular and Volunteer Experience Alpha Epsilon Delta University Park, PA Sept. 2013-Present President – May 2015-May 2016  The Health Preprofessional Honor Society – Pennsylvania Beta Chapter

HUB-Robeson Center University Park, PA Aug. 2013-Present Aquarium Volunteer  Assist in the maintenance of the 650-gallon Indo-Pacific aquarium in the Penn State student center  Train new volunteers and supervise all volunteers

Biomedical Sciences Club University Park, PA Aug. 2012-Present Webmaster – May 2015-May 2016 President – May 2014-May 2015

Mount Nittany Medical Center State College, PA Jan. 2014-Present Emergency Department Volunteer- August 2014-Present  Assist ER staff by transporting patients to and from tests (CT, MRI, X-Ray, & Ultrasound)  Deliver paperwork, bloodwork, medications, equipment, and more to requested individuals Patient Floors Volunteer- January 2014-Present  Interact with patients and families to provide directions, assistance locating staff, and more

Publications and Presentations Fowler, BD; Araujo, JPM; Hughes, DP. (29 March 2016) Quantifying the transmission potential of an entomopathogenic fungus that controls the behavior of ants. Gamma Sigma Delta Research Exhibition. The Pennsylvania State University, University Park, PA. Fowler, BD; Araujo, JPM; Hughes, DP. (2015) Quantifying the transmission potential of an entomopathogenic fungus that controls the behavior of ants. Submitted for review and publication. Fowler, BD; Evans, J; McLaughlin, J. (2014 March) Give Life a CHANCE. Pennsylvania Environmental Resources Consortium Student Symposium

Honors and Awards Honors National Inductee of Alpha Epsilon Delta: The Health Preprofessional Honor Society Dean’s List for 7 of 7 semesters (2012-present)

Research Grants Dec. 2015 - College of Agricultural Sciences Undergraduate Research Grant for spring 2016 May 2015 - College of Agricultural Sciences Undergraduate Research Grant for fall 2015 April 2015 - Schreyer Honors College Travel Grant for global health research in Senegal Dec. 2014 - College of Agricultural Sciences Undergraduate Research Grant for spring 2015 Nov. 2014 - Schreyer Honors College Travel Grant for field research in Brazil Oct. 2014 - College of Agricultural Sciences Tag Along Grant for field research in Brazil April 2014 - College of Agricultural Sciences Undergraduate Research Grant for summer 2014 April 2013 - Schreyer Honors College Travel Grant for field research in Panama

Scholarships Schreyer Honors College Academic Excellence Scholarship – 2012-2016 John N. Adam Jr. Scholarship for Excellence in Agriculture - 2015-2016 Theola F. Thevaos Honors Scholarship in the College of Agricultural Sciences - 2014-2015 Eva B. and G. Weidman Groff Memorial Scholarship - 2013-2014 Oswald Scholarship - 2012-2013