The Pennsylvania State University
The Graduate School
Intercollege Graduate Degree Program in Ecology
SOCIAL INSECTS AS SOLITARY VEHICLES
A Dissertation in
Ecology
by
Emilia Solá Gracia
2017 Emilia Solá Gracia
Submitted in Partial Fulfillment of the Requirements for the Degree of
Doctor of Philosophy
August 2017
ii The dissertation of Emilia Solá Gracia was reviewed and approved* by the following:
David P. Hughes Assistant Professor of Entomology and Biology Dissertation Advisor Chair of Committee
Nina Jenkins Senior Research Associate
Victoria Braithwaite Professor of Fisheries and Biology
Ephraim Hanks Assistant Professor
Jason Kaye Professor of Soil Biogeochemistry Professor in Charge
*Signatures are on file in the Graduate School
iii ABSTRACT
Parasite-host interactions affect more than the two main characters. Organisms living in tight-knit communities depend on each other in order to survive. Individuals harboring a parasite can affect the delicate balance within these communities. While on the other hand the community could also have a strong effect on a parasite’s life cycle. I explore the effects infected individuals have on the within-nest dynamics of ant colonies.
Additionally, I investigate the effects of infectious and non-infectious cadaver exposure have on worker behavior, as well as determine the effects workers have on the fungus protruding from the infectious cadavers. Using both a coevolved fungus, Ophiocordyceps unilateralis sensus lato, and a generalist fungus, Beauveria bassiana, I found infected workers do not strongly affect the within nest dynamics of Camponotus castaneus, the natural host of O. unilateralis s. l. Furthermore C. castaneus and C. pennsylvanicus manage infectious and non-infectious cadavers differently. Workers from C. castaneus colonies are adept in quickly removing cadavers with internal and external fungal development. While C. pennsylvanicus workers must come in contact with fungal tissue in order to recognize the disease threat. However, I found C. pennsylvanicus workers to be highly effective in reducing the infectivity of hazardous fungal conidia (i.e. asexual spores). Such findings could be caused by the wood nesting ecology this species has, as well as the higher aggression towards infectious cadavers. My research lays the foundation for future studies which focus on using semi-natural settings to study parasite- host and community dynamics.
iv TABLE OF CONTENTS
LIST OF FIGURES ...... vi
LIST OF TABLES ...... ix
ACKNOWLEDGEMENTS ...... x
DEDICATION ...... xi
Chapter 1: Introduction ...... 1
Background ...... 1 Behavioral mechanisms for parasite defense ...... 2 Objectives...... 6
Chapter 2: Within the fortress: the tolerance of a specialized parasite in a social insect society ...... 8
Introduction ...... 9 Materials and Methods ...... 12 Results ...... 19 Discussion ...... 22 Conclusion ...... 28
Chapter 3: Observations on a generalist entomopathogen within a social insect fortress...... 36
Introduction ...... 37 Materials and Methods ...... 40 Results ...... 45 Discussion ...... 47 Conclusion ...... 51
Chapter 4: Dynamics surrounding the dead: Do ant colonies recognize disease threats? ...... 60
Introduction ...... 61 Materials and Methods ...... 65 Results ...... 76 Discussion ...... 79 Conclusion ...... 84
Chapter 5: The effectiveness of social behaviors in reducing fungal pathogenicity ...... 103
Introduction ...... 104
v Materials and Methods ...... 107 Results ...... 114 Discussion ...... 116 Conclusion ...... 118
Chapter 6 Conclusion anf future directions ...... 128
Appendix A: Supplementary material for Chapter 2 ...... 132
Appendix B: Supplementary material for Chapter 3 ...... 137
Appendix C: Supplementary material for Chapter 4 ...... 140
Appendix D: Supplementary material for Chapter 5 ...... 143
References ...... 146
vi LIST OF FIGURES
Figure 2.1- Proportion of time spent in trophallaxis 6 days post-injection...... 29
Figure 2.2- Extended trophallaxis observations ...... 30
Figure 2.3- Mean distance between focal individuals and nest entrance ...... 32
Figure A .1- Mortality cuves ...... 132
Figure A.2- Time spent within the nest ...... 133
Figure 3.1- Proportion of time workers spent in trophallaxis while inside the nest ...... 53
Figure 3.2- Number of trophallaxis events occuring within the nest ...... 55
Figure 3.3- Percent of the time individuals spent within the nest during our
observational bouts ...... 57
Figure 3.4- Proportion time individuals performed vigilance behaviors while inside the
nest ...... 58
Figure B.1- Nest vigilance categorization ...... 137
Figure B.2- Healthy Beauveria bassiana growth from cadaver ...... 138
Figure B.3- Survival rate per treatment...... 139
Figure 4.1- Time spent within th nest, percent of contacts, mortality, and infection for
Camponotus castaneus nests which are open for the first 12 hours of exposure ...... 86
vii Figure 4.2- Time spent within th nest, percent of contacts, mortality, and infection for
Camponotus pennsylvanicus nests which are open for the first 12 hours of exposure .... 88
Figure 4.3- Time spent within th nest, percent of contacts, mortality, and infection for
Camponotus castaneus nests which are closed for the first 12 hours of exposure ...... 90
Figure 4.4- Time spent within th nest, percent of contacts, mortality, and infection for
Camponotus pennsylvanicus nests which are closed for the first 12 hours of
exposure ...... 92
Figure 4.5- Rate of gaster bending within nests which are closed for the first 12 hours of
exposure ...... 94
Figure 4.6- Rate of gaster bending within nests which are open for the first 12 hours of
exposure ...... 95
Figure 4.7- Survival post exposure ...... 97
Figure 4.8- Percent of individuals near the cadaver within nests containing Camponotus
castaneus workers ...... 99
Figure 4.9- Percent of individuals near the cadaver within nests containing Camponotus
pennsylvanicus workers ...... 101
Figure C.1- Schematic representation of the nest architecture and cadaver treatments
used for Camponotus castaneus ...... 140
Figure C.2- Schematic representation of the nest architecture and cadaver treatments
used for Camponotus pennsylvanicus ...... 141
viii Figure C.3- Confirmation of conidia production on external mycosis cadavers ...... 142
Figure 5.1- Images of a cadaver introduced to an empty nest for 24 hours ...... 120
Figure 5.2- Images of a cadaver introduced to a closed nest with workers for 24 hours ...... 121
Figure 5.3- Images of a cadaver placed within an open nest with workers for 24 hours ...... 122
Figure 5.4- Mean worker survival post exposure ...... 124
Figure 5.5- Mean conidia concentration per sample ...... 125
Figure 5.6- Mean conidia germination rate per treatment ...... 127
Figure D.1- Before and after photographsof an exemplar cadaver used for the “nest + no
workers” treatment ...... 143
Figure D.2- Before and after photographsof an exemplar cadaver used for the “open nest
+ workers” treatment ...... 144
Figure D.3- Before and after photographsof an exemplar cadaver used for the “closed
nest + workers” treatment ...... 145
ix LIST OF TABLES
Table 2.1- Results from mixed-effect model on trophallaxis 6 days post-injection...... 33
Table 2.2- Analysis performed on our extended dataset...... 34
Table 2.3- Analysis for the proximity of individuals and the entrance of the neston day 6
post-injection...... 35
Table A.1- Sample sizes for present time spent within th nest ...... 134
Table A.2- Colony and subcolony observation use ...... 134
x
ACKNOWLEDGEMENTS
My personal life at the beginning of this program was one that required a lot of strength. Continuing my graduate degree became a source certainty, but it was and still is challenging. Over the course of this roller coaster ride, I have met many people and am honored to have met them. I cannot possibly thank enough all of the people I have relied on, but I will focus on a few key individuals Within my academic development, I must first recognize my advisor, David. I started this program as a young, wide-eyed undergraduate, and now I feel comfortable saying I have become a scientist. Others who have had a strong hand in molding my knowledge and scientific mind are my committee members, Nina, Ephraim, and Victoria. The input, support, and critiques have made me a better academic; I am grateful. I would also like to thank those who allowed me to continue my academic pursuit by helping me behind the scenes. I thank NSF and PSU for their financial support. I am also grateful to all the staff and faculty within the Ecology Program, Entomology Department, and the Center for Infectious Disease Dynamics. Within my personal life, there are far too many people to thank. A complete list would be far too overwhelming. Laura Drew became my adoptive sister over the course of my PhD, and I am forever grateful to have her be my sister. Wes E. Neal has become the wise elder who has helped me develop numerous soft and hard life skills, including editing and budgeting. And I cannot leave out the group of people who have supported, listened, and sometimes calmed me down when science was not cooperating. I extend my deepest appreciation and gratitude to Ryan Bringenberg, Kelsee Baranowski, Melissa Ishler, Lauren Quevillion, João Araújo, Raquel Loreto, Charissa de Bekker, Maridel Federicksen, and other Hughes lab members, past and present. Sometimes my focus blinded me from solutions, and you are the people who helped me find the path to answers. Thank you.
xi DEDICATION
I would like to dedicate my accomplishment to two people who became my rocks in the initial stages of my journey. Without their initial support, guidance, love, and encouragement, I would have embarked on a different path. My mother, Irma M. Gracia
Solá, pushed me to become independent and strong. Without her as an example, I would not be the woman I am today. I have no earthly idea where you are at this moment, Mom, but I hope our paths cross again. I would also like to dedicate my accomplishment to José
J. Muñoz Díaz, my first love. You left too soon, but your positive and loving energy stayed with me. You have made me want to fight harder.
Words struggle to capture the emotions that emerge when I think about the both of you. I know my accomplishment makes you proud, and I hope you can somehow see me reach the finish line we talked about so many times. The fact that I was able to cross the finish line without you is the embodiment of your support, strength, and love which have become a part of me.
1
Chapter 1: Introduction
Background
Parasite-host interactions go beyond the two main characters, host and parasite. If the host lives within a social group, an infection could have an effect on how that individual interacts with other group members. The parasitic infection can then precipitate a conflict between a parasitized individual and the social group to which it belongs. Such individuals can infect other group members (e.g. intestinal infection in bees, Otterstatter & Thomson, 2007), reduce group productivity (e.g. gregarines affecting the rate of reproductives produced within a social wasp nest, Bouwma, Howard, &
Jeanne, 2005), or sequester group resources towards parasite development (e.g. tapeworm affecting the energy management within an ant colony, Scharf, Modlmeier, Beros, &
Foitzik, 2012). Such conflicts imply the group would have an advantage if it were able to recognize an infection and eradicate the sources of conflict. A parasite would require the ability to hide in plain sight to take advantage of such a socially complex system (i.e. ant colony). Endoparasites are likely to have an advantage by disguising themselves within their host. Living undetected within the group would allow a parasite to take advantage of the social group’s resources with little to no harm to its development and spread. Testing how complex social groups like ants handle the presence of infected individuals and
2 manage sources of infection within the nest could give us insight into how such groups manage disease.
Ants’ terrestrial lifestyle, coupled with the constant contact and high relatedness among colony members, presents ideal conditions for parasites to take advantage of resources of the group (W. D. Hamilton, 1987). Workers are infected by a wide variety of endoparasites, including fungi (e.g. Pandora, Ophiocordyceps), trematodes (e.g.
Brachylecithum, Dicrocoelium), cestodes (e.g. Raillietina, Anomotaenia), nematodes
(e.g. Mermis, Tetradonema), strepsipterans (e.g. Caenocholax), dipterans (e.g.
Pseudaceton, Styletta), and hymenopterans (e.g. Alucha, Eucharomorpha) (Schmid-
Hempel, 1998). Furthermore, ant colonies are also infected with social parasites, organisms which take advantage of socially managed resources. Such social parasites include beetles (e.g. Ecitophya, Dinarda), flies (e.g. Forcipomyia, Malaya), caterpillars
(e.g. Maculinea, Epizeuxis), other species of ants (e.g. Protomognathus, Polyergus), and mollusks (e.g. Nagurus) (Foitzik, DeHeer, Hunjan, & Herbers, 2001; Guillem, Drijfhout,
& Martin, 2014; Kistner, 1982). The diversity and high prevalence of parasites amongst ant species would suggest ant colonies constantly succumb to parasites (Boomsma,
Schmid-Hempel, & Hughes, 2005). However, the defense mechanisms used by ant colonies against parasites have enabled ants to become one of the most abundant and diverse organisms on the planet (Bert Hölldobler & Wilson, 1990, pp. p.1-3).
Behavioral mechanisms for parasite defense
Ant colonies have multiple layers (worker versus colony) of defense against parasites. Ant workers are in constant contact with potential parasite sources, either
3 through managing nest material (Reber & Chapuisat, 2012), consuming food (Poinar &
Yanoviak, 2008; Reid & Nugara, 1961), or foraging (Elizalde & Julia Folgarait, 2012;
Raquel G. Loreto, Elliot, Freitas, Pereira, & Hughes, 2014). Workers have evolved behavioral and physiological adaptations to reduce infection. Such adaptations include the performance of grooming behaviors and production of substances with antibiotic properties. Grooming is the first line of defense for ants against parasites (Zhukovskaya,
Yanagawa, & Forschler, 2013).
The removal of infectious material from the cuticle reduces worker and brood mortality. Workers exposed to asexual fungal spores (i.e. conidia) have higher chances of survival when placed within a group of nestmates (Theis, Ugelvig, Marr, & Cremer,
2015). Brood and workers that are exposed to conidia are groomed more than individuals exposed to inert material (Myles, 2002; Reber, Purcell, Buechel, Buri, & Chapuisat,
2011). Tragust et al. (2013) demonstrated conidia removal is important for brood survival after exposure to conidia. Furthermore, brood survival increased when workers had the ability to also use their metapleural gland and acidopore (i.e. poison gland)
(Simon Tragust, Mitteregger, et al., 2013).
In order to reduce the viability and infectivity of parasites, ants have evolved physiological defense mechanisms, which include production of antibiotic substances from glands within their bodies. Most ant species have the metapleural gland, which secretes a substance with antibiotic properties. Workers spread these secretions while grooming, developing a thin layer of antibiotics over the cuticle (Fernandez-Marin,
Zimmerman, Rehner, & Wcislo, 2006; B. Hölldobler & Engel-Siegel, 1984). Leaf-cutting ants heavily depend on the metapleural gland and their microbial symbionts in order to
4 maintain the health of their fungal garden by reducing the garden’s exposure to fungal pathogens (Little, Murakami, Mueller, & Currie, 2006; Poulsen, Bot, Nielsen, &
Boomsma, 2002). However, ants within the genera Camponotus, Dendromyrmex,
Oecophylla, and Polyrhachis do not have a metapleural gland, but rather use the formic acid produced within in their poison gland for defense (Bert Hölldobler & Wilson, 1990, p. 30).
Although formic acid is used as a defense against predation and as an alarm pheromone, formic acid also has antibiotic properties (Attygalle & Morgan, 1984; Blum,
Walker, Callahan, & Novak, 1958; Storey, Vandermeer, Boucias, & McCoy, 1991) and is highly effective at inhibiting parasite development (Blum et al., 1958; Li, Jin, & Chen,
2012; Storey et al., 1991). After Lasius neglectus brood had been exposed to conidia of
Metarhizium anisopliae, workers groomed and sprayed the brood with formic acid to increase brood survival (Simon Tragust, Mitteregger, et al., 2013). To manipulate the colony, parasites need to overcome workers’ grooming behavior and the colony’s social defenses.
The presence of parasites within workers has been shown to affect colony dynamics. When a subset of Camponotus pennsylvanicus workers has a bacterial infection, they increase their performance of trophallaxis. During the process of socially sharing food, workers transfer agents with antimicrobial properties, including cathepsin D
(C. Hamilton, Lejeune, & Rosengaus, 2011). However, after workers are exposed and potentially infected with fungus, antimicrobial exchange would be inadequate, since the fungal tissue develops within the hemolymph. In such a scenario, a worker could change tasks and spatially isolate in order to reduce the exposure of other workers. When Lasius
5 neglectus workers return to the nest after being exposed to M. anisopliae (a lethal entomopathogen), the colony changes how brood are cared for, only workers who have not been exposed nurse the brood, while those exposed spend less time within the brood chamber and significantly reduce their contact with brood (Ugelvig & Cremer, 2007).
However, disease threats can also come from within the nest. Waste produced within the nest can harbor parasites and potentially cause disease outbreaks.
Waste management is another method used to reduce the threat of infection
(reviewed in: Lopez-Riquelme & Fanjul-Moles, 2013). Ants perform a variety of behavioral repertoires to manage colony members that have died. Such behaviors include the removal of dead bodies (i.e. necrophoresis) (Diez, Lejeune, & Detrain, 2014), the placement of dead bodies within a refuse pile (Bot, Currie, Hart, & Boomsma, 2001), and in some occasions the consumption of dead bodies (Howard & Tschinkel, 1976). The performance of these behaviors could inhibit parasite development and increase colony health. Brood survival has been shown to increase when cadavers are properly managed within the ant nest (Diez, Deneubourg, & Detrain, 2012). Colony health and survival depends on the ability of workers to detect and react to possible sources of disease.
However, the majority of studies performed to observe how ants manage disease have been made using artificial arenas, staged encounters, and short observation bouts
(for example: N. Bos, Lefevre, Jensen, & d'Ettorre, 2012; Heinze & Walter, 2010;
Konrad, Grasse, Tragust, & Cremer, 2015; Konrad et al., 2012; Masri & Cremer, 2014;
Theis et al., 2015; Simon Tragust, Mitteregger, et al., 2013; Simon Tragust, Ugelvig,
Chapuisat, Heinze, & Cremer, 2013; Ugelvig & Cremer, 2007; Westhus et al., 2014).
Although the simplicity is convenient and leads to many publications entering the
6 literature (R.G. Loreto & Hughes, 2016 Figure 1), this type of methodology does not necessarily answer biologically relevant questions. Collecting behavioral data for 10 minutes per day is not sufficient to capture the changes in behavior over time.
Furthermore the dependence on petri dishes and staged encounters between workers creates artificial scenarios that likely alter the results of the experiment. Such methodology could lead to overly simplistic conclusions which are not informed by biologically meaningful behavior. Using semi-natural nests or arenas coupled with longer observation bouts could lead to more biologically relevant conclusions.
Objectives
The basis of my PhD dissertation is to explore the effects parasites have on within-nest social dynamics of ant colonies. In order to make these observations, I used two ants species (Camponotus castaneus and Camponotus pennsylvanicus), along with two fungal parasites (Ophiocordyceps unilateralis sensu lato and Beauveria bassiana).
The observations made in Chapter 2, using O. unilateralis s. l. and C. castaneus, allowed me to better understand how individuals infected with a co-evolved, behavior manipulating fungal parasite affect the social dynamics within the nest. I was particularly interested in determining if infected individuals become resource sinks for the colony’s food resources by increasing the rate of social food exchange, or if these individuals are recognized as social parasites and potentially removed from the nest by uninfected ants.
After performing my observations for Chapter 2, I became interested in determining if my results were replicable within a different system with a non-coevolved pathogen. Thus, in Chapter 3, I infected C. castaneus workers with B. bassiana to
7 determine if infected individuals changed their behavior over the course of fungal development. After seeing no negative reactions towards live workers infected with O. unilateralis s. l. or B. bassiana, I became interested in testing the ability workers have in detecting fungal growth within nestmate cadavers.
In Chapter 4, I tested the abilities of workers from C. castaneus and C. pennsylvanicus species to detect B. bassiana development within the cadavers of nestmates. I used three different perspectives (cadaver, exposed, and group) to ascertain the ability for a worker to recognize and reduce disease spread within the nest after the exposure to infectious material. During my observation in Chapter 4, I noted strong changes in fungal and cadaver structure after being exposed to live workers.
I became interested in empirically testing the effects live workers have on fungal reproduction. In Chapter 5, I tested whether the reproduction and structure of B. bassiana was influenced by exposure to live C. pennsylvanicus workers. Finally, in Chapter 6, I discussed the findings from the previous chapters and suggested future research directions.
Chapter 2: Within the fortress: the tolerance of a specialized parasite in a
social insect society
Emilia Solá Gracia 1, 2, 3, Charissa de Bekker 6, Ephraim M. Hanks 2, 5, and David P.
Hughes 2, 3, 4
1. Ecology Program, Huck Institutes of Life Sciences, Pennsylvania State
University, University Park, Pennsylvania, 16802
2. Centre for Infectious Disease Dynamics, Huck Institutes of Life Sciences,
Pennsylvania State University, University Park, Pennsylvania, 16802
3. Department of Entomology, Pennsylvania State University, University Park,
Pennsylvania, 16802
4. Department of Biology, Pennsylvania State University, University Park,
Pennsylvania, 16802
5. Department of Statistics, Pennsylvania State University, University Park,
Pennsylvania, 16802
6. Department of Biology, University of Central Florida, Orlando, Florida, 32816
9 Introduction
Cooperation is a major theme in biological organization as different units, from cells to individuals, come together to form a whole which is greater than the sum of its parts (A. F. G. Bourke, 2011). Social insect societies are considered to be paragons of cooperative behavior where individual units (i.e. workers) forgo direct fitness to increase the reproductive output of other individuals (i.e. queens and males). Such altruism is evolutionarily stable because colonies of social insects are composed of kin groups (W.
D. Hamilton, 1963, 1964a, 1964b). However, such an altruistic system is inherently susceptible to cheating if the cheaters can mimic the cues displayed by bona fide colony members.
Within ant societies, diverse organisms, ranging from other ants to beetles, flies, caterpillars, and even mollusks, have evolved ways to break the colony code and become social parasites (Foitzik et al., 2001; Guillem et al., 2014; Kistner, 1982). Social parasitism is “a relationship between two species where the parasite benefits in many ways from brood care or other socially managed resources of its host” (Schmid-Hempel,
1998, pp. 111-112). Social parasites enter the society, in this case ant nests, to obtain resources but do not enter the bodies of colony members (Lachaud, Lenoir, & Witte,
2012; J. A. Thomas, Schonrogge, & Elmes, 2005).
Social parasites must develop strategies to enter and remain inside the colonies they infect. Maintaining residence within a colony can be a challenge for these parasites because they can be detected by colony members through chemical cues. Social parasites typically must disguise their presence through chemical mimicry. For example,
10 caterpillars of the blue butterfly (Maculinea rebeli) mimic the chemical profile of larval ants and are carried into the nest by foraging ant workers. Within the nest, the caterpillars then feed on food foraged by the workers and have been observed to consume colony brood (i.e. larval ants and the eggs) (Akino, Knapp, Thomas, & Elmes, 1999; Elmes,
Thomas, & Wardlaw, 1991; Elmes, Wardlaw, & Thomas, 1991; Guillem et al., 2014).
On the other hand, parasites can also infiltrate the nest by using the body of a colony member. Ant endoparasites include diverse genera, which include fungi (e.g.
Pandora, Ophiocordyceps), trematodes (e.g. Brachylecithum, Dicrocoelium), cestodes
(e.g. Raillietina, Anomotaenia), nematodes (e.g. Mermis, Tetradonema), strepsipterans
(e.g. Caenocholax), dipterans (e.g. Pseudaceton, Styletta), and hymenopterans (e.g.
Alucha, Eucharomorpha) (Schmid-Hempel, 1998). Each of these endoparasites contend not only with their host’s social environment, but also with the individual host’s body
(Rolff & Siva-Jothy, 2003). Little work has been done to observe and understand whether individuals infected by a specialized endoparasite are detected by the colony.
It may be reasonable to hypothesize that the colony will detect infected workers, either because the infected individuals represent a source of infection (current or future), or because they represent a drain to the colony’s resources (nutrients go to the developing parasite and not to the reproductives of the colony). An alternative hypothesis is that specialized parasites coevolve with their hosts to develop strategies which limit the colony’s ability to detect the parasite’s presence. By avoiding detection, the parasite takes advantage of both the host and the host’s society without adverse effects on the parasite’s development.
11 To explore these hypotheses, we examined an ant-parasite system which is amenable to a controlled laboratory setting where infections and behavioral dynamics within the nest can be observed. We chose to expose an ant society to the fungal endoparasite Ophiocordyceps unilateralis sensu lato (O. unilateralis s. l.) to test these hypotheses. The coevolved fungal endoparasite O. unilateralis s. l. manipulates worker behavior just before death and can become a future disease threat to colony foragers. We tested whether uninfected members of a carpenter ant colony, species Camponotus castaneus, recognized siblings infected by O. unilateralis s. l.
In recent years, a number of studies have demonstrated that species within the species complex O. unilateralis s. l. manipulate the behavior of Camponotus and
Polyrhachis workers by causing infected individuals to leave the colony and bite into vegetation before dying (Andersen et al., 2009; Hughes et al., 2011; Pontoppidan,
Himaman, Hywel-Jones, Boomsma, & Hughes, 2009). The fungus transitions from growing within the ant's body to growing a stalk from which spores (i.e. infectious particles) are produced and released onto the forest floor (Andersen et al., 2009). Spores are large, heavy, and not dispersed by wind (Araújo & Hughes, 2016; Evans, Elliot, &
Hughes, 2011), and therefore fall in a concentrated area under the cadaver. The fungus uses the cadaver as a platform to release spores. These spore platforms are placed near the host’s colony and foraging trails, potentially increasing disease transmission to the siblings of the now-dead individual (Raquel G. Loreto et al., 2014; Pontoppidan et al.,
2009).
12 Infected individuals inside the nest transition to spore producing cadavers outside of the nest, making it beneficial for the colony to detect infected individuals and take action in order to reduce future disease threats. In this study we tested if infected individuals: 1) are attacked by nestmates, 2) spend more or less of their time socially exchanging food, and 3) are spatially separated from the colony. We hypothesize these infected workers will be recognized by the colony, and will be removed by colony members. We furthermore hypothesize infected workers will spend significantly more time socially exchanging food with other workers, and become spatially isolated from other members colony. Infected individuals must sustain themselves and the developing parasite, potentially requiring an increase in food consumption or proportion of time spent in trophallaxis.
Materials and Methods
Ant collection and stock colony maintenance
We collected Camponotus castaneus colonies in Douglas, South Carolina during
April and May of 2012 by digging the soil near the nest entrance. Collecting queen-right colonies is difficult because colony architecture is hard to predict, and it is commonplace to find deep tunnels, which makes the queen hard to extract. We used three colonies within this experiment. Colony 1, collected April 2012, consisted of unmated reproductives that had not yet left for their natal flight, brood, and about 120 workers.
Colony 2, collected May 2012, consisted of brood and about 100 workers. Colony 3,
13 collected May 2012, consisted of brood and about 100 workers. We provided all colonies with water and 10 % sugar water ad libitum, which we replenished once a week. We also provided dead crickets (supplied by Fluker’s Farms) as a source of protein for the developing brood.
Injection and infection techniques
Due to the limited knowledge of abiotic and biotic factors necessary for the fungus to produce sexual structures in the laboratory, we were unable to use the natural pathway of infection and therefore employed artificial methods. We followed the O. unilateralis s. l. infection protocols in a similar fashion as those successfully employed in previous studies (de Bekker et al., 2015; de Bekker et al., 2014). Fungal hyphae from a single fungal colony were placed in a sterile 2 mL tube with two 0.63 cm metal balls
(Wheels Manufacturing, Inc.) and 200 µL Grace’s medium (Sigma) freshly supplemented with 10% Fetal Bovine Serum (FBS, PAA Laboratories, Inc.). We lysed the fungal colony tissue using a TissueLyser II (Qiagen) at room temperature for 60 seconds at 30 cycles per second. This process enabled us to obtain small segments of fungal hyphae, which we then used at a mean concentration of 3.9x107±1.1x107 hyphae per mL for injection. We infected workers by injecting 1 µL fungal hyphal solution with a laser- pulled 10 µL micropipette (Drummond) and aspirator tube (Drummond) into the thorax underneath the prothoracic legs. Sham treatments were done in similar fashion using 1
µL Grace’s medium supplemented with 10% FBS without fungal tissue.
Establishing infection status
14 In order to determine proper infection we performed daily mortality observations after injections occurred (Figure A.1). Any cadavers collected during the experiment were surface sterilized by oscillating them within 70% ethanol for 20 seconds. After surface sterilization, we placed each cadaver in a sterile petri dish (100×15 mm dimension) containing a Whatman 541 (70 mm diameter) filter paper moistened with 250
L of sterile water and incubated them at 28˚C. In order to determine if these individuals died due to an O. unilateralis s. l. infection we monitored the cadavers for fungal growth once a day.
Treatments and individual identification of C. castaneus workers
From each of the three colonies (colonies 1-3), we established two subcolonies for a total of 6 subcolonies in all (depicted in Table A.2). Each of the 6 subcolonies contained a total of 35 adult workers. We partitioned individuals into three groups: untreated individuals (“healthy” treatment, n=15 workers); individuals injected with O. unilateralis s. l. plus media (“infected” treatment, n=10 workers); and individuals injected with media alone (“sham” treatment, n=10 workers). In order to follow individuals through time, we marked them with unique dot patterns on their head, thorax, and gaster using Edding®, number 751 paint markers. Each treatment followed a specific pattern, either using color or dot location.
Within-nest behavioral observations
We began our observations three days after we performed our injections to give the subcolonies time to settle after the treatment. To observe within-nest behavior, we
15 used a modified GoPro camera (Hero 2 fitted with both an infrared [IR] lens and a 4.6 mm macro lens). We kept colonies under a 12:12 day-night light cycle with visible spectrum lights from 0600-1800 and infra-red light for the remainder of the 24 hours.
Since ants cannot detect light in the infra-red range, it appears dark to them. We situated the camera on top of the colony chamber, and we recorded for 24 hours except for three daily changes of the memory cards (which took between 5 and 15 seconds per change).
We housed our experimental subcolonies in a wooden chamber ranging from 14.93 to
15.46 cm2. Each wooden chamber was placed in an individual 452 cm3 arena with a sandy floor, which served as a foraging arena. We gave all experimental subcolonies water and 10 % sugar water ad libitum. We performed behavioral observations by following focal individuals within the nest (scoring aggression: total 585 hours, trophallaxis: 2,399 hours of observations, and spatial data: collected within an 8 hour time frame). The number of focal individuals that we followed within each treatment was determined by the number of infected individuals within the nest. After counting the number of known infected individuals within the nest, we followed the same number of sham and healthy individuals.
Scoring aggression
We measured aggression towards infected individuals over the course of the internal fungal development. In order to reduce observational bias, a single observer
(ESG) watched the videos. We classified aggressive behavior as mandible spreading, gaster bending, and jolting. We followed focal individuals within the nest until all infected individuals within the subcolony left the nest. Since the time of nest departure
16 was variable between subcolonies the duration of observation time was also different between each subcolony we observed. The observation periods for the first subcolony of each genetic colony were as follows: colony 1 - 152 hours, days 3-8 post-injection; colony 2 - 177 hours, days 3-12 post-injection; and colony 3 - 256 hours, days 3-11 post- injection. The total time was 585 hours of video. Video was played at 10 times the normal speed and paused or played at normal speed if any abnormal behavior occurred.
Scoring food exchange
We followed focal individuals from each of the treatments within the nest to determine if social food exchange (termed trophallaxis) between infected and uninfected individuals differed while O. unilateralis s. l. developed within the hosts’ bodies. A single observer (ESG) collected the behavioral data to reduce observational bias. We classified trophallaxis as starting when the labrum and labium (i.e. mouthparts of the maxillo-labial complex) were exposed and distended between the two individuals. The event ended when the mouthparts separated and mandibles closed. In order to adequately collect these data, we focused on one worker at a time. We watched the behaviors of one focal individual over the course of the entire video. When we finished collecting the data for one individual we would go back to the beginning to follow a new focal individual.
We collected data on day 6 after injection during the daylight hours (0900-1700) from six experimental subcolonies (three distinct genetic colonies), collecting a total of
655 hours of observation amongst all the colonies. We chose day 6 post-injection because previous dissections of infected ants revealed that the fungal development was apparent and active by that time (unpublished data). We followed a total of 89 focal individuals,
17 pooling both subcolonies of each genetic colony together: colony 1 (10 infected, 9 sham, and 9 healthy), colony 2 (10 infected, 11 sham, and 11 healthy), colony 3 (9 infected, 10 sham, and 10 healthy).
Furthermore, we collected trophallaxis data over the course of seven days
(hereafter referred to as the extended dataset). To determine if trophallaxis behaviors differed over the course of the initial fungal development within three of our subcolonies
(colony 2: subcolony 1; colony 3: subcolony1 and subcolony2). We only scored trophallaxis behavior in colonies 2 and 3, having lost the video footage on colony 1 and part of colony 2 due to a failed computer hard drive. We continuously followed 46 focal individuals throughout the daylight hours of 0900-1700. The distributions of focal individuals are as follows: colony 2 (7 infected, 5 sham, and 5 healthy) and colony 3 (12 infected, 10 sham, and 10 healthy). Over the course of seven days, we made observations on a total of 2,399 hours.
Since workers could leave the observation arena (nest chamber) freely, we expressed trophallaxis as a proportion: time spent in trophallaxis is equivalent to total time spent in trophallaxis divided by the total time spent within the nest; taking into account the significant differences in time spent within the nest amongst individuals and treatments. We found infected individuals spent significantly less time within the nest in comparison to other treatments (Figure A.2).
To analyze the data collected on day 6 post-injection and our extended dataset, we used permutation tests followed by data analysis using mixed-effect models. Since our response variable was a pair-wise event (trophallaxis between two individual ants), the
18 usual assumption of independence between observations was violated. Therefore, we used permutation tests, a conservative approach, to determine significance in our analysis. The permutation test allowed us to create 10,000 datasets in which the treatment of each focal individual was randomly assigned. After the creation of these datasets, we performed the analysis using a mixed-effect model, comparing our collected data to the randomly created data sets. Further explanations can be seen within the pseudocode found in our supplemental material. We analyzed the square root transformation of our proportion of time spent in trophallaxis in relationship to treatment as a fixed effect, while we deemed colony and subcolony identification as random effects. The extended dataset analysis was similar but required taking into account how time may affect our variables. We used three different mixed-effect models to take into account how time affects our data: 1) treating day post-injection as a factor, and maintaining the interaction term between day post-injection and treatment; 2) treating day post-injection as a factor and eliminating the interaction term between day post-injection and treatment; and 3) not treating day post-injection as a continuous variable (not as a factor) and maintaining the interaction term between day post-injection and treatment.
Spatial analysis of focal ants inside the nest
To determine spatial separation from the social group, we measured the distances between focal individuals and the entrance of the nest. Ants closest to the entrance of the nest, and therefore closest to the outside, could easily leave the nest. A single observer collected all the data (ESG) to reduce observational biases in data collection. We used all three colonies and subcolonies, with the exception of one (colony 2: subcolony 1), within
19 this dataset: colony 1 (10 infected, 9 sham, and 9 healthy individuals), colony 2 (7 infected, 6 sham, and 6 healthy individuals), and colony 3 (9 infected, 10 sham, and 10 healthy individuals); total n was 26 infected, 25 sham, and 25 healthy individuals. We measured distances between the head-thorax juncture for each focal ant inside the nest and the center of the nest entrance every minute from 0900-1700 on day 6 post-injection
(n=35,691 data points). In order to collect these data, we created a Python script which created frames (screenshots) of the video every minute (code available upon request).
These frames allowed us to follow our focal individuals within the image. The program allowed us to collect the x-y coordinates by using the mouse to click on the focal individual of interest. The analysis of this spatial dataset was performed by using permutation tests and a mixed-effect model in which we deemed treatment as a fixed effect, while colony and individual identification were treated as random effects. Within this dataset, the assumption of independence is violated because if one ant moves it likely causes another ant to move. Therefore, we used a permutation test, which allowed us to perform a conservative analysis of our data.
Results
Survival analysis
We injected the specialized fungal parasite O. unilateralis s. l. into 60 workers.
However, we did not achieve any external development post mortem. Due to an abundance of environmental fungi in the genus Aspergillus, which quickly colonized the
20 cadavers, we were unable to obtain an ideal scenario to develop O. unilateralis s. l. from our cadavers. We are confident these fungi are not parasitic as no records exist of them infecting ants (Araújo & Hughes, 2016). But to lend support to the assertion that O. unilateralis s. l. was the cause of death we analyzed the survival probability by using a
Kaplan-Meier analysis and found infected individuals had a lower survival probability than sham and healthy treatments (Figure A.1; Kaplan-Meier log rank: P<0.05). The time of death is within the same range as we observed in other studies (de Bekker et al., 2014).
Since sham treated individuals had a low level of mortality (below 20%), we can conclude our injections with O. unilateralis s. l. and not the injection process itself was the cause of mortality.
Infected ants are not attacked by siblings
We did not observe aggression within the nest between healthy and infected treatments nor between sham and infected treatments (n=585 hours of observations).
Furthermore, we did not see aggressive interactions within or between other treatments.
Infected ants engaged in trophallaxis with siblings
When comparing treatments on day 6 post injection, we found individuals within the sham treatment spent significantly less time in trophallaxis compared to individuals within the healthy treatment (GLMM: P< 0.05; Figure 2.1 and Table 2.1). We found no significant difference between individuals within the infected treatment and uninfected treatments. In order to verify our results, we performed the same analysis but removed the outlier within the infected treatment. However we arrived at similar results: the sham
21 treatment is significantly different to the healthy treatment, results which are likely due to the wide daily and individual variation in the proportion of time spent in trophallaxis.
In the extended dataset (i.e. data collected across 7 days), we found the proportion of time in trophallaxis between infected and healthy individuals is significantly different throughout all three of our mixed effect models (GLMM: P< 0.05; Figure 2.2 and Table
2.2). We found that, overall, individuals within the infected treatment spent less time in trophallaxis in comparison to the healthy treatment. Within our second model, we found all treatments to be significantly different from one another. However, we must recognize the daily variations seen over the course of these 7 days were large.
Our results suggest individuals within the infected treatment change their social food sharing rate within the nest over the course of O. unilateralis s. l. development.
However, when we take a single day into account, the high variability in behavior can change the relationship between treatments. In this case, we saw individuals within the sham treatment spending a significantly smaller proportion of time socially sharing food in comparison to the healthy treatment.
Infected ants are not socially isolated within the nest
To determine spatial isolation within the nest, we used worker proximity to the nest entrance as a proxy for isolation. Similar to trophallaxis, these data violate the assumption of independence. In order to take this into account, we used permutation tests coupled with mixed effect models for data analysis. Within the dataset, we found infected individuals to be significantly closer to the nest entrance in comparison to healthy
22 individuals (GLMM: P<0.05; Figure 2.3 and Table 2.3) suggesting infected individuals experience spatial isolation inside the nest, facilitating their ability to leave the colony more readily. However, this also placed the infected individuals in a location where interactions with others happen at a steady rate. Furthermore, this spatial isolation could be cause by either workers isolating themselves or subtle behaviors performed by uninfected nestmates which causes the isolation. It is unclear which behavioral mechanism is taking place within our observations.
Discussion
Our experiment set out to answer if infected individuals: 1) were attacked by their nestmates, 2) spent more (or less) of their time socially sharing food than their uninfected peers, and 3) became spatially isolated from the colony. We had two distinct hypotheses.
First, we hypothesized infected individuals would be recognized and attacked by nestmates to decrease the future threat to the colony. We expected that any increased aggression would also lead to infected individuals becoming spatially isolated. Secondly, we hypothesized infected individuals may have an increased food exchange reflective of the increased need for resources as the infection progresses within each worker’s body. In order to test these hypotheses, we observed the interactions between healthy C. castaneus workers and those infected with O. unilateralis s. l. within a semi-natural cage setting
(containing a nest with other workers and a sandy foraging arena).
23 We observed no within-nest attacks towards individuals infected with O. unilateralis s. l. We continuously followed individuals within the nest, enabling us to have a large number of behavioral observations over the course of parasite development, approximately 585 hours total. Aggression towards infected individuals has been suggested in other social insect systems, such as bees infected with deformed wing virus and termites infected with an entomopathogenic nematode (Baracchi, Fadda, &
Turillazzi, 2012; Fujii, 1975, pp. 35-36). However, the evidence presented by Baracchi et al. (2012) and Fujii (1975, pp. 35-36) is not conclusive . In the Baracchi et al. (2012) study, the observations were very short (10 minutes), analyzed along with other behaviors in a principal component analysis framework, and the percent change in aggression was not reported. Observations performed by Fujii (1975, pp. 35-36) were cursory (“antenna and legs of moribund termites bitten off by nestmates”), and the description of the behavior towards infected termites before becoming moribund is not present. However, observations in ant-parasite systems have shown the presence of infected individuals can affect group behavior. For example, Temnothorax nylanderi workers infected with the
Anomotaenia brevis tapeworm receive no aggression, and the presence of such infected individuals has been shown to reduce colony aggression (Beros, Jongepier, Hagemeier, &
Foitzik, 2015).
Our data suggests that, at least for the O. unilateralis s. l. system, infected workers are not attacked. We propose that aggression in other studies of parasites affecting ants and other social insects should be determined based on extensive observations similar to those performed here. Ideally, observations should be double
24 blind and include nocturnal observations. Within this experiment, we were unable to accomplish this ideal scenario due to our interest in establishing fungal growth post mortem and the change in color of the cuticle and paint caused by infrared lights.
However, aggression is not the only metric for understanding how individuals within the nest are treated.
Information on how individuals socially exchange food is also important for informing us on how within-nest dynamics are affected by parasites. Workers of the species T. nylanderi infected with A. brevis have been shown to become social parasites to their own colony by increasing their food begging behaviors and trophallaxis events
(Beros et al., 2015; Scharf et al., 2012). In our study, we expected infected individuals would become a functional social parasite by increasing rate of social food exchange
(similar to A. brevis tapeworms). From previous field studies, we know that infected individuals die close to foraging trails, likely leading to infection of other colony members (Raquel G. Loreto et al., 2014). The colony should act to prevent infected individuals being fed by their uninfected relatives, since the infected individuals will later become a source of disease for the colony.
Our observations suggest that sham individuals engage significantly less in trophallaxis than other treatments groups 6 days post-injection, as measured by the proportion of time spent in trophallaxis while under observation (Figure 2.1 and Table
2.1). However, when we extend our observations to a period of multiple days, we see a significant difference between healthy and infected treatments (Figure 2.2 and Table 2.2).
Furthermore, we can clearly see daily fluctuating patterns in trophallaxis, along with wide
25 individual variation. Our results highlight the importance of not making conclusions based on one day of observation. The duration of trophallaxis is not a measure of food intake, as ants constantly share the contents of their crop with other ants via trophallaxis, as well as exchange other resources (C. Hamilton et al., 2011; Bert Hölldobler & Wilson,
1990). Other researchers have reported an increase in trophallaxis after an exposure to infectious material or their components (Aubert & Richard, 2008; De Souza, Van
Vlaenderen, Moret, & Lenoir, 2008; C. Hamilton et al., 2011; Qiu, Lu, Zalucki, & He,
2016).
While both De Souza et al. (2008) and Hamilton et al. (2011) either isolated or starved their treated individuals, we maintained our ant treatments together with ad libitum food and water. The treatments performed by De Souza et al. (2008) and
Hamilton et al. (2011) could have led to an increase in trophallaxis due to starvation and stress, a behavior which has been demonstrated in previous experiments (Franks, Bryant,
Griffiths, & Hemerik, 1990). Additionally, unlike our use of O. unilateralis s. l., a natural pathogen of ants, Hamilton et al. (2011) used Serratia marcescens, which is not naturally found infecting ants. Although we cannot say the infected ants in our study received less food than controls, what we can suggest based on our data is that C. castaneus workers infected by the specialist parasite O. unilateralis s. l. socially exchange food with others, and the rate of trophallaxis does not increase over time. Suggesting infected individuals do not act as resource sinks within the nest.
Our data indicates infected workers are not recipients of aggression from their uninfected counterparts, nor are the infected excluded from the social exchange of
26 resources. In order to determine if spatial isolation occurred within the nest, we collected point-distance data for every focal individual within the nest every minute during the day light hours (900-1700) of day 6 post-injection. Previous studies performed in the absence of parasites have shown spatial fidelity of ants within the nest (Mersch, Crespi, & Keller,
2013; Quevillon, Hanks, Bansal, & Hughes, 2015; Sendova-Franks & Franks, 1995).
Here we have added the complexity of disease, giving us a different insight into how ant societies interact with their parasitized siblings. Finding a significant difference in the proximity to the entrance between infected and healthy individuals (Figure 2.3 and Table
2.3), in which infected individuals spent more time closer to the entrance; suggests infected individuals are spatially isolated from their nestmates within the nest and place themselves in an ideal location to leave the nest. Our observations on aggression and trophallaxis provide evidence that uninfected nestmates are incapable of detecting O. unilateralis s. l. infection. Taking into account the extensive time infected individuals pent outside of the nest (Figure A.2) and the results of our spatial analysis we can see infected individuals removing themselves from colony dynamics. Such observations have been previously made within other systems (N. Bos et al., 2012; Chapuisat, 2010).
However, within this experiment we were unable to reach the point of manipulation, nor could we recover O. unilateralis s. l. from the cadavers of those injected with fungal tissue.
Although we can observe worker manipulation under controlled conditions following artificial infection, manipulation does not occur in all cases and did not occur here (de Bekker et al., 2015; de Bekker et al., 2014). The correlation in overall mortality
27 rates between this and previous infection studies, as well as the low mortality among sham and healthy treatments (Figure A.1), convinced us that mortality in this study was generally attributable to O. unilateralis s. l. (de Bekker et al., 2015; de Bekker et al.,
2014). Furthermore, in the de Bekker et al. (2014) study, infected ants left after 9 days, as observed here. As such, we are confident that we successfully infected our workers with
O. unilateralis s. l. and the within-nest behaviors observed within this experiment are biologically real. Due to technological challenges, we were unable to follow individuals during the night, limiting us to diurnal observations. However, in future experiments we will address these issues in order to perform detailed observations and to ascertain how manipulation affects the ant society in both nocturnal and diurnal settings.
Future detailed studies are needed to confirm the lack of discrimination toward infected nestmates and to evaluate whether it extends to other coevolved parasites of social hosts, such as trematodes, cestodes, nematodes, strepsipterans, and parasitoids.
None of these pathogens transmit inside the nest. Instead, infection occurs when foragers leave to collect resources and are directly infected or carry back propagules that infect larval ants that then mature into infected adults (for example Schmid-Hempel, 1998).
28 Conclusions
Within this experiment, we tested the response of the colony to the presence of a sibling infected with the specialized fungal parasite O. unilateralis s. l. We found infected individuals are not attacked, socially exchange food throughout the course of fungal development, and spent significantly more time closer to the nest entrance. When considering the significance of these results, we must consider the biology of both the parasite and its host. Only individuals that leave the nest to collect food are infected by the fungus O. unilateralis s. l. and the infection cannot be transmitted within the nest
(Raquel G. Loreto et al., 2014). Foragers only account for a small portion of the colony’s population, roughly 30%, and likely only a small portion of foragers get infected
(Andrew F. G. Bourke & Franks, 1995). This implies that the costs of parasitism at the colony level is low and can be buffered by the colony if mature enough (Hughes, Pierce,
& Boomsma, 2008). A mature ant colony therefore may not necessarily have a strong evolutionary pressure to perform the initially hypothesized behaviors of aggression, exclusion, and isolation. We therefore suggest that such parasites are able to fly under the radar of the colonies’ protective defenses.
29
Figure 2.1- Proportion of time spent in trophallaxis 6 days post-injection. The black whisker plots depict the mean ± standard error for each treatment group. The colored circles depict the proportion of time each focal individual spent in trophallaxis on day 6 post-injection. Each color represents a different treatment group: red represent healthy focal individuals (n=30), green represent infected individuals (n=29), and blue represent sham individuals (n=30). We permuted the treatment of each focal individual 10,000 times and then analyzed the data using a mixed-effect model. Individuals within the sham treatment spent significantly less time in trophallaxis in comparison to the healthy
30 treatment (GLMM: P<0.005). We corroborated our results by analyzing our data without the outlier within the infected treatment and arrived to similar results.
Figure 2.2- Extended trophallaxis observations. Black whisker plots depict the mean ± standard error for each treatment and day. The colored circles represent the proportion of time focal individuals spent in trophallaxis while inside the nest. Each color represents a
31 different treatment group: red represent healthy focal individuals, green represent infected, and blue represent sham. The sample sizes for each day can be found above each group. These observations were performed throughout 7 days of observation (3-9 days post-injection), from 0900-1700. Because trophallaxis is a pair-wise variable, the assumption of independence is violated, so we used a permutation test which randomized individual treatments 10,000 times before comparing treatments. We used three distinct mixed-effect models to analyze our data. Throughout all the models, we find a significant difference between healthy and infected treatments (GLMM: P<0.05). However, within our second model, we found all treatments to be statistically different from one another.
32
Figure 2.3- Mean distance between focal individuals and nest entrance. The black whisker plots depict the mean ± standard error for each treatment group. The grey shapes represent the mean distance between focal individuals and the entrance of the nest on day
6 post-injection. Each shape represents a different treatment group: circles represent healthy focal individuals, triangles represent infected individuals, and squares represent sham individuals. We followed 25 healthy individuals, 26 infected, and 25 sham focal individuals. We analyzed these data by permuting treatments before performing the
33 mixed-effect model for 10,000 iterations. We found a significant difference between healthy and infected treatment groups (GLMM: P<0.05) informing us that infected individuals spent significantly more time closer to the nest entrance. The results can be found on Table 2.3.
P-values For Contrasts Beta-hat values
Models H -I I - S H - S Infected Sham Healthy
Square root(Trophallaxis 0.067 0.330 0.034 0.126 0.143 0.160 proportion)~ 0+ Treatment + (1|Colony.subcolony) +(1|Ant identification)
Table 2.1- Results from mixed-effect model on trophallaxis 6 days post-injection.
Our data violates the assumption of independence. In order to correct for this violation we used a permutation test which permuted the treatment of each focal individual 10,000 times to obtain a dataset for comparison. We used the mixed-effect model denoted within this table to analyze our dataset. Letters differentiate amongst treatments: “H” for the healthy, “I” for the infected, and “S” for the sham treatment. When contrasting the treatments, we found individuals within the sham treatment are significantly different from the healthy treatment (GLMM: P<0.05).
34
P-values for contrasts Beta-hat value
Models H -I I - S H - S Infected Sham Healthy
Square root(Trophallaxis 0.035 0.096 0.061 0.134 0.182 0.152 proportion)~ 0+ Treatment * factor(Day post-injection) + (1|Colony.subcolony) +(1|Ant identification)
Square root(Trophallaxis 0.002 0.024 0.021 0.152 0.164 0.153 proportion)~ 0+ Treatment + factor(Day post-injection) + (1|Colony.subcolony) +(1|Ant identification)
Square root(Trophallaxis 0.035 0.090 0.056 0.149 0.194 0.167 proportion)~ 0+ Treatment(Day post-injection) + (1|Colony.subcolony) +(1|Ant identification)
Table 2.2- Analysis performed on our extended dataset. We permuted the treatment of
each focal individual 10,000 times in order to obtain our results since the assumption of
independence is violated. Using the permuted data we analyzed our results using three
distinct mixed-effect models. The differences amongst the given models are underlined
within the table. Within the contrasts section, we used letters to differentiate amongst
treatments: “H” for healthy, “I” for infected, and “S” for sham treatment. When
contrasting the treatments, we found that across all models individuals within the infected
treatment are significantly different from those in the healthy treatment (GLMM:
P<0.05).
35
P-values for contrasts Beta-hat values
Models H -I I - S H - S Infected Sham Healthy
Distance to Entrance~ 0+ 0.023 0.252 0.294 468.735 521.853 571.535 Treatment + (1|Colony.subcolony)
Table 2.3- Analysis for the proximity of individuals and the entrance of the nest on day 6 post-injection. We randomized the treatments of all focal individual 10,000 times before obtaining results from our mixed effect model (depicted in table). Within the contrasts, section letters differentiate amongst treatments: “H” signifies the healthy, “I” for infected treatment, and “S” for sham treatment. We found infected individuals were significantly closer to the nest entrance in comparison to individuals within the healthy treatment (GLMM: P<0.05).
Chapter 3: Observations on a generalist entomopathogen within a social
insect fortress
Emilia Solá Gracia 1, 2, 3, Benjamin Budas 4, and David P. Hughes 2, 3, 5
1. Ecology, Huck Institutes of Life Sciences, Pennsylvania State University,
University Park, Pennsylvania, 16802
2. Center for Infectious Disease Dynamics, Huck Institutes of Life Sciences,
Pennsylvania State University, University Park, Pennsylvania, 16802
3. Department of Entomology, Pennsylvania State University, University Park,
Pennsylvania, 16802
4. Eberly College of Science, Pennsylvania State University, University Park,
Pennsylvania, 16802
5. Department of Biology, Pennsylvania State University, University Park,
Pennsylvania, 16802
37 Introduction
Eusocial insect colonies are often viewed as a “factory within a fortress” (E. O.
Wilson, 1968) and this structures inhibits pathogen establishment and spread (Cremer,
Armitage, & Schmid-Hempel, 2007; R. B. Rosengaus, Traniello, Chen, Brown, & Karp,
1999; Sadd & Schmid-Hempel, 2006; Traniello, Rosengaus, & Savoie, 2002). Ants, termites, bees and wasps are known for their ability to use multiple strategies which reduce disease spread and establishment within their colonies. These strategies include increased self-grooming after exposure to infectious material (Oi & Pereira, 1993; Theis et al., 2015), increased social grooming after disease exposure (Oi & Pereira, 1993;
Westhus et al., 2014; Zhukovskaya et al., 2013), increased aggression towards individuals exposed to disease (Baracchi et al., 2012; Fujii, 1975; Myles, 2002), self-isolation (N.
Bos et al., 2012; Heinze & Walter, 2010), self-medication (Abbott, 2014; Nick Bos,
Sundstrom, Fuchs, & Freitak, 2015; Erler & Moritz, 2016), and social ‘vaccination’ (R.
B. Rosengaus et al., 1999; Traniello et al., 2002). Such behaviors are considered to be important mechanisms which function to decrease disease exposure and prime the physiological immune system towards a disease threat. Enriching our understanding of these complex behaviors, along with their effects on the individuals within these highly- related colonies, would allow us to study mechanisms of disease defense within complex social dynamics.
The complexities of behaviors performed within eusocial insect colonies have rendered some biological control agents, which have been successful within a laboratory setting, ineffective in the field. The use of entomopathogenic fungi, such as Beauveria bassiana and Metarhizium anisopliae, became the emphasis of more than 100
38 experiments due to the promise these entomopathogens showed as biological control agents in a laboratory setting (reviewed in: Thomas Chouvenc, Su, & Grace, 2011; R.G.
Loreto & Hughes, 2016). However, when used in the field B. bassiana and M. anisopliae showed limited success in eradicating eusocial insect colonies (reviewed in: Thomas
Chouvenc et al., 2011; R.G. Loreto & Hughes, 2016). Such failures mean these fungi are impractical biological control agents, yet their use in the laboratory setting to study host- parasite ecology continues (R.G. Loreto & Hughes, 2016). Both fungal pathogens have become tools for understanding disease management within eusocial insect colonies.
Social structures and worker dynamics within ant species have been tested using
B. bassiana and M. anisopliae, but the typically employed methodology lacked biological relevance (reviewed in R.G. Loreto & Hughes, 2016). Loreto and Hughes (2016) found reports of wild ants infected with M. anisopliae and B. bassiana in the field to have significantly smaller number of samples than reports referring to wild ants infected with a coevolved fungal parasite, Ophiocordyceps. In nature ant workers are not readily infected with generalist entomopathogens like B. bassiana and M. anisopliae (R.G. Loreto &
Hughes, 2016, Figure 3). However, evolutionary processes are invoked within the literature which uses B. bassiana and M. anisopliae. Furthermore ants are mostly placed within sterile environments like petri dishes, and short observation bouts have become commonplace (for example: N. Bos et al., 2012; Heinze & Walter, 2010; Konrad et al.,
2015; Konrad et al., 2012; Masri & Cremer, 2014; Theis et al., 2015; Simon Tragust,
Mitteregger, et al., 2013; Simon Tragust, Ugelvig, et al., 2013; Ugelvig & Cremer, 2007;
Westhus et al., 2014). This is not necessary as Loreto and Hughes (2016) showed. It is understandable that simplification allows for more control over the variables affecting
39 behavior. However, it is unrealistic to infer evolutionary processes when important biological factors have been removed. The use of short observational bouts, which are generally chosen for convenience, can inhibit our understanding of important overarching behavioral processes which in turn may affect our understanding of complex systems.
Hence, it is of value to perform an experiment where ant colonies are in a semi-natural environment and observed for longer periods of time, increasing the biological relevance within the results.
Here we set out to use a semi-natural nest design together with long observational bouts to determine the effects that workers infected with B. bassiana have on within- colony interactions. This experiment is a positive control for the observations performed on nests containing workers infected with the specialized entomopathogenic fungus
Ophiocordyceps unilateralis sensu lato (Chapter 2). Allowing us to determine if our previous observations could was due to sickness behaviors or the behavior manipulating parasite. Within this experiment, we are interested in determining if infected individuals change: 1) their rate and number of social food exchanges, 2) the amount of time spent within the nest, and 3) the amount of time spent in nest vigilance behavior. We hypothesized that infected workers will have higher rates of social food exchange
(trophallaxis) to compensate for the physiological costs of infection. We expected infected workers would spend less time within the nest, a behavior observed in previous work (Chapter 2, N. Bos et al., 2012; Heinze & Walter, 2010). Furthermore, we expected workers to reduce the amount of time spent in nest vigilance behaviors due to their infection.
40 Materials and Methods
Ant colony collection and maintenance
We collected two colonies of the soil dwelling carpenter ant, Camponotus castaneus, colonies (Cast1 and Cast13) from Due West, South Carolina in July of 2013.
The methods of colony collection are described in Chapter 2. Cast1 and Cast13 contained approximately 400 workers, and no queen was collected (it is very difficult to collect the queen). After collection, we provided each colony with 10% sugar water and water, changing them once a week.
Beauveria bassiana infection technique
We created Beauveria bassiana (strain BI 93-825) hyphal cultures by placing a conidia solution between two IsoporeTM membrane filters (filter type: 0.1 μm, VCTP).
After three days, which is the required time for hyphal development, we followed similar infection methods as previously described (Chapter 2, de Bekker et al., 2015; de Bekker et al., 2014). Briefly, we cut a 1 cm2 section of the fungal tissue, which we then placed into a sterile 2 mL tube with two 0.63 cm metal balls (Wheels Manufacturing, Inc.) and
200 µL Grace’s medium (Sigma) that we freshly supplemented with 10% Fetal Bovine
Serum (FBS, PAA Laboratories, Inc.). We placed our sample in the TissueLyser II
(Qiagen) at room temperature for 60 seconds at 30 cycles per second in order to create small segments of hyphae (mean concentration of 3.9 × 107 ± 1.1 × 107 hyphae per mL) for infection. We injected a subset of workers with 1 µL of the hyphal solution with a laser-pulled 10 µL micropipette (Drummond) and aspirator tube (Drummond) into the thorax underneath the front pair of legs. We injected another subset of workers with 1µL
41 of medium (Grace’s insect media freshly supplemented with 10% FBS) without hyphae, serving as a sham treatment for injection.
Treatments and individual identification
We randomly selected 30 workers from two stock colonies (Cast1 and Cast13) to create experimental groups for observation. We marked each worker with a unique paint pattern using Edding® number 751 paint markers as previously described (Chapter 2).
After three days of baseline observations, we randomly placed workers into three treatment groups: Healthy (untreated workers, n=10), Sham (workers injected with 1 µL of media, n=10), and Infected (workers injected with1 µL of media and B. bassiana hyphal tissue, n=10). Observers only knew individual identity and date of death. They had no knowledge of individual treatment, allowing us to reduce observational bias during data collection.
Mortality observations
In order to determine the effectiveness of our injections, we monitored mortality once a day. We collected and surface sterilized the cadavers collected throughout the experiment by placing the cadavers within 70% ethanol for 20 seconds, which was then allowed to dry. After surface sterilization, we placed each cadaver in a sterile petri dish
(100 mm × 15 mm dimension) with a moist filter paper (Whatman 541, 70 mm diameter) and incubated the cadavers at 28˚C. In order to determine if these individuals died due to a B. bassiana infection, we monitored the cadavers for fungal growth once a day.
Behavioral observations
42 Within this experiment, we were interested in determining if workers infected with B. bassiana change in: 1) how often and at what rate infected ants consume food from their siblings via mouth to mouth exchange of liquid food (i.e. trophallaxis), 2) the amount of time spent within the nest, and 3) performance of an altruistic behavior (i.e. nest vigilance). As with previous work, we made within-nest observations by using a
GoPro camera (Hero 2 fitted with both infrared (IR) lens and 4.6 mm macro-lens) held above the artificial nest (nest dimension of 21.79 cm2). The apparatus used to hold the camera was covered with red Mylar, which kept the workers in darkness as ants cannot perceive red light. We connected the semi-natural nest to a foraging arena (dimension of
1,224 cm3) which had a thin layer of plaster on the floor to increase the humidity within the foraging arena. The foraging arena contained feeding tubes with 10% sugar water and water both given at libitum. Colonies had a 12:12 day-night cycle. Daylight hours began at 0600 and ended at 1800 (EST). In order to make nocturnal observations within the nest, we used infrared lights. Recording was done continuously for a total of 8 days (4 days before and 4 days after injection), except for when the memory cards were changed.
These card changes occurred twice a day, taking approximately 5 to 15 seconds each time.
Trophallaxis
Adult worker ants cannot consume solid food and get energy from the exchange of liquid food from mouth to mouth. This is mostly carbohydrate rich but can also be proteins derived from larval stage siblings who can eat solid food and convert that to liquid (Bert Hölldobler & Wilson, 1990, pp. 164-168). We observed workers from the
Cast13 colony three days before, and three days post injection. Data collection occurred
43 for 30 minutes every hour on the hour (0000-0030, 01000-0130, and so forth, until 2300-
2330 each day). We observed a total of 72 hours over the course of 6 days. Trophallaxis began when the labrum and labium (i.e. maxillolabial complex of the ant’s mouth) were distended between two individuals, and ended when these mouth parts separated. We monitored the durations and number of trophallaxis events for all workers within the nest at any given time point within out observation bout. We accomplished this by individually monitoring each worker within the nest meaning each bout was watched multiple times.
Within our experimental arena, workers were able to enter and exit the nest freely.
We therefore needed to express trophallaxis as a proportion in which time spent in trophallaxis is equivalent to total time spent in trophallaxis divided by the total time spent within the observation arena during the observational bouts. These calculations allowed us to control for variation in the amount of time workers spent within the nest. To analyze these data, we created a mixed-effect model in which treatment and day of observation are treated as fixed effects, while ant identification is a random effect.
One issue is independence of data, since trophallaxis requires two workers interacting. As such the data are not independent. Another issue is sample size; we only used data collected from one colony. This work is presented with these caveats in mind in the expectation that they inform other studies. In order to evaluate any significant effects we have within the model, we used the least-squares means post hoc analysis and compared between treatments. We also analyzed the proportion of time spent in trophallaxis on a day-by-day basis, where we used Kruskal-Wallis analysis, coupled with a Bonferroni correction and Nemenyi post hoc analysis. We further explored the effects
44 infection could have on trophallaxis by determining if the number of events changed over the course of fungal development. We used a mixed-effect model, which treated day of observation and treatments as fixed effects, while we treated ant identification as a random effect. We used the least-square means post hoc analysis to compare amongst treatments. Furthermore, we analyzed each day separately using Kruskal-Wallis analysis coupled with a Bonferroni correction and a Nemenyi post hoc analysis.
Time spent within the nest
Past work has suggested sick individuals spend more time outside of the nest which was interpreted as a form of self-isolation (N. Bos et al., 2012; Heinze & Walter,
2010). Within this experiment, we tested the change in time workers spent within the nest, comparing treatments and days after injection. We collected data in 30 minutes intervals every hour on the hour each day (0000-2300) from colonies Cast1 and Cast13.
In order to analyze our data, we used a simple mixed-effect model, treating day of observation, colony, and treatment as fixed effects, while individual identification was a random effect. We also analyzed each day individually by using a Kruskal-Wallis analysis coupled with a Bonferroni correction and Nemenyi post hoc analysis.
Nest vigilance
A nest worker places herself in front of the entrance to detect nest intruders and possible threats to the colony. We wanted to determine if C. castaneus workers infected with B. bassiana stopped performing nest vigilance behaviors. To do, this we recorded the identity and times spent in front of the internal nest entrance 3 days before and after injection for both Cast1 and Cast13 colonies (see classification for nest vigilance behaviors in Figure B.1). We collected these data in 30-minute bouts every hour on the
45 hour, providing a total of 12 hours of observation/day. Vigilance behavior began when individuals placed themselves in front of the nest entrance located within the nest and placed their antennae parallel to the entrance walls (Figure B.1). In order to reduce observational bias, one observer (BB) recorded these data. We used a mixed-effect model to analyze our data. We treated day of observation, colony identification, and treatment as fixed effects, and we treated individual identification as a random effect. Furthermore, we used Kruskal-Wallis analysis and a Nemenyi post hoc analysis in order to make isolated daily analyses of our data. We used a Bonferroni correction to account for multiple comparisons.
Data analysis
All individuals within each colony were labeled as healthy over the course of the three days before injection, while after injection, each individual was placed into one of three treatment categories (healthy, sham, or infected). We used the R program, including the packages lmer and lsmeans, to analyze our data sets (Bates, Maechler, Bolker, &
Walker, 2015; Lenth, 2016). We used the kruskal.test function to perform out Kruskal-
Wallis analysis, while we used the PMCMR package to perform our Nemeyi post hoc analysis (Pohlert, 2016).
Results
Beauveria bassiana infection
We were able to properly infect Camponotus castaneus workers using Beauveria bassiana hyphal tissue. We injected 20 individuals with B. bassiana within our experiment. Four days after injection, a total of 19 individuals died and 18 of these
46 cadavers had the regular pattern of B. bassiana growth. We classified proper B. bassiana development when the cadaver showed clear signs of external hyphal growth with proper conidia production (see image in Figure B.2). Furthermore, workers within the infected treatment died at a higher rate than those within the healthy and sham treatments
(Kaplan-Maier log rank: P= 7.79e-08; Figure B.3).
Trophallaxis
We found infected individuals to spend significantly less time in trophallaxis on the day after treatment, in comparison to individuals within other treatment groups
(GLMM: P=0.03). We further analyzed these data by comparing amongst treatments each day using a Kruskal-Wallis analysis, corrected our p-values with a Bonferroni correction
(for significance P<0.017), and used the Nemenyi post-hoc analysis (Pohlert, 2016). We found a significant difference on day 1 of observation between infected and healthy treatments (Kruskal-Wallis: P=0.011; Figure 3.1). In order to determine if the number of trophallaxis events was also affected, we performed analyses on the number of events performed by each individual while under observation.
The number of trophallaxis events workers perform over the course of fungal development was not statistically significant amongst treatments when using the mixed- effect model (GLMM: P>0.05). However, when analyzing each day separately using a
Kruskal-Wallis test, we found infected individuals to be significantly different from individuals within the healthy treatment on days 1 and 3 post injection (Kruskal-Wallis:
P<0.017). Suggesting individuals infected with B. bassiana reduce the number of times they perform social food exchange right after infection and the day before death (Figure
47 3.2). However, it is critical to stress that the data only comes from one colony and as such results are not robust.
Time spent within the nest
Infected individuals have been observed to spend significantly more time outside of the nest (N. Bos et al., 2012; Heinze & Walter, 2010). Within our mixed-effect model, we found no significant differences amongst treatments or days within our analysis
(GLMM: P>0.05). However, when we analyzed each day separately with a Kruskal-
Wallis analysis, we found a significant difference on day 3, in which infected individuals are significantly different from sham and healthy treatments (Kruskal-Wallis: P<0.017).
The high variation in the amount of time spent within the nest affected our ability to detect any significant results using a mixed-effect model. However, the analysis performed on a day-by-day basis suggests individuals infected with B. bassiana spend more time outside of the nest the day before death (Figure 3.3).
Nest vigilance
We found no significant differences amongst treatments or days within our mixed-effect model analysis (GLM: P>0.05). Furthermore, the Kruskal-Wallis analysis on each individual day and also found no significant differences (Kruskal-Wallis:
P>0.017). It is important to note the stochastic nature of our data and small sample sizes inhibited our ability to detect any statistical differences amongst treatments (Figure 3.4).
Discussion
We set out to test the effects a generalist entomopathogen has on individuals within the nest by collecting data on: 1) rate and number of social food exchange events
48 between workers, 2) time individuals spend within the nest, and 3) the amount of time individuals spent in nest vigilance behaviors. We hypothesized infected Camponotus castaneus workers would increase the amount of time spent in social food exchanges due to the physiological pressure Beauveria bassiana poses on the host. Furthermore, we hypothesized infected workers would increase their time spent outside of the nest. We also expected infected workers would decrease their time spent performing nest vigilance behaviors, since such behaviors are not in the interest of the growing parasite as nest vigilance is a costly behavior that could expose the host to mortality factors. In order to test these hypotheses, we recorded within-nest behaviors for C. castaneus workers three days before and after injecting a subset with B. bassiana.
We found individuals infected with B. bassiana had small alterations in their behaviors. We recognize these results come from small sample sizes and therefore must be viewed with caution. However, our detailed behavioral observations are biologically meaningful and can inform future work. We found infected individuals spent on average
0.015% of their time in trophallaxis the day after we treated them, while workers within the healthy treatment spent 0.31% of their time in trophallaxis on the same day, suggesting infection affects the percent of time spent in trophallaxis for a small period of time (Figure 3.1). When taking into account the number of trophallaxis events, we found infected individuals performed less trophallaxis on days 1 and 3 post injections.
Individuals within the healthy treatment had a wide variation in the number of trophallaxis events they performed (Figure 3.2). Furthermore, we found infected individuals spent less time within the nest the day before death. However, in part due to the stochastic performance of vigilance behaviors, we found no significant difference
49 amongst treatments. Our results suggest B. bassiana infection had a small effect on social food exchange in the ant society we tested. This is in line with previous observations made in other systems.
Observations made within other endoparasite systems have seen changes in the performance of trophallaxis behavior. For example, the tapeworm Anomotaenia brevis infecting Temnothorax nylanderi workers has been shown to increase the rate of trophallaxis and begging behavior of those infected (Scharf et al., 2012). Furthermore, these infected individuals have been shown to affect behavior at the colony level (Beros et al., 2015). When Camponotus pennsylvanicus workers are infected with the gram- negative bacteria Serratia marcescens, they increase the amount of trophallaxis they perform (C. Hamilton et al., 2011). Hamilton et al (2011) demonstrated that workers exchanged cathepsin D which has antibiotic properties, suggesting trophallaxis has an important role in colony defense against disease.
It is possible that a fungal infection could have different effects on worker and colony dynamics. Qiu et al (2016) used Metarhizium anisopliae, a generalist fungal pathogen, to show that infected individuals change their feeding behaviors, finding infected workers only change their behavior three days after inoculation. On the third day after inoculation workers are closer to dying; the change in trophallaxis would not have an immunizing effect but could be fueling the development of M. anisopliae. Our previous study found workers infected with O. unilateralis s. l., a specialized fungal entomopathogen, engage in trophallaxis events throughout the course of fungal development, in which infected workers were significantly different from the healthy treatment (Chapter 2). However, our results had no continuous pattern due to individual
50 and daily variations within our data set. Within this experiment, we find workers infected with B. bassiana continuously perform trophallaxis with their nestmates over the course of fungal development. However, when analyzing our data using a mixed-effect model, we found infected individuals had a significantly lower rate of trophallaxis the day after injection. We further corroborated these results by using a Kruskal-Wallis analysis, which also showed a significant difference between infected and healthy treatments. Our findings suggest a strong immunological response towards the injection of hyphal material. In order to further explore the effects a B. bassiana infection has on individual behavior, we determined how much time workers spent within the nest.
Workers infected with generalist entomopathogenic fungi have been shown to spend significantly more time outside of the nest (N. Bos et al., 2012; Heinze & Walter,
2010). Authors suggest this behavior reduces disease exposure and spread within the nest, thus maintaining the colony health. However, B. bassiana and M. anisopliae infect via contact with conidia, which are produced after host death. One possibility is that erratic walking behavior increases the chances of the infected ant dying in a place that is suitable for fungal development , similar to a behavior seen in crickets infected with hairworms where infected wander before the parasite is ready to mature (Sanchez et al., 2008).
Within this experiment, we only performed observations on within-nest behaviors, but future work should challenge this argument by following infected workers outside of the nest. We found no significant difference amongst treatments in the amount of time individuals spent within the nest. However, when we analyzed each day separately, we found infected individuals to be inside the nest significantly less than other treatments.
The amount of time workers spent within the nest was highly variable. In the future, we
51 suggest these observations be done on multiple colonies with multiple replicates of the same colonies to decrease the effect of colony and individual variation.
In order to determine if the colony is being attacked, one worker must place herself at the entrance to examine the individuals entering the nest and detect changes in the external environment. Within this experiment, we made observations on the number of individuals performing vigilance behaviors, along with the duration of each vigilance bout. Overall, the performance of vigilance behaviors was highly variable both in the individuals that performed it and the amount of time spent performing vigilance behaviors. During the development of the fungal endoparasite, infected workers continued to perform nest vigilance behaviors. Our observations suggest B. bassiana does not have an effect on the performance of vigilance behaviors. However, within future experiments, we would suggest agitating the nest to simulate a nest disturbance, and then compare the identity of those individuals that attacked the object versus those that placed themselves at the nest entrance, allowing us to determine if individuals performing vigilance behaviors are likely to perform nest guarding behaviors.
Conclusion
Within our experiment, we found C. castaneus workers infected with B. bassiana do not have large changes in behavior. Infected workers receive socially shared food, spend more time outside of the nest the day before death, and sporadically perform vigilance behaviors. We suggest that C. castaneus colonies are capable of buffering the negative effects the presence of infected individuals might have on colony productivity.
The results obtained from our observations suggest B. bassiana has an effect on
52 individual behavior (i.e. less time within the nest and smaller number of trophallaxis events), but does not affect the interactions amongst workers (e.g. infected workers continuously exchanged food socially). Our experiment should be repeated with more colonies to corroborate our results, and further our understanding of social dynamics inside ant nests. Similar to our previous findings using a specialized entomopathogen
(Chapter 2), we found workers are cannot detect the entomopathogen developing within infected nestmates. Allowing B. bassiana and O. unilateralis s. l. to take advantage of the colony as soon as it infects a colony member.
53
Figure 3.1- Proportion of time workers spent in trophallaxis while inside the nest.
Trophallaxis data was collected from three days before injection (negative x-axis values) and three days after injection (positive x-axis values) from one colony (Cast13). Black whisker plots depict the mean ± standard error for each treatment and day. The colored circles represent the proportion of time individuals within each treatment spent in trophallaxis while inside the nest. Each color represents a different treatment group: red represents healthy, green represents infected, and blue represents sham. The sample sizes for each day can be found above each group. We used letters above or below the sample size to depict significant differences between treatments, on the days we found statistical
54 differences amongst our treatments. Within these observations we found infected individuals spent significantly less time in trophallaxis in comparison to individuals within the healthy treatment (GLMM: P=0.03; Kruskal-Wallis: P=0.011).
55
Figure 3.2- Number of trophallaxis events occurring within the nest. The days before injection treatment have negative values, while the days after the injection treatments occurred have positive values. We collected these data from one genetic colony, Cast13.
Black whisker plots depict the mean ± standard error for each treatment and day. The colored circles represent the number of trophallaxis events performed by each individual within their treatment group. Each color represents a different treatment group: red represents healthy, green represents infected, and blue represents sham. The sample sizes for each day can be found above each group. We used letters above or below the sample
56 size to depict significant differences between treatments, on the days we found statistical differences amongst our treatments. We found individuals within the infected treatment performed significantly less trophallaxis events, in comparison to individuals within the healthy treatment on days 1 and 3 post injections (Kruskal-Wallis: P<0.017).
57
Figure 3.3- Percent of the time individuals spent within the nest during our observational bouts. Negative days of observation represent the days before injection, while positive values represent the days after injection treatments occurred. We collected these data from two colonies, Cast1 and Cast13. Sample sizes of each treatment can be found about the column of data, while the black whisker plots depict the mean ± standard error for each day and treatment. Each color represents a different treatment group: red represents healthy, green represents infected, and blue represents sham. Significant differences between treatments are depicted using letters above the sample sizes, on days we found a significant difference amongst treatments. Within these data we found
58 infected individuals spent significantly less time within the nest the day before death
(Kruskal-Wallis: P<0.017).
Figure 3.4- Proportion of time individuals performed vigilance behaviors while inside the nest. We collected these data from two colonies, Cast1 and Cast13. Each color represents a different treatment group: red represents healthy, green represents infected, and blue represents sham treatment. Sample sizes of each treatment can be found about the column of data, while the black whisker plots depict the mean ± standard
59 error for each day. We found no significant differences amongst treatments (GLMM:
P>0.05; Kruskal-Wallis: P>0.017).
Chapter 4: Dynamics surrounding the dead: Do ant colonies recognize
disease threats?
Emilia Solá Gracia 1, 2, 3, Raquel Loreto 2, 3, Ryan Bringenberg 2, Ephraim Hanks 2, 5 ,
and David P. Hughes 2, 3, 4
1. Ecology, Huck Institutes of Life Sciences, Pennsylvania State University,
University Park, Pennsylvania, 16802
2. Center for Infectious Disease Dynamics, Huck Institutes of Life Sciences,
Pennsylvania State University, University Park, Pennsylvania, 16802
3. Department of Entomology, Pennsylvania State University, University Park,
Pennsylvania, 16802
4. Department of Biology, Pennsylvania State University, University Park, PA 16802
5. Department of Statistics, Pennsylvania State University, University Park,
Pennsylvania, 16802
61
Introduction
Eusocial insect colonies live within microclimates that are considered ideal for infectious disease transmission (Schmid-Hempel, 1998). Therefore, the appropriate management of potential sources of diseases is critical within wasp, bee, termite, and ant colonies, since workers within the nest are highly related and constantly in close proximity to one another. A disease which enters the colony could spread inside the nest.
In order for a colony to maintain health it must develop strict defense mechanisms against parasites. Cadavers within the nest can harbor the development of bacterial decomposers bacteria or fungal entomopathogens (Lopez-Riquelme & Fanjul-Moles, 2013).
Therefore the cadavers within the nest must be properly managed to reduce disease outbreaks within the nest. The behaviors used by eusocial insects to manage cadavers within the nest have been well documented and include: cadaver removal
(Julian & Cahan, 1999; Oi & Pereira, 1993; Trumbo & Robinson, 1997; Visscher, 1983), burial (T. Chouvenc, Robert, Semon, & Bordereau, 2012; Myles, 2002; Neoh, Yeap,
Tsunoda, Yoshimura, & Lee, 2012; Sun, Haynes, & Zhou, 2013), placement of the cadaver within refuse pile (Bot et al., 2001), cadaver isolation (Neoh et al., 2012; Oi &
Pereira, 1993; Rebeca B. Rosengaus, Traniello, & Bulmer, 2011), and cannibalism
(Thomas Chouvenc & Su, 2012; Oi & Pereira, 1993). Colony lifestyle and ecology play an important role in the type of behaviors used to manage cadavers within the nest.
Wasp, bees, termites and ants have different life histories and ecologies which impact the approach workers employ to manage cadavers. For example wasps and bees typically live within suspended nests and can spatially isolate themselves from cadavers
62
using flight. Paper wasps (Polistes spp.) which live in open nests allow waste to fall out of the nest (Turillazzi, 2012). While Vespula rufa, which live in an enclosed nest, allow the natural accumulation of waste at the bottom of the nest (Ross & Matthews, 1991).
Such behaviors quickly removes sources of disease within the nest with little effort and do not require a lot of contact with hazardous material. Unlike wasps, honey bees (Apis mellifera) remove cadavers from the nest in a behavior referred to as necrophoresis.
Honey bee workers pick up a dead nestmate and fly it 10 to 100 meters away from the nest before allowing it to fall (Visscher, 1983). Such management requires extensive effort and contact with potentially hazardous material. However, the terrestrial lifestyle of termites and ants place them in constant contact with sources of disease (Boomsma et al.,
2005).
The inability of termites and ants to fly inhibits them from being able to spatially separate themselves from hazardous material without substantial contact with the deceased. The energy and time spent managing cadavers could increase the chances of becoming infected. Termites perform two typical behaviors in order to manage cadavers within the nest, which include cannibalism (Thomas Chouvenc & Su, 2012) and burial or avoidance (T. Chouvenc et al., 2012). While the behavioral repertoire used by ants is extensive, similar to wasp and bees the behaviors used to manage cadavers is dependent on species ecology and life-history. According to Bot et al. (2001) leaf-cutter ants (Atta cephalotes) collect moribund or dead nestmates and place them at the edge of the refuse pile. The workers which manage the refuse pile collect the deceased and mix their bodies into the colony’s waste (Bot et al., 2001). Such behaviors could be efficient in reducing
63
the exposure to potential disease. While Myrmica rubra remove cadavers and place them in remote at distance away from the nest (Diez et al., 2012). The removal of cadavers from ant colonies increases larval survival and acts as a prophylactic measure for the colony (Diez et al., 2014).
Workers are capable of recognizing and managing some sources of infectious material. Brood and nestmates powdered with conidia (asexual spores) are treated differently than individuals powdered with inert material (Reber et al., 2011; Simon
Tragust, Mitteregger, et al., 2013). The ability for nestmates to remove infectious agents off the cuticle or surface of other increases nestmate survival after exposure to deadly conidia (Theis et al., 2015). Although the simple ability to remove infectious conidia off the cuticle of other nestmates increases survival, the use of the metapleural gland (i.e. complex glandular structure which produces substances with antibiotic properties) and acidopore (i.e. poison gland) have also shown to be important defense mechanisms against disease.
Both the metapleural gland and acidopore produce substances with antibiotic properties which significantly reduce microorganism viability post exposure. The metapleural gland is a synapomorphic trait in ants; substances produced by this organ are spread over the body as a chemical defense against parasites (B. Hölldobler & Engel-
Siegel, 1984; Yek & Mueller, 2011). Leaf-cutting ants depend on this gland to maintain the health of their fungal garden, by killing fungal pathogens which could parasitize their food source (Fernandez-Marin et al., 2006; Kermarrec, Febvay, & Decharme, 1986;
Richard & Errard, 2009). However, ants within the genera Camponotus, Dendromyrmex,
64
Oecophylla, and Polyrhachis have secondarily lost the metapleural gland (Bert
Hölldobler & Wilson, 1990, p. 30). Ant species within these genera use formic acid produced within their poison gland apparatus to protect themselves against parasites (Bert
Hölldobler & Wilson, 1990, p. 229).
The chemical cocktail produced within the poison gland is used under multiple circumstances. Although the secretions produced within the poison gland function as an alarm pheromone and predator deterrent, the antibiotic properties of these secretions have not gone unnoticed (Attygalle & Morgan, 1984; Blum et al., 1958; Storey et al., 1991).
The ability for these substances to inhibit parasite development has been tested in vitro
(Blum et al., 1958; Li et al., 2012; Storey et al., 1991) and in vivo (Simon Tragust,
Mitteregger, et al., 2013). Tragust et al. (2013) demonstrated the ability for Lasius neglectus workers to reduce the infectivity of Metarhizium anisopliae with the use of chemicals produced within the poison gland. However, the ability for ants to recognize and protect themselves from infectious cadavers has not been empirically tested.
Within this experiment we set out to test if workers belonging to two distinct species of carpenter ant, Camponotus castaneus and Camponotus pennsylvanicus, are capable of recognizing and differentiating between hazardous and non-hazardous cadavers. We exposed a treatment group to one of the three cadaver treatments. Our first cadaver treatment is a nest worker we killed by freezing and posed no disease threat. Our second treatment is a cadaver which succumbed to a Beauveria bassiana infection, but poses as a latent disease threat since the B. bassiana development is not mature and unlikely infect workers. However, our third treatment poses an immediate disease threat,
65
since the external hyphal development along with the mature conidia production of B. bassiana can easily infect the workers which handle the cadaver. We hypothesize these cadavers would receive the highest levels of aggression, be removed the fastest, and received the smallest proportion of worker contacts. Workers will evade exposure to harmful conidia by not contacting the cadaver or being close to the source of disease.
However, we do not expect workers to recognize our second cadaver treatment as a disease threat. It is possible that workers are unable to detect the internal hyphal development occurring within these cadavers. In previous worker we saw no aggressive behavior towards workers with internal fungal development (Chapter 2 and Chapter 3).
Materials and Methods
Camponotus castaneus collection
We collected eight C. castaneus colonies in Due West, South Carolina by digging the soil near nest entrances. Since we are unable to know wild colony architecture and in some cases their location is difficult to dig, we are unable to collect queen-right colonies frequently, making most colonies void of a queen. The colonies’ demography and date of collection are as follows: KFM14-13, collected July 2014, consisted of brood and approximately 300 workers; KFM14-15, collected July 2014, consisted of brood and approximately 250 workers; KFM14-22, collected July 2014, consisted of queen, brood, and approximately 350 workers; KFM15-8, collected June 2015, consisted of brood and approximately 450 workers; KFM15-9, collected June 2015, consisted of brood and approximately 300 workers; KFM15-11, collected June 2015, consisted of brood and
66
approximately 300 workers; KFM15-12, collected June 2015, consisted of queen, brood, and approximately 450 workers; KFM15-15, collected June 2015, consisted of brood and approximately 450 workers.
Camponotus pennsylvanicus collection
We collected eight C. pennsylvanicus colonies in State College, Pennsylvania by cutting into fallen trees, decaying logs, and other woody debris. Being a polydomous species, the colony uses multiple nest locations, making the collection of a complete colony difficult at times, and therefore not all colonies are queen-right. The demography and date of collection for each colony are as follows: SGL176-SU16/1, collected August
2016, consisted of approximately 500 workers; KG5, collected June 2016, consisted of brood and approximately 1,300 workers; SGL176-G2, collected June 2016, consisted of brood and approximately 1,500 workers; PWN1-FA16, collected October 2016, consisted approximately 2,000 workers; SGL16-G4, collected June 2016, consisted of brood and approximately 800 workers; SGL176-F16/2, collected September 2016, consisted of approximately 1,000 workers; PWN2-FA16, collected October 2016, consisted approximately 600 workers; PWN6-FA16, collected October 2016, consisted of approximately 4,500 workers.
Individual identification and treatment group habituation
In order to follow multiple individuals over an extended period of time we printed individualized labels on polyester paper. We labeled each worker by securing them onto a soft foam board with monofilament line (South Bend brand, item: M144). During this period we placed a small drop of cyanoacrylate (Super-Gold+TM odorless) onto the
67
worker’s gaster and attached their unique label. Each treatment group consisted of 20; however at first we labeled a total of 25 to account for mortality and nest rejection. We labeled 6 treatment groups for each of the 16 genetic colonies used within this experiment
(8 genetic colonies per species). After labeling, we placed the workers within a cage
(length 29.99 cm × width 19.36 cm × depth 9.52 cm; Pioneer plastics, item: 195C) and provided a housing tube, water and 10% sugar water. We monitored mortality for 24 hours, if more than 5 workers died within this period we would repeat the labeling procedure for a new group. If mortality was low we would introduce each group to their semi-natural nest.
We placed surviving workers into their assigned semi-natural nest by using soft forceps, we maintained the workers within the closed nest for 1 hour. This period allowed the workers to calm down after being handled and begin to accept the semi-natural nest as theirs. After this hour long period we allowed workers to move freely between their cage and semi-natural nest. Treatment groups containing more than 20 individuals had excess individuals removed after 24 hours. During this removal we focused on individuals which rejected the nest, allowing us to increase the number of workers which accepted the nest. However, if more than 5 workers rejected the semi-natural nest we replaced the group. The week long habituation period before the exposure to the cadavers began the day after we culled our treatment groups. Over our habituation period we monitored for mortality and changed their 10% sugar water and water.
Semi-natural nest construction and design
68
The semi-natural nest used for our observations had two chambers (plaster: length
3.61±0.2 cm × width 4.96±0.2 cm; wood: length 3.78 cm × width 5.05 cm) and was connected by using a one corridor (plaster: length 2.39±0.2 cm × width 1.12±0.2 cm; wood: length 3.45 cm × width 1.27 cm). However, the nest ecology and behaviors of workers are significantly different from one another. Requiring different nest material and design for each species. The humidity requirements of C. castaneus workers required a nest which would retain water for an extended period of time. Therefore we used plaster to create the nests used for this species (length 4.89± 0.2 cm × width 9.33± 0.2 cm × depth 1.75± 0.2 cm). We used 0.1-10 μL pipette tip box lid, along with a foam mold to imprint our nest design into the plaster to create the nests used for our observations (nest schematic can be found in Figure C.1). We maintained the ants within the nest by using acrylic Plexiglas and red Mylar to keep workers in full darkness. We placed the aperture used for cadaver introduction above the corridor connecting the chambers, which we then closed by using clear transparency film to stop the workers from escaping. However, we made a significant amount of changes for C. pennsylvanicus nests in order to accommodate nesting differences and higher levels of aggression.
To build the semi-natural nests for C. pennsylvanicus we used a laser to cut sections of medium-density fiberboard, obtaining nests with equal dimension (length 11.0 cm ×width 5.7 cm × depth 0.6 cm) and structure (nest schematic can be found in Figure
C.2). We kept workers contained within the nest with a fitted acrylic Plexiglas lid and in full darkness by using red Mylar. Due to higher level of aggression of C. pennsylvanicus
69
we placed the aperture for cadaver introduction on one side of the corridor connecting the chambers. We kept this aperture closed with a locking mechanism.
Cadaver preparation
Two of our cadaver treatments (internal mycosis and external mycosis) came from individuals that had succumbed to a Beauveria bassiana infection. In order to obtain these cadavers we used an infection cohort system, in which we selected two groups of 5 workers from each stock colony to infect with B. bassiana. We first created a highly concentrated conidia suspension with saline-Tween (0.05%). We placed 500 μL of the suspension onto two Whatman 541 filter papers (70 mm diameter) within a petri dish
(100 mm × 15 mm dimension). After the suspension had been absorbed into the filter paper we placed a cohort within the petri dish and exposed them to the conidia for a 24 hour period. After each 24 hour exposure period the cohort was released into a cage with a housing tube, 10% sugar water and water ad libitum. We exposed the first cohort the same day their colony began habituation, while the second cohort was exposed three days after habituation had started. We monitored for mortality twice a day and collected the cadavers. We surface sterilized each cadaver with 70% ethanol for 20 seconds and placing them into a sterile petri dish (60 mm × 15 mm dimension) containing filter paper saturated with 250 L of sterile water. We incubated all cadavers at 28˚C, and daily monitored for B. bassiana growth.
We monitored the fungal development by placing the cadavers under a dissecting microscope. To obtain the treatment which represented a latent infection threat to the colony (termed internal mycosis) we needed cadavers which only had internal hyphal
70
growth. The internal mycosis cadavers had no signs of external hyphal growth and took approximately 2-3 days to obtain. While the cadavers used for our treatment representing an immediate threat (termed external mycosis) needed to have external hyphal growth along with mature conidia production (confirmed conidia production shown in Figure
C.3). These cadavers took approximately 5-6 days to obtain. Since there is variability in the rate of growth and time of death, we needed to stop the development fungal development on cadavers which had the ideal growth for introduction, before we had the sufficient number of cadavers to run our observations. As soon as a cadaver obtained the ideal growth we refrigerated them at 4C, until we obtained the cadavers needed to run the colony observations (maximum of 4 days). Placing did not kill the entomopathogenic fungus, but halted any further growth. Our third cadaver treatment was obtained using a faster technique.
The third cadaver treatment allowed us to observe the natural behaviors towards dead nestmates. To obtain these cadavers which did not pose a disease threat (termed unexposed) we placed workers into the -80˚C freezer for 10 minutes and allowed to thaw.
This treatment controlled for the well documented cadaver management behaviors seen within ant colonies (reviewed in: Lopez-Riquelme & Fanjul-Moles, 2013). Before the introduction of each treatment we took images of the cadavers, allowing us to compare the structure of the cadaver and fungal growth after exposure to workers. Images of exemplar cadavers used for our observations can be found in Figure C.1 and Figure C.2.
Cadaver exposure
71
Due to differences in nest types and ability for the ants to escape each species had a slightly different arena. We placed the C. castaneus semi-natural nests within a cage
(length 17.14 cm ×width 12.22 cm × depth 6.67 cm; Pioneer plastics, item: 049C) containing 10% sugar water and water. While we connected the C. pennsylvanicus nest to the same cage, which we used as a foraging arena. After 24 hours of habituation within the observation arena, we began the nest treatments, inserting the assigned cadavers, and recording the behaviors occurring within the nest.
Within our observations we tested if workers are capable of differentiating between cadavers which are hazardous from non-hazardous. Furthermore we tested the importance of necrophoresis behaviors in the reduction of secondary infections, by either allowing the workers to remove the cadaver within the first 12 hours of exposure (open at
0900) or allow them to remove the cadaver within the second 12 hours of exposure (open at 1800). All of the treatments began at 0900 and ended the next day at 0900, with exception of two C. pennsylvanicus colonies (KG5 and SGL176-F16/2) which required for us to begin at 1230. When we finished recording, we took images of each cadaver and noted their status (broken or complete) along with their location (within or outside of the nest). To eliminate the possibility of contamination we cleaned the forceps used for introduction with 10% bleach (20 seconds), water (oscillation for 5 seconds, repeated twice), and 70% ethanol (30 seconds).
We used a GoPro camera (Hero 2 fitted with both infrared (IR) lens and 4.6 mm macro-lens) to record within nest behaviors. We recorded within nest behaviors for a total of 24 hours, however after 12 hours of recording we stopped recording for a period
72
lasting 5-15 seconds in which we changed the memory card and nest treatment for each treatment group. For each species we used a different apparatus to hold the camera.
Within our C. castaneus recordings we used a foam apparatus covered with red Mylar to hold the camera within the arena, while we used a wooden apparatus for C. pennsylvanicus. Our wooden apparatus was also used to connect our semi-natural nest to the foraging arena, and had built in infrared bulbs. When we introduced cadavers into the nest we turned the lights off and only used red laps to guide us, maintain the nest in full darkness at all times.
Cadaver perspective
Time each cadaver treatment spent within the nest- Removing a source of disease is the first line of defense for a colony. One observer (ESG) watched approximately 1,672 hours of footage to determine the amount of time each treatment group kept the cadaver within the semi-natural nest. We classified cadaver removal to occur when the entire body of the introduced cadaver had been removed from the artificial nest. If the head, thorax, or gaster of the introduced cadaver were kept within the nest we did not classify the cadaver as being removed. To determine differences amongst treatments we used a
Kaplan-Meier analysis.
Percent of individuals whom contacted the cadaver- To determine if workers avoided the exposure to hazardous material, one observer (ESG) watched approximately 1,672 hours of footage to determine the percent of individuals within the nest which contacted the cadaver. We defined a contact event occurring when the antennae and mandibles were above or touching the cadaver, along with any direct body contact workers had with the
73
cadaver introduced. Events in which workers used their antennae to examine the cadaver from a distance were not classified as contact events. We analyzed these data from each species by using a mixed-effect model, in which cadaver treatment was a fixed effect while colony was a random effect. After which we used a least-squares means post hoc analysis with a pair-wise contrast to discern differences amongst our cadaver treatments.
We analyze data from each species separately.
Rate of gaster bending events towards the cadaver- Workers can use formic acid to reduce the infectivity of B. bassiana conidia. We used gaster bending as a proxy to measure the rate at which cadavers were sprayed with formic acid within the nest. In order to reduce observational bias a single observer (ESG) watched approximately 1,672 hours of footage to determine the number of times each worker bent her gaster towards the introduced cadaver. Before analyzing these data we created a proportion for each worker whom performed gaster bending, which allowed us to take into account the variation in the number of times an individual would bend its gaster towards the introduced cadaver and the amount of time each cadaver was kept within the semi-natural nest. The number of gaster bending events performed per individual was divided by the amount of time the introduced cadaver was kept within the nest. In order to analyze these data we used a mixed-effect model in we used cadaver treatment as a fixed effect and colony as a random effect. After which we used a least-squares means post hoc analysis with a pair-wise contrast to discern differences amongst our cadaver treatments. We analyze data from each species separately.
Exposed perspective
74
Infection rate post exposure- After exposing workers to our cadaver treatments we monitored mortality for 15 days after exposure and determined the number of secondary infections. We used three C. castaneus colonies (KFM14-13, KFM14-15, and KFM14-
22) to study the effects of isolation on worker survival and infection rate. We isolated each worker within a petri dish (100 mm × 15 mm dimension) and monitored mortality for 15 days. However, within the rest of our experimental colonies we maintained the workers within their experimental groups and cages (containing water and 10% sugar water which we changed once a week).
Any worker that died within this 15 day period was surface sterilized and monitored daily for B. bassiana growth. We surface sterilized these individuals by placing their bodies in 70% ethanol for 20 seconds (same processed used to obtain the cadavers used for observation), only making an exception for cadavers which had a significant amount of damage to their bodies. After we surface sterilized each cadaver we removed their gaster, placed it within a 2 mL cryogenic tube, and stored within a -80C freezer. While we placed the head and thorax in a sterile petri dish (100 mm × 15 mm dimension) containing filter paper (Whatman 541 filter, 70 mm diameter) saturated with
250 L of sterile water. Which we then incubated at 28C and monitored growth. We used the gaster as a safety net when our incubated sections had Aspergillus sp., bacterial, or in rare occasions no growth. In such cases we mounted the contents of the gaster onto a slide and stained with lactofuchsin (0.1 g of acid fuchsin in 100 mL of lactic acid) and determine the presence of B. bassiana within the gaster. Any fungal material takes on a pink tint, which allows us to reliably conclude the infection status of the worker. In order
75
to determine if cadaver treatment had an effect on mortality we used a Kaplan-Meier analysis, and analyzed each species separately.
Group perspective
Movement for the first 25 minutes after introduction- In order to determine how the cadaver treatment affected worker movement within the nest we collected the x-y coordinates every second for the first 25 minutes after cadaver introduction. We used a
Python code to create frames every second (code available upon request). We obtained the location of each individual using the same code to manually track each worker and cadaver within the nest, using the head thorax juncture as our center. We obtained over
1.7 × 106 x-y coordinates, which we then used to obtain the pair-wise distances between worker and cadaver for every second of observation. In order for us to determine the percent of individuals coming to close proximity of the cadaver we created a radius of
140 pixels around the cadaver, creating a binomial response variable. In order to analyze these data without having an issue with temporal autocorrelation, we created an average proportion of times each individual spent in proximity to the cadaver and performed a square root transformation. We then analyzed these averages with a mixed-effect model, which treated colony as a random effect and treatment as a fixed effect. After which we used a least-squares means post hoc analysis with a pair-wise contrast to discern differences amongst our cadaver treatments. We analyze data from each species and nest treatment separately.
Data analysis
76
We used the R program, including the packages survival, lmer ,and lsmeans, to analyze our data sets (Bates et al., 2015; Lenth, 2016).
Results
Cadaver perspective
Time each cadaver treatment spent within the nest- Cadaver removal is the first behavioral defense against parasites. We found Camponotus castaneus did not significantly differ in the amount of time a cadaver was kept within the nest, independently of nest treatment (Figure 4.1 and Figure 4.3, Kaplan-Meier log rank:
P>0.05). Within our Camponotus pennsylvanicus treatment groups with the ability to remove the cadaver within the first 12 hours of exposure, removed cadavers with external mycosis at a higher rate than other cadaver treatments (Figure 4.2, Kaplan-Meier log rank: P=0.00). When the option of removal is eliminated for the first 12 hours of exposure, all the introduced cadavers were kept within the nest independently of their treatment (Figure 4.4). We suggest C. pennsylvanicus workers must have direct contact with fungal tissue in order to recognize the disease threat. However, when workers are unable to remove the cadaver for 12 hours, they could become habituated to the introduced cadaver or no longer recognize the cadaver as a disease threat, in which case they kept the cadaver within the nest.
Percent of individuals whom contacted the cadaver- Avoidance is another method workers could use to reduce their exposure to Beauveria bassiana. We found C. castaneus treatment groups, independent of their nest treatment, had smaller groups of
77
individuals contacting cadavers with external mycosis in comparison to unexposed cadavers (Figure 4.1, GLM: P=0.00; Figure 4.3, GLM: P=0.00). We also found smaller groups of workers contacting the internal mycosis cadaver when the nest was open for the first 12 hours of exposure (Figure 4.1, GLMM: P=0.03). However, within open C. pennsylvanicus nests we found no significant difference amongst the treatments (Figure
4.2, GLMM: P>0.05). While workers unable to remove the cadaver within the first 12 hours of exposure had significantly smaller groups of individuals coming into contact with the external mycosis cadaver treatment in comparison to the unexposed treatment
(Figure 4.4, GLMM: P=0.04). We suggest workers from both species avoid cadavers which pose an immediate disease threat.
Rate of gaster bending events towards the cadaver- We found C. castaneus groups with independent of their nest treatment bent their gaster more frequently towards cadavers with external mycosis in comparison to unexposed cadavers (Figure 4.1, GLMM: P=
0.05; Figure 4.3, GLMM: P=0.02). Suggesting, C. castaneus workers use of the secretions produced within their poison gland towards hazardous cadavers. While finding
C. pennsylvanicus treatment groups able to remove the cadaver within the first 12 hours of exposure bent their gaster significantly more to cadavers with external mycosis in comparison to other treatments (Figure 4.6B, GLM: P>0.05). Treatment groups unable to remove the cadaver within the first 12 hours of exposure had higher rates of gaster bending towards cadavers with external mycosis in comparison to the unexposed treatment (Figure 4.5B, GLM: P=0.04). Suggesting C. pennsylvanicus workers first use the chemicals within their poison gland as an anti-fungal agent.
78
Exposed perspective
Infection rate post exposure- We found isolation after exposure leads to a higher mortality, independently of the treatment and therefore the colonies in which we isolated the workers will not be used for analysis. However, mortality within the treatment groups in which we did not isolate post exposure was affected by their treatment. Treatment groups exposed to cadavers with external mycosis had a lower survival rates than other treatments independently of their nest treatment (Figure 4.7, Kaplan-Meier log rank:
P<<0.05). We exposed 178 C. castaneus workers to cadavers with external mycosis and from these groups we had a total of 76 deaths. Within our 76 cadavers, we confirmed 72 of them having B. bassiana infection (Figure 4.1 and Figure 4.3). While we exposed 270
C. pennsylvanicus to cadavers with external mycosis, within these treatment groups we had a total of 54 deaths and only 38 of these cadavers had confirmed B. bassiana growth
(Figure 4.2 and Figure 4.4). We did have one anomaly within our C. pennsylvanicus observations, where within our internal mycosis treatments for which we exposed 282 workers, we had 15 deaths, and had 1 confirmed B. bassiana infection.
Group perspective
Movement for the first 25 minutes after introduction- Within our contact data we can see workers avoid the introduced cadaver, reducing their exposure to hazardous material.
However, looking at these patterns over time can inform us of how the group as a whole reacts to the introduction of our treatments. We found external mycosis and unexposed cadaver treatment groups to be significantly different from one another in C. castaneus treatment groups which had the ability to remove the cadaver from the nest within the
79
first 12 hours of exposure (Figure 4.8A, GLMM: P=0.00). Workers within these treatment groups avoided the cadaver with external mycosis, while individuals within the unexposed treatment group spent more time closed to the introduced cadaver. When we removed the ability for workers to remove the introduced cadaver within the first 12 hours of exposure we found the unexposed treatment groups to be significantly different from other (Figure 4.8B, GLMM: P=0.00). Overall C. castaneus workers spent more time close to introduced cadavers which had no fungal growth. We also found a significant difference between cadaver treatments with C. pennsylvanicus treatment groups.
Treatment groups able to remove the cadaver within the first 12 hours of exposure had more individuals coming into contact with cadaver that had external mycosis (Figure
4.9A, GLMM: P= 0.00). While workers in treatment groups unable to remove the cadaver spent more of their time closer to the cadaver with internal mycosis (Figure 4.9B,
GLMM: P=0.00). Suggesting C. pennsylvanicus workers are either attracted to fungal growth or act aggressively towards cadavers with fungal development.
Discussion
Our experiment set out to answer if Camponotus castaneus and Camponotus pennsylvanicus workers are capable of recognizing, and protecting themselves from hazardous material within the nest. We recorded behavioral changes within the nest after inserting one of three cadaver treatments: unexposed, internal mycosis, and external mycosis. The unexposed cadaver posed no disease threat to the colony; since these cadavers are frozen colony members. The internal mycosis cadavers represented a latent
80
disease threat, since the Beauveria bassiana development was only internal and immature. While our external mycosis treatment represented an immediate disease threat, the cadaver had external hyphal growth and mature conidia production (i.e. producing a high concentration of asexual spores). Any contact with the external mycosis cadavers could lead to an infection. Furthermore, we tested the importance of necrophoresis (i.e. cadaver removal) in the prevalence of secondary infections, by either allowing or inhibiting workers from removing the introduced cadaver within the first 12 hours of exposure. We hypothesized small groups of workers would remove the cadaver with external mycosis as soon as possible from the nest, groups which are in open nests will spend significantly less time with the cadaver, reducing the possibility of becoming infected. We also expect external mycosis cadavers to receive higher levels of aggression
(i.e. higher rates of gaster bending events), while workers isolate themselves from the external mycosis cadaver.
The removal of pathogenic material from the nest is one of the first defense mechanisms ant workers can use to decrease the exposure to disease. Within Solenopsis invicta colonies workers remove cadavers with B. bassiana development faster than frozen cadavers (Fan, Pereira, Kilic, Casella, & Keyhani, 2012). However, within our observations we found each species differs in their removal time of the inserted cadavers.
We found no significant differences between the C. castaneus treatment groups, however this is likely caused by the fast removal of internal and external mycosis cadavers within open nests (Figure 4.1). Within C. pennsylvanicus treatment groups with open nests removed the cadavers with external mycosis faster and tend to keep cadavers with
81
internal growth in the nest (Figure 4.2). The stark difference in behavior between the two species is likely caused by their nesting ecology. The soil dwelling nature of C. castaneus exposes these colonies to fungal spores and conidia constantly (Reber & Chapuisat,
2012), their ability to recognize a fungal threat could become fine-tuned over evolutionary time. Enabling workers to recognize different B. bassiana developmental stages and quickly remove the disease threat from the nest. However, C. pennsylvanicus colonies live in wood which is a cleaner environment (Bert Hölldobler & Wilson, 1990, p. 30), the recognition of pathogenic fungi could be significantly dampened. Requiring workers to physically contact the fungal tissue in order to recognize the threat and remove it from the nest. However, reducing the numbers of individuals exposing themselves to harmful material could be beneficial for the colony.
We expected to see small groups of workers contacting the cadavers who impose an immediate disease threat, in order to reduce the exposure and spread of the hazardous material in the nest. We found both C. castaneus and C. pennsylvanicus had significantly lower number of individuals contacting the cadaver with external mycosis, avoiding exposure to the immediate disease threat. However, larger groups of individuals came into contact with the cadavers within the unexposed treatment (Figure 4.1, Figure 4.2,
Figure 4.3, and Figure 4.4), likely caused by the oleic acid accumulation in our unexposed cadavers, which induce worker grooming (Edward O. Wilson, Durlach, &
Roth, 1958). Counting the number of individuals does not take into account how contact changes over time and which individuals spent their time with the cadaver. Within our observations we tended to see an initial high contact rate, which then decreased over the
82
course of time the cadaver was kept within the nest (Figure 4.8 and Figure 4.9).
Suggesting workers interest change over time and potentially dues to the reduce disease threat imposed by the cadaver.
Determining how contact changes over time would allow us to better understand the role of avoidance in disease management. We found C. castaneus treatment groups exposed to cadavers with external mycosis avoid the cadaver longer, while they spend more time closer to unexposed cadavers (Figure 4.8). While, C. pennsylvanicus workers with the ability to remove the cadaver within the first 12 hours of exposure had more individuals coming into contact with cadavers with external mycosis growth (Figure
4.9A). When these workers did not have the ability to remove the cadaver from the nest, they spent more time closed to cadavers with internal mycosis. However, we recommend such observations should be performed for a longer period of time to determine the efficiency of avoidance in reducing infection. Furthermore, we suggest looking at how conidia movement changes over time. Such an experiment we could test if workers avoid the locations with highest conidia concentration. Within our observations we saw significant differences in how workers used space when confronted with an infectious agent; however avoidance did not reduce infections.
After a worker has contact with hazardous material they can use the chemical cocktail within their poison gland as an antimicrobial, reducing conidia viability
(Attygalle & Morgan, 1984; Blum et al., 1958; Li et al., 2012; Storey et al., 1991; Simon
Tragust, Mitteregger, et al., 2013). We found C. castaneus colonies bent their gaster significantly more towards cadavers with external mycosis, independently of their nest
83
treatment (Figure 4.5A and 4.6A). Similarly, C. pennsylvanicus workers bent their gaster significantly more towards cadavers with external mycosis independently of their nest treatment (Figure 4.5B and Figure 4.6B). Suggesting C. castaneus uses formic acid as a secondary defense when unable to remove the cadaver from the nest, while C. pennsylvanicus acts aggressively towards fungal material. We are interested in determining the efficiency these attacks had in reducing the number of individuals which became infected with B. bassiana post exposure.
Many experiments both in vitro and in vivo have shown the antifungal properties possessed by the chemical cocktail produced within the poison gland (Attygalle &
Morgan, 1984; Blum et al., 1958; Oi & Pereira, 1993; Storey et al., 1991; Simon Tragust,
Mitteregger, et al., 2013). However, these experiments do not empirically test the survival and rate of infection of workers post exposure. After our 24 hour exposure to our cadaver treatment, we followed each treatment group for 15 days and determined the number of individuals who had succumbed to B. bassiana. Within our observations both
C. castaneus and C. pennsylvanicus workers exposed to cadavers with external mycosis had a higher mortality and confirmed B. bassiana infection. Smaller number of workers came into contact with external mycosis cadavers, as well as removed them from the nest quickly workers. However, workers still became infected with B. bassiana. Our data suggests workers are capable in reducing the number of infections, but cannot eliminate the possibility of becoming infected.
Within our observations we challenged workers with different cadavers, testing their ability to recognize and defend themselves from infection. We found C. castaneus
84
and C. pennsylvanicus can recognize disease threats when the B. bassiana growth has become external. However, C. castaneus workers are capable of also recognizing B. bassiana development within the cadaver, while C. pennsylvanicus cannot. Furthermore workers change their use space when confronted with infectious cadavers. We found C. castaneus workers isolate themselves, while C. pennsylvanicus increase their proximity to the cadaver. Furthermore, groups exposed to cadavers with external mycosis did become infected with B. bassiana, even though workers employed multiple defense strategies. Suggesting the defense mechanisms are effective in reducing disease, but unable in eradicating the possibility of infection.
Future work within this area should focus on how the entomopathogen is affected by the exposure to workers. Determining how conidia concentration, germination rate, and hyphal structure are affected by the exposure to workers. Such work could focus more on the pathogen’s biology, and inform our understanding of ant colony dynamics.
We also propose an experiment which determines how the movement of conidia inside the nest affects the movement of workers, to determine the capability of workers to avoid high conidia concentrations within the nest. Further observations which look at the simple biological questions of host and pathogen are needed before we extrapolate any observation to an evolutionary adaptation.
Conclusion
We found nest ecology to be a significant behavioral driver of cadaver management. The soil nesting ecology of C. castaneus enables them to recognize both
85
mature and immature stages of fungal development. While an ant species living in wooden nests, C. pennsylvanicus can only detect the fungal disease threat when they come in contact with fungal material. Although workers used multiple defense strategies to avoid infection we still found workers exposed to cadavers with external mycosis died at a higher rate. The defense mechanisms employed by ant workers are flawed.
86
87
Figure 4.1 – Time spent within the nest, percent of contacts, mortality, and infection for Camponotus castaneus nests which are open for the first 12 hours of exposure.
Each row on the graph represents data collected from a different colony. The colony labels can be found on the y-axis. The color of each graph denotes the cadaver treatment: external mycosis (red), internal mycosis (green), and unexposed (blue). The length of the bars within the “Time spent in nest” section represent the amount of hours the cadaver was kept within the nest, while the nest was open (i.e. workers had the ability to remove the cadaver). We found no significant difference amongst cadaver treatments (Kaplan-
Meier log rank: P=0.25). The bar graphs within the “Contact” section represents the number of individuals which came into contact with the cadaver while inside the nest.
We have placed the percent of contacting individuals within the bar, along with the actual value. We found smaller groups of workers come into contact with cadavers that have fungal growth (GLMM: P<0.05). The percentages within the “Mortality” column denote the percent of individuals that died in each group over the course of 15 days after exposure. We then measured the numb which had confirmed Beauveria bassiana growth and placed them in the “Infection” column. Workers exposed to an external mycosis cadaver and died within out 15 day observation period, had a high probability of being infected with B. bassiana.
88
89
Figure 4.2 – Time spent within the nest, percent of contacts, mortality, and infection for Camponotus pennsylvanicus nests which are open for the first 12 hours of exposure. Each row on the graph represents data collected from a different colony. The colony labels can be found on the y-axis. The color of each graph denotes the cadaver treatment: external mycosis (red), internal mycosis (green), and unexposed (blue). The length of the bars within the “Time spent in nest” section represent the amount of hours the cadaver was kept within the nest, while the nest was open (i.e. workers had the ability to remove the cadaver). Workers removed cadavers with external mycosis at a faster rate
(Kaplan-Meier log rank: P=0.00). The bar graphs within the “Contact” section represents the number of individuals which came into contact with the cadaver while inside the nest.
We have placed the percent of contacting individuals within the bar, along with the actual value. We found no significant difference in the number of workers which came into contact with the different cadaver treatments. The percentages within the “Mortality” column denote the percent of individuals that died in each group over the course of 15 days after exposure. We then measured the numb which had confirmed Beauveria bassiana growth and placed them in the “Infection” column. Workers exposed to an external mycosis cadaver and died within out 15 day observation period, had a high probability of being infected with B. bassiana. However, we did have one anomaly in which a worker within the internal mycosis treatment also became infected with B. bassiana.
90
91
Figure 4.3- Time spent within the nest, percent of contacts, mortality, and infection for Camponotus castaneus nest which are closed for the first 12 hours of exposure.
Each row on the graph represents data collected from a different colony. The colony labels can be found on the y-axis. The color of each graph denotes the cadaver treatment: external mycosis (red), internal mycosis (green), and unexposed (blue). The length of the bars within the “Time spent in nest” section represent the amount of hours the cadaver was kept within the nest, while the nest was open (i.e. workers had the ability to remove the cadaver). After cadavers have been within the nest for more than 12 hours the likelihood of workers removing the cadaver significantly decreases (Kaplan-Meier log rank: P>0.05). The bar graphs within the “Contact” section represents the number of individuals which came into contact with the cadaver while inside the nest. We have placed the percent of contacting individuals within the bar, along with the actual value.
Smaller groups of workers within the external mycosis treatment contacted the cadaver, in comparison to the unexposed treatment (GLMM: P=0.00). The percentages within the
“Mortality” column denote the percent of individuals that died in each group over the course of 15 days after exposure. We then measured the numb which had confirmed
Beauveria bassiana growth and placed them in the “Infection” column. Workers exposed to an external mycosis cadaver and died within out 15 day observation period, had a high probability of being infected with B. bassiana.
92
93
Figure 4.4- Time spent within the nest, percent of contacts, mortality, and infection for Camponotus pennsylvanicus nest which are closed for the first 12 hours of exposure. Each row on the graph represents data collected from a different colony. The colony labels can be found on the y-axis. The color of each graph denotes the cadaver treatment: external mycosis (red), internal mycosis (green), and unexposed (blue). The length of the bars within the “Time spent in nest” section represent the amount of hours the cadaver was kept within the nest, while the nest was open (i.e. workers had the ability to remove the cadaver). After cadavers have been within the nest for more than 12 hours workers do not remove the cadaver (Kaplan-Meier log rank: P>0.05). The bar graphs within the “Contact” section represents the number of individuals which came into contact with the cadaver while inside the nest. We have placed the percent of contacting individuals within the bar, along with the actual value. Smaller groups of workers came into contact with the external mycosis cadavers in comparison to the unexposed cadaver treatment (GLMM: P=0.04). The percentages within the “Mortality” column denote the percent of individuals that died in each group over the course of 15 days after exposure.
We then measured the numb which had confirmed Beauveria bassiana growth and placed them in the “Infection” column. Workers exposed to an external mycosis cadaver and died within out 15 day observation period, had a high probability of being infected with
B. bassiana.
94
Figure 4.5 – Rate of gaster bending within nests which are closed for the first 12 hours of exposure. Each colored circle represents the number of gaster bending events per colony: red (external mycosis), green (internal mycosis), and blue (unexposed). The black whisker plots depict the mean ± standard error for each treatment group. Sample sizes are above each column of data, along with the letters denoting significant differences amongst treatments. (A) Represents the data collected from Camponotus castaneus groups. While (B) depicts the data collected from Camponotus pennsylvanicus groups. We found C. castaneus and C. pennsylvanicus workers bent their gaster more frequently towards cadavers with external mycosis (GLMM: P<0.05).
95
Figure 4.6 – Rate of gaster bending within nests which are open for the first 12 hours of exposure. Each colored circle represents the number of gaster bending events per colony: red (external mycosis), green (internal mycosis), and blue (unexposed). The black whisker plots depict the mean ± standard error for each treatment group. Sample sizes are above each column of data, along with the letters denoting significant differences amongst treatments. (A) Represents the data collected from Camponotus castaneus groups. While (B) depicts the data collected from Camponotus pennsylvanicus
96
groups. We found C. castaneus and C. pennsylvanicus workers bent their gaster more frequently towards cadavers with external mycosis (GLMM: P<0.05).
97
Figure 4.7 – Survival post exposure. Within this graph we summarize the mortality of workers exposed to three different cadaver treatments: external mycosis (red), internal mycosis (green), and unexposed (blue). Graphs A and C contain the data collected from 5 genetically distinct colonies of Camponotus castaneus. (A) Depicts the mean mortality data ± standard deviation from treatment groups which had the ability to remove the cadaver within the first 12 hours of cadaver introduction. (C)
Depicts the mean mortality data ± standard deviation from treatment groups which had the ability to remove the cadaver 12 hours after cadaver introduction. While graphs B and D depict data collected from 8 genetically distinct colonies of
98
Camponotus pennsylvanicus, exposed to one of three cadaver treatments. (B) Depicts the mean mortality data ± standard deviation from treatment groups which had the ability to remove the cadaver within the first 12 hours of cadaver introduction. (D)
Depicts the mean mortality data ± standard deviation from treatment groups which had the ability to remove the cadaver 12 hours after cadaver introduction. Within both species we found that individuals exposed to a cadaver with external mycosis had a higher incidence of mortality (Kaplan-Meier log rank: P<0.05).
99
100
Figure 4.8- Percent of individuals near the cadaver within nests containing
Camponotus castaneus workers. Within these graphs we depict the average percent of individuals within the 140 pixels to the cadaver within each of the 1500 frames. Black lines with colony names denote the removal of a cadaver at a given time point. (A)
Represents data we collected from nest we which had the ability to remove the cadaver from the nest for the first 12 hours of exposure. Within these data we found individuals within the external mycosis treatment isolated themselves from the introduced cadaver, in comparison to workers within the unexposed treatment which spent more time closer to the cadaver (GLMM: P=0.00). (B) Represents data we collected from nest we allowed to remove the cadaver 12 hours after introduction. Workers within the unexposed treatment group spent more of their time closer to the cadaver in comparison to other treatments
(GLMM: P=0.00). Our results suggest C. castaneus workers are attracted to cadavers which do not have a fungal infection.
101
102
Figure 4.9- Percent of individuals near the cadaver within nests containing
Camponotus pennsylvanicus workers. Within these graphs we depict the average percent of individuals within the 140 pixels to the cadaver within each of the 1500 frames. Black lines with colony names denote the removal of a cadaver at a given time point. (A) Represents data we collected from nest we which had the ability to remove the cadaver from the nest for the first 12 hours of exposure. Within these data we found individuals within the external mycosis treatment more time closed to the introduced cadaver in comparison to the other treatments (GLMM: P=0.00). (B) Represents data we collected from nest we allowed to remove the cadaver 12 hours after introduction.
Workers within the internal mycosis treatment group spent more of their time closer to the introduced cadaver in comparison to other treatments (GLMM: P=0.00). Our results suggest C. pennsylvanicus workers could be attracted to fungal growth or are highly aggressive towards cadaver with fungal development.
Chapter 5: The effectiveness of social behaviors in reducing fungal
pathogenicity
Emilia Solá Gracia 1, 2, 3, Nina Jenkins 3, and David P. Hughes 2, 3, 4
1. Ecology, Huck Institutes of Life Sciences, Pennsylvania State University,
University Park, Pennsylvania, 16802
2. Center for Infectious Disease Dynamics, Huck Institute of Life Sciences,
Pennsylvania State University, University Park, Pennsylvania, 16802
3. Department of Entomology, Pennsylvania State University, University Park,
Pennsylvania, 16802
4. Department of Biology, Pennsylvania State University, University Park,
Pennsylvania, 16802
104 Introduction
Eusocial insects (bees, wasps, termites, and ants) must defend themselves against parasites. Their nesting ecology and lifestyles are ideal for parasites to take advantage.
Colonies contain readily available hosts to infect and the social interactions between nestmates enable parasite transmission. Eusocial insect colonies are comprised of highly- related kin groups (W. D. Hamilton, 1963, 1964a, 1964b). Workers rely on one another in order to survive. Eusocial insect colonies are best described as a machine built for reliable performance, workers act in parallel to accomplish different tasks for the benefit of the colony (Oster & Wilson, 1978; Schmid-Hempel, 1998, pp. 119-124; E. O. Wilson,
1968). In order to maintain cohesion amongst workers a constant exchange of information is needed, which in most cases is in the form of chemical communication.
Such interactions are used for task switching (Gordon, 1989), communicate food location
(Mc Cabe, Farina, & Josens, 2006; Pinter-Wollman et al., 2013), nestmate recognition
(Esponda & Gordon, 2015; Mitra, Ramachandran, & Gadagkar, 2014), and grooming
(Zhukovskaya et al., 2013). Eusocial insect colonies are comprised of individuals with very similar genetic makeup who are in constant contact with one-another, creating the ideal circumstances for parasites to take advantage of a large group of hosts. While eusocial insect colonies seem vulnerable to parasites, they do have well established defense mechanisms in place to reduce parasite infection and transmission. However, the first step in reducing parasitic infection is recognition.
Chemical communication is imperative within eusocial insect colonies (Richard
& Hunt, 2013). Eusocial insect colonies use their keen sense of smell to protect their nest entrance (e.g. bees) (Nouvian, Reinhard, & Giurfa, 2016) and establish territories (e.g.
105 ants) (Bert Hölldobler & Wilson, 1990, pp. 286-289). Workers can use their antennae to detect slight changes in cuticular hydrocarbon profile, in order to differentiate others as kin or conspecifics (Esponda & Gordon, 2015; Murakami, Nunes, Desuo, Shima, &
Mateus, 2015; Nascimento & Nascimento, 2012; Sharma et al., 2015; Singer & Espelie,
1992). The ability to detect minute changes in chemical profiles allows eusocial insects to detect microparasites before they infect and transmit inside the colony. For example, honey bees (Apis mellifera) are capable of detecting brood infected with Bacillus larvae, infected larvae and pupae are removed from the nest (Spivak & Gilliam, 1998), while ants and termites increase allogrooming behaviors when detecting lethal entomopathogenic conidia ( asexual fungal spores) on nestmates (Myles, 2002; Reber et al., 2011; R. B. Rosengaus, Maxmen, Coates, & Traniello, 1998; Simon Tragust,
Mitteregger, et al., 2013).
Grooming is the first line of defense against parasites (Farish, 1972; S. Tragust,
2016; E. O. Wilson & Regnier, 1971; Zhukovskaya et al., 2013). Eusocial insects are known for their ability to recognize and actively remove parasites. Honey bee, A. mellifera, colonies have a designated caste responsible of removing the mites off the wing axis and petioles of other workers (Bozic & Valentincic, 1995). While termites exposed to lethal Metarhizium anisopliae conidia (asexual spores) drum their gaster on the nest floor and are groomed by nestmates to remove the conidia, these activities can also turn into cannibalism or burial of the exposed worker (Thomas Chouvenc & Su,
2012; Fujii, 1975, pp. 35-36; R. B. Rosengaus et al., 1998; Rebeca B. Rosengaus et al.,
2011, pp. 179-182). Termites can also defend themselves using the antimicrobial chemicals produced within the sterna gland to reduce conidia viability (R. B. Rosengaus,
106 Traniello, Lefebvre, & Maxmen, 2004). Similar to termites, ants have the ability to actively remove microparasites from their cuticle, and have chemical defenses against pathogens (Attygalle & Morgan, 1984; Myles, 2002; Reber et al., 2011; S. Tragust,
2016).
Specialized exocrine glands has evolved independently twice within ants (Bert
Hölldobler & Wilson, 1990, p. 30; S. Tragust, 2016). The metapleural gland is a synapomorphic trait in ants; this organ is located at the posterior of the ants’ thorax and varies in size across species (B. Hölldobler & Engel-Siegel, 1984; S. Tragust, 2016).
Leaf-cutting ants highly dependent on the metapleural gland and their actinomycete
(gram positive bacteria) symbiont to maintain their fungal garden free of parasites (Little et al., 2006). Ant species within the genera Camponotus, Dendromyrmex, Oecophylla, and Polyrhachis have lost their metapleural gland, instead these species use the chemicals produced within the poison gland (i.e. acidopore) to defend themselves against microparasites (Bert Hölldobler & Wilson, 1990, p. 30). The chemical cocktail produced inside the poison gland is highly effective in reducing the viability of fungal tissue
(Attygalle & Morgan, 1984; Blum et al., 1958; Cole, Blum, & Roncadori, 1975; Storey et al., 1991; Simon Tragust, Mitteregger, et al., 2013).
Actively removing and reducing the viability of conidia (asexual spores) increases the survival of exposed colony members. Brood exposed to M. anisopliae conidia have a higher likelihood of survival when workers are capable of actively removing the conidia and reducing their viability with the secretions from their poison gland (Simon Tragust,
Mitteregger, et al., 2013). The chemicals produced within the poison gland have been mostly performed in vitro (Attygalle & Morgan, 1984; Blum et al., 1958; Cole et al.,
107 1975; Storey et al., 1991). Our study empirically tests how Beauveria bassiana is affected by being exposed to Camponotus pennsylvanicus within a semi-natural nest environment.
We set out to determine: (1) the number of secondary infections produced by one cadaver, (2) the effect workers have on conidia concentration, (3) the effect workers have on conidia germination, and (3) how fungal structure changes post exposure. In this study we exposed our samples to four different scenarios: infectious stage fungi growing from cadavers never exposed to either the nest environment or workers (termed untreated), infectious stage fungi growing from cadavers placed within an empty nest (termed nest + no workers), infectious stage fungi growing from cadavers exposed to approximately 20 workers within a closed nest (termed closed nest + workers), and infectious stage fungi growing from cadavers exposed to approximately 20 workers within an open nest (termed open nest + workers). We expected to see a significant increase in mortality in nests where workers unable to remove our hazardous sample. We also expect to see lower conidia concentrations and germination rates. Samples placed within these treatments would also have significant fungal tissue damage, caused by the prolonged exposure to aggressive workers.
Materials and Methods
Camponotus pennsylvanicus collection
We collected three C. pennsylvanicus colonies in State College, Pennsylvania by cutting into fallen trees, decaying logs, and other woody debris. Since this species is polydomous, meaning the colony uses multiple nest sites as it grows, not all colonies are
108 queen-right. We collected all the colonies in October 2016 and their demography is as follows: PWN1-FA16, consisted of approximately 2,000 workers and brood; PWN6-
FA16, consisted of approximately 4,000 workers, sexuals, and brood; PWN8-FA16, consisted of approximately 4,500 workers and brood.
Nest design
We laser cut medium-density fiberboard to created wooden nests which have the same dimensions (length 11.7 cm, width 5.7 cm, depth 0.6 cm) and structure. The design consisted of two chambers (length 3.78 cm × width 5.05 cm), which we connected with one centralized corridor (length 3.45 cm × width 1.27 cm). In order to insert our samples with minimal disturbance to the workers, we created an aperture on the side of the corridor. We kept workers contained by placing a fitted acrylic Plexiglas over the nest and used a wooden locking mechanism for our corridor aperture. We kept colonies in full darkness using red Mylar.
Experimental colony habituation and maintenance
Within each genetic colony we had five replicate observations. Treatments containing workers (i.e. closed nest + workers and open nest + workers) had an average of 19 ± 2 (± SD) workers inside of the nest. This variation was caused by mortality and escapes over the course of our experiment. In order to obtain approximately 20 workers accepting the semi-natural nest, we would cool down 25 workers in 4C refrigerator for approximately 10 minutes and then force them into the semi-natural nest. We locked the workers inside the nest for 1 hour, after which we allowed the workers to move freely between the cage and the semi-natural nest. The nest was placed within a Sistema storage container (length 35.56 cm × width 25.40 cm × height 12.70 cm) containing water and
109 10% sugar water given ad libitum. The day after introduction we removed the workers which had rejected the nest, maximum of 5 workers per treatment group, to increase the number of workers which would stay inside the nest for our observations. We began our habituation period the day after this last disturbance.
We habituated each group to the experimental nest for 6-10 days, depending on the infection process. While the ants habituated to the nest, we also infected workers to produce the samples needed for our observations. Due to the variation in mortality and B. bassiana growth between workers, we had a range of habituation periods. However, when we reached the correct number of samples for our observations we placed the workers into our exposure arena 24 hours before introduction. The exposure arena consisted of the semi-natural nest adhered, with modeling clay and straw, to a smaller foraging arena (length 17.14 cm × width 12.22 cm × height 6.03 cm; Pioneer plastics, item: 049C) containing 10% sugar water and water. In order to keep our nest in full darkness while preparing them for sample introduction we turned the lights off and only used lamps with red light.
Cadaver preparation
In our experiment we are interested in understanding how an entomopathogenic fungus, Beauveria bassiana, is affected by the behaviors of C. pennsylvanicus workers, using a semi-natural arena. In order to obtain the samples used for our observations we developed a cohort system. We randomly selected 5 workers to be a part of one cohort.
We used three cohorts, distributed over the course of 5 days to obtain the samples we needed. We exposed the first cohort to B. bassiana conidia four days before we began
110 habituation on our ant treatment groups. Our second cohort was exposed two days before habituation and the third cohort a day after habituation began.
We carried out infections by creating a highly concentrated conidia suspension in saline-tween (0.05 %). We placed 500 μL of the suspension onto two Whatman 541 filters (70 mm diameter) within a petri dish (100×15 mm dimension). After the filter paper absorbed the liquid, we placed one cohort into the plate and exposed them to conidia for 24 hours. We then released the cohort into a cage with a housing tube, 10% sugar water and water given ad libitum. We performed daily mortality observations and collected the deceased.
In order to reduce contamination, we surface sterilized cadavers by placing them in 70% ethanol for 20 seconds. After surface sterilization we placed each cadaver into an individual petri dish (60 × 15 mm dimension) with a saturated (250 L of sterile water) filter paper (Whatman 541, 47 mm diameter). We incubated cadavers at 28˚C and made daily observations to detect B. bassiana development. Cadavers that had abnormal growth, contamination (Aspergillus spp. contamination), or the presence of any other organism (mites or nematodes) were not used in our observations. Due to the variability in time of death and fungal growth, we needed to halt the fungal development on cadavers which had the ideal growth for our observations. We refrigerated these samples at 4C, until we obtained the correct number of samples needed to run the colony observations. We only allowed a sample to be refrigerated for a total of 4 days, if we did not obtain the total numbers of samples needed in time to use the refrigerated sample it was discarded. Placing our samples into the fridge did not kill the entomopathogenic fungus, but halted any further growth.
111 Exposure and post exposure
All of the samples used for our observation had external B. bassiana development with mature conidia production (Figure D.1, Figure D.2A, and Figure D.3A), which took approximately 5-6 days to establish. We assigned each sample to one of four treatments: untreated, nest + no workers, closed nest + workers, and open nest + workers. Before we placed our sample into their treatment we took images of their fungal development for a comparison post treatment, with exception of the unexposed treatment. Treatments began approximately at 1500-1530 and ended at the same time the next day.
After each treatment we took images of the nests and samples to record sample location (in or outside of the nest) and damage (broken or complete). We also took images of each sample under a stereomicroscope to collect more detailed information on cadaver damage post exposure (Olympus SZX16). Then we tested the conidia concentration and germination rate for each sample post exposure, with exception of the untreated samples. We also made daily mortality observations on the ants we exposed to our samples. In order to analyze our mortality data we used the Kaplan-Meier analysis, and tested the effect nest treatment (open versus closed) has on worker mortality.
Cadaver solution
We used four different treatments to determine how the conidia concentration and germination rate are affected by the exposure to workers: never exposed to nest or ants
(untreated), exposed to an empty nest for 24 hours (not exposed), exposed to a closed nest with workers for 24 hours (closed), and an open nest with workers for 24 hours
(open). Untreated samples allowed us to determine the baseline concentration and germination rate of our samples. While the cadavers we introduced to the nest without
112 workers allowed us to determine the effects of handling or a dry environment have on conidia concentration and germination. While nests containing workers allowed us to capture the effects exposure to workers has on conidia concentration and germination.
In order to determine the effects ant workers have on conidia concentration and germination we created a solution of each sample using saline-tween (0.05%). We first added 100 L of saline-tween (0.05%) to a 2 mL micro-centrifuge tube containing our cadaver sample, and vortexed for 10 seconds at maximum intensity. If we saw a high concentration of fungal material, we repeated these steps until we reached an ideal concentration for data collection. We used a median of 200 L of saline-tween (0.05 %) for our solutions. However, we did have one sample which required 1,400 L of saline- tween (0.05 %). Before using the solution for data collection we removed the cadaver from the micro-centrifuge tube, to reduce the possibility of changing the conidia concentration while handling our solutions.
Conidia concentration
In order to detect how workers’ affect the conidia concentration within our sample we placed 20 L (10 L per side) of our solution onto a hemocytometer to determine the conidia concentration. We obtained the number of conidia within each 0.04 mm2 area on the slide. We measured the number of conidia per sample using the formula depicted below. We used a logarithmic transformation, before using a mixed-effect model to analyze out data. Within our model we treated colony and replicate as random effects, while treatment was a fixed effect. We used a least-squares means post hoc analysis with a pair-wise contrast to determine the significant differences amongst treatments.
113 Conidia concentration per sample=
total sum of conidia counted on the hemocytometer volume of saline tween added 104 2 1000
Germination rate
To detect the effects our treatments had on conidia germination we placed 10 L of our solution onto three petri dishes (60 × 15 mm dimension) containing sabouraud dextrose agar media, supplemented with Kanamycin (Sigma-Aldrich) and Penicillin
Streptomycin (Corning). We evenly spread our solution over the media, using a cell spreader (VWR), and then incubated o at room temperature (21˚C) for 16 hours. We assessed germination using a compound microscope (magnification × 400) and classified a conidium as germinated when the germ tube was at least a third of the length of the conidium. In order to obtain the germination rate per plate we examined at least 300 conidia. We calculated the sample’s germination rate by averaging the germination rates within the three plates. In order to examine the effects our treatments had on the germination rate we square root arcsine transformed our data before analysis (similar transformation performed in M. B. Thomas & Jenkins, 1997). We used a mixed-effect model to test our dataset. We treated colony and replicate as random effects, while treatment was a fixed effect. We used a least-squares means post hoc analysis with a pair- wise contrast to determine the significant differences amongst treatments.
Fungal structure
For scanning electron microscopy we allowed the samples to desiccate post treatment for approximately 3 weeks. We then placed each sample on a stub and coated
114 with gold using Denton Vacuum desk II, which we then viewed using a Zeiss
GeminiSEM. We also created z-stack images using the Axion Zoom V16 microscope.
Data analysis
We used the R program, including the packages survival, lmer ,and lsmeans, to analyze our data sets (Bates et al., 2015; Lenth, 2016).
Results
Post exposure
Samples exposed (i.e. cadavers with mature Beauveria bassiana development) to workers experienced severe damage in comparison to those which we did not expose to workers (see Figure 5.1, Figure 5.2, Figure 5.3, Figure D.1, Figure D.2, and Figure D.3).
All of the samples within treatment groups with an open nest were removed from inside the nest. We also found a significant effect of colony and treatment on worker mortality
(Kaplan-Meier log rank: P=0.00; Figure 5.4). Workers within the open nest treatment died at a higher rate (lowest survival rate of 78%), than those within the closed nest treatment (lowest survival rate of 89%). Suggesting workers within closed nests spent more time avoiding the cadaver, reducing the exposure to deadly B. bassiana conidia.
The workers exposed to our samples became infected with B. bassiana. After exposing 347 workers within the open nest arena, we had a total of 41 deaths which all but 3 had confirmed B. bassiana infection. Within our closed nest treatment we exposed
340 workers, only had 25 deaths, and all of them had confirmed B. bassiana infections.
115 Our observations suggest workers are capable of significantly reducing secondary infections, but cannot completely avoid infection.
Conidia concentration
Reducing the number of conidia on the sample could be a behavioral strategy used to lower the probability of infection. Although conidia production is variable amongst samples (Silva & Diehl-Fleig, 1988), we did see a clear pattern within our data
(Figure 5.5). Samples exposed to workers had significantly less conidia, than samples which we did not expose to workers (GLMM: P<0.05). Suggesting workers are capable of reducing the conidia concentration they are exposed to after introduction. However, active removal of conidia is not the only explanation for the significant decrease.
Workers could easily dislodge conidia off of the cadaver while handling the hazardous material, spreading the conidia inside of the nest potentially increasing the exposure to lethal conidia. We can also see a large variation of concentration in our empty nest samples (i.e. nest + no workers treatment). Suggesting sample handling also affects conidia concentration.
Germination rate
Determining the percent of conidia capable of germinating (i.e. produce a germ tube), allows us to approximate the infectivity of our samples before or after exposure to our treatments. We found samples exposed to workers had a significantly lower germination rates, in comparison to those not exposed to workers (GLMM: P=0.00;
Figure 5.6). However, we did see a lot of variability within the samples placed inside an empty nest (i.e. nest + no workers treatment). This variation is likely caused by the lack of humidity within our semi-natural nests, affecting the viability of our conidia.
116 Fungal structure
While cadavers were clearly affected by being exposed to workers (see
Supplementary Figure D.1, Figure D.2, and Figure D.3) we used scanning electron microscopy and z-stack images to show changes in hyphal structure. We found significant changes in structure depending on the treatment. Samples exposed to workers experience significant more damage while the other treatment groups did not (Figure 5.1,
Figure 5.2, and Figure 5.3).
Discussion
We set out to examine how Beauveria bassiana is affected by the exposure to
Camponotus pennsylvanicus workers. Within our observations we used four cadaver treatments: untreated, exposed to an empty nest, exposed to ants within a nest unable to remove the introduced sample, and exposed to ants in an open nest. We collected data on sample location, sample damage (i.e. broken or intact), workers mortality, conidia concentration, conidia germination, and fungal structure changes post treatment. We expected to see samples placed inside closed nests with workers to receive the most damage, highest worker mortality, have the lowest conidia concentration, and germination rates. Since these samples would have a longer exposure time to worker aggression and be sprayed more than samples which were placed inside an open nest (see
Chapter 4).
Within our observations we found a significant effect of colony and treatment on worker mortality after exposure (Figure 5.4). Workers placed within the open nest treatment had significantly higher mortality, than workers within closed nests. Although
117 workers within open nest could quickly remove the cadaver from the nest, their movement inside of the nest could have spread the conidia across the nest floor and expose a large group of workers to the lethal conidia. While workers within the closed nest could isolate themselves from the cadaver, and restrict their movement to one chamber significantly reducing their exposure to conidia and increasing their chances of survival. An experiment which tests conidia dispersal caused by worker movement would test our explanation for the higher mortality seen within our open nest treatments.
The conidia produced on our samples could easily be aerosolized and moved accord the nest floor, if a worker is unable to detect these loose conidia they could easily become infected. Another form of defense could be reducing the number of conidia on the introduced sample.
Reducing the concentration of conidia on the cadaver would reduce the possibility of infection when coming into contact with the cadaver. Within our observations we found cadavers exposed to workers had significantly lower concentrations of conidia
(Figure 5.5). The reduction in conidia could also be indicative of conidia dispersal inside of the nest, potentially exposing more workers to lethal conidia. Conidia concentration is not the only variable which determines the lethality of our sample. The ability a conidium has to produce a germ tube dictates its ability to infect a host.
We expected to see the lowest germination rates within the nests unable to remove our sample. The workers inside these nests would have more time to use more of the chemicals produced within their poison gland, significantly reducing conidia viability
(Attygalle & Morgan, 1984; Blum et al., 1958; Cole et al., 1975; Storey et al., 1991;
Simon Tragust, Mitteregger, et al., 2013). However, our data suggests any exposure to
118 workers drastically decreases conidia germination rates (Figure 5.6). It would be interesting to test if there are environmental factors within the nest that also affect the germination of B. bassiana conidia. Within our observation we found a change in humidity affects conidia germination rate (Bouamama, Vidal, & Fargues, 2010; Fargues
& Luz, 2000; Hong, Gunn, Ellis, Jenkins, & Moore, 2001). Hölldobler and Wilson
(1990, p. 30) suggest living within a wooden environment is cleaner. We propose that the lack of humidity relieves these colonies from being in constant contact with viable fungal material. It would be interesting to perform these observations within the parameters of a soil-dwelling species, since they live in higher humidity. We would suggest conidia would not be limited by humidity and possibly be able to infect workers for longer. We suggest performing experiments in a semi-natural setting in order to better understand the underlying biological factors affecting parasites and their host communities.
Conclusion
We found Beauveria bassiana is greatly affected by being exposed to
Camponotus pennsylvanicus workers. The significant reduction in conidia concentration, germination rate, and drastic changes in fungal structure suggests workers are highly capable of neutralizing a fungal threat inside the nest. Removal, avoidance, and spraying are mechanisms workers can use to reduce their exposure and the viability of lethal conidia (Chapter 4). Within this experiment we provide evidence, using a semi-natural arena, that workers are capable of greatly affecting the lifecycle of a generalist entomopathogen. Furthermore, the nesting ecology of C. pennsylvanicus could also be
119 providing a strong advantage. The drier environment inside of the nest could be reducing conidia viability (Bouamama et al., 2010; Fargues & Luz, 2000; Hong et al., 2001), allowing workers to live in an environment which has less hazardous fungal material. In order to disentangle the complexities within host-parasite interactions we must understand the basic biology of each organism.
120
Figure 5.1- Images of a cadaver introduced into an empty nest for 24 hours. Image
(A) illustrates the full body of an exemplar sample used for our nest + no workers treatment. (B) Shows the legs of the same cadaver, notice the numerous hyphae and extensive conidia production. While image (C) is a SEM image of the gaster, notice the extensive hyphal production. And image (D) is an illustration of the hyphal material protruding from the cadaver at a higher magnification.
121
Figure 5.2- Images of a cadaver introduced to a closed nest with workers for 24 hours. Image (A) depicts how the head of a sample placed in the closed nest + workers treatment is damaged. The antennae were removed and the growth extending from the antennae has significant damage. While (B) illustrated the damage done to the gaster, the fungal growth has been completely removed.
122
123 Figure 5.3- Images of a cadaver placed within an open nest with workers for 24 hours. (A) An image of the head of the exemplar cadaver used within the open nest + workers treatment. We can only see one small section of fungal tissue in the head-thorax junction. (B) An illustration of the thorax, where most legs have been removed and the fungal tissue is severely damaged. While image (C) depicts the damaged done to the gaster, the fungal tissue has been completely removed.
124
Figure 5.4- Mean worker survival post exposure. Each color represents the mean survival within the five replicates in each of our genetic colonies: red is data from
PWN1-FA16, green is data from PWN6-FA16, and blue denotes the data collected from
PWN8-FA16. We found treatment and colony to have a strong effect on worker mortality post exposure (Kaplan-Meier log rank: P= 0.00). Individuals within the open nest treatment died at a higher than individuals within the closed nest treatment. Suggesting workers within closed nest avoided the cadaver and conidia more effectively than workers inside the open nest.
125
Figure 5.5- Mean conidia concentration per sample. Each colored circle represents one data point, while each color represents a different treatment: purple represents our untreated samples used to determine baseline Beauveria bassiana health (n=17), green represents the samples we placed within a nest with no ants (n=15), red represents the samples we placed within a nest containing ants unable to remove the cadaver for 24 hours (n=15), and blue represents the samples we exposed to workers for 24 hours with the ability to remove our sample from the nest (n=15).The black whisker plots denote the
126 mean ± standard error for each treatment. Significant differences amongst treatments are indicated with letters above the sample sizes. Samples exposed to workers had significantly lower conidia concentrations, than samples we did not expose to conidia
(GLMM: P<0.05).
127
Figure 5.6- Mean conidia germination rate per treatment. Each colored circle represents one data point, while each color represents a different treatment: purple represents our untreated samples used to determine baseline Beauveria bassiana health
(n=17), green represents the samples we placed within a nest with no ants (n=15), red represents the samples we placed within a nest containing ants unable to remove the cadaver for 24 hours (n=15), and blue represents the samples we exposed to workers for
24 hours with the ability to remove our sample from the nest (n=15).The black whisker plots denote the mean ± standard error for each treatment. Significant differences amongst treatments are indicated with letters above the sample sizes. We found samples exposed to workers had significantly lower germination rates than those we did not expose to workers (GLMM: P<0.05).
128
Chapter 6: Discussion and future directions
Ant colonies are composed of highly related individuals in constant contact with one another living under conditions ideal for pathogen infection and development
(Schmid-Hempel, 1998). Although ant colonies are assumed to be constantly affected by parasites, there are limited records of colonies dying off due to disease in the wild (R.G.
Loreto & Hughes, 2016). However, when a worker is infected with a pathogen, the conflict is not solely between the worker and the infection it harbors; the colony also becomes a host (Sherman, Seeley, & Reeve, 1998). I explore how within nest dynamics are affected both by the presence of an infected individual and the infectious agent.
In Chapter 1, I introduced the complex social dynamics within an ant nest, highlighting the dynamic nature between parasite, host, and colony. In Chapter 2, I explored the effects that a specialized behavior-manipulating entomopathogen,
Ophiocordyceps unilateralis sensu lato, has on the social dynamics within the nest of a soil-dwelling carpenter ant, Camponotus castaneus. From my observations, I concluded that infected workers continuously performed social food exchange over the course of fungal development, spent significantly more time outside of the nest, and spent more time closer to the nest entrance. In Chapter 3, I used a generalist entomopathogen,
Beauveria bassiana, as a positive control for my previous observations. Similar to individuals infected with the specialist entomopathogen, workers infected with B. bassiana performed social food exchange over the course of fungal development.
129 However, these infected workers sporadically performed nest vigilance behaviors and spent more time outside of the nest when close to death. From Chapters 2 and 3, I concluded that fungal infections have a limited effect on the social dynamics within the nest.
After finding evidence that suggested healthy workers are unable to detect developing fungal growth within their living nest mates, I became interested in determining if workers are capable of recognizing different fungal development within cadavers. In Chapter 4, I asked if workers from different ant colonies in the same genus but with different life histories are capable of detecting early- or late-stage fungal development within cadavers of their nestmates. Within this experiment, I observed
Camponotus pennsylvanicus and C. castaneus workers to explore the importance of nest ecology and cadaver management. I suggested the soil-dwelling lifestyle of C. castaneus is likely the reason for the species’ ability to detect and removal fungal sources from within the nest. In contrast, workers from the species C. pennsylvanicus, a wood-nesting species, removed cadavers with external fungal growth but did not remove cadavers with internal fungal growth from their nest. This behavior suggests C. pennsylvanicus workers are only capable of detecting immediate disease threats. Workers within these colonies could be neutralizing the disease threat by using the chemical cocktail produced within their poison gland.
In order to test the idea of disease neutralization, I observed how B. bassiana is affected by exposure to workers. In Chapter 5, I tested the efficiency of workers in reducing fungal reproduction. I found workers had a strong effect on conidia
130 concentration and germination rate. These results let me to conclude C. pennsylvanicus workers are highly capable in mitigating the effects of exposure to lethal material.
Within my experiments, I explored how behaviors of C. castaneus and C. pennsylvanicus changed when placed within different disease scenarios. I used extensive video recording, coupled with spatial data collection, observational bouts, and mortality observations. Furthermore, I strived to develop semi-natural nest arenas and collect behavioral data from extensive behavioral observation.
Numerous experiments testing how ant workers react to disease have been done in a simplistic manner. Observations being performed within a petri dish, staged interactions, and short observational bouts have been used extensively throughout the literature (for example: N. Bos et al., 2012; Heinze & Walter, 2010; Konrad et al., 2015; Konrad et al., 2012; Masri & Cremer, 2014; Theis et al., 2015; Simon Tragust, Mitteregger, et al., 2013;
Simon Tragust, Ugelvig, et al., 2013; Ugelvig & Cremer, 2007; Westhus et al., 2014). Such methodology could be ideal for high volumes of publications but does not answer biologically relevant questions. In most cases, these studies suggest their findings have evolutionary bases when the scenarios used for observation are artefactual. The use of longer observation periods and building more natural nests or arenas are challenging, however, these techniques allow for the development of biologically relevant observations. Therefore, I suggest future work should emphasize the use of longer observation periods and semi-natural arenas to ask biologically relevant questions.
The behavioral approach of my dissertation provides a deeper understanding of ant behavior, disease management within a complex society, and a more comprehensive analysis of how a parasite’s life history is affected by the differences in behavior and
131 ecology amongst ant species. Such a novel approach had not been previously attempted. I suggest these approaches could enable us to observe ant colonies within a biologically relevant framework, for example: determining what behaviors are performed outside the nest by infected workers within a complex area, or determining how a colony manages cadavers within their foraging area by increasing the distance between the foraging arena and the nest. Within my thesis, we can see there is still a lot more to be discovered within parasite-host and community interactions. In order to advance our understanding in such complex systems, we must integrate the use of prolonged observations and use multiple perspectives to unravel the answers within.
132 Appendix A: Supplementary material for Chapter 2
Figure A.1- Mortality curves. The comparison of mortality rates across infected, sham, and healthy treatments over the course of 20 days. The initial sample sizes are the following: healthy (n=91, grey line), infected (n=60, black line), and sham/control (n=59, light grey line). We found a significant difference in mortality within the O. unilateralis s. l. treatment (Kaplan-Meier log rank: P<0.05).
133
Figure A.2- Time spent within the nest. Proportion of time spent within the nest while under observation. We performed observations over the course of seven days (3-9 days post-injection), during the daylight hours (0900-1700). Each line represents a different treatment. Black represents the proportion of time focal individuals within the healthy treatment spent inside the nest, while grey represents individuals within sham treatment, and light grey represents individuals within the infected treatment. Infected individuals spent less time within the nest overall (ANOVA: F2,18=223, P<005). The sample sizes for these data can be seen in Table A.1.
134
Day post-injection
Treatment 3 4 5 6 7 8 9
Infected 20 19 19 18 17 15 13
Sham 20 20 19 19 19 19 19
Healthy 31 30 29 29 28 28 28
Table A.1- Sample sizes for percent time spent within the nest. We performed the
observations over the course of seven days (3-9 days post-injection), during the daylight
hours (0900-1700). We collected these data from both subcolonies of genetic colony 3.
Colony Subcolony Aggression Trophallaxis Extended Time in Spatial Mortality trophallaxis nest data Colony 1 1 X X X X Colony 1 2 X X X Colony 2 1 X X X X Colony 2 2 X X X Colony 3 1 X X X X X X Colony 3 2 X X X X X
Table A.2- Colony and subcolony observation use. Within this table, we have marked
(with an “X”) the colonies and subcolonies used for specific observations and analyses.
135 A.1- Pseudocode for trophallaxis and spatial isolation data analysis
As our response variable was a pair-wise event (ants feed each other, and when one ant moves, it can affect the movement of another), the usual assumption of independent observations was violated. We thus used permutation tests for significance in our analysis. A permutation test consists of permuting treatments under the null hypothesis that all treatments have the same effect. The point estimates of parameters from the permuted data are a sample from the sampling distribution of model parameters under the null hypothesis and take into account the intrinsic dependence in the data. Pseudocode for performing this permutation test is as follows.
1. Obtain point estimates for regression parameters from the observed
trophallaxis events or spatial location using “lmer” in the “lme4” package in
R.
2. Randomly permute the treatment labels for all ants, maintaining the number of
ants given each treatment (Healthy, Sham, or Infected).
3. Obtain point estimates for regression parameters using observed trophallaxis
events or spatial location and the permuted treatment labels in “lmer”.
4. Repeat steps #2-3 10,000 times. The result is 10,000 samples from the
sampling distribution of the regression parameters under the null hypothesis
that treatment has no effect.
5. Run permuted data 10,000 times through the general mixed-effect model
created to analyze the complexity of our dataset.
136 6. Obtain beta-hat from each permutation and place in a separate data frame.
7. Used obtained beta-hat data frame to calculate the p-values of interest.
8. Calculate an empirical p-value for the regression parameter estimates from the
original treatment assignments by finding the proportion of the 10,000
permutation regression parameters that have absolute value larger than the
regression parameter estimate from the original data.
137 Appendix B: Supplementary material for Chapter 3
Figure B.1- Nest vigilance categorization. Within the images A and B workers are classified as performing nest vigilance behaviors. The workers within these first two images have their antennae parallel to the entrance walls and are facing the nest exterior.
While images C through E depict individuals that are not performing nest vigilance behaviors.
138
Figure B.2- Healthy Beauveria bassiana growth from cadaver. Within this image we have a Camponotus castaneus cadaver with external extensive hyphal growth and conidia production.
139
Healthy Control Infected 100 90 80
70 60 50 40
Survival rate Survival 30 20 10 0 -4 -3 -2 -1 0 1 2 3 4 Day of observation
Figure B.3- Survival rate per treatment. Individuals injected with B. bassiana died at a higher rate than individuals within the other treatments (Kaplan-Maier: P= 7.79e-08). We collected these data from two colonies, Cast1 and Cast13. Each color represents a treatment: blue represents healthy, red represents controls, and green represents individuals within the infected treatment.
140 Appendix C: Supplementary material for Chapter 4
Figure C.1- Schematic representation of the nest architecture and cadaver treatments used for Camponotus castaneus. We made the nest using plaster, for which we then used a mold to imprint our nest design. (A) Depicts the nest treatments able to remove the introduced cadavers within the first 12 hours of exposure. The ant images within these nests are exemplar samples we introduced into the nest. (B) Depicts the nest treatments unable to remove the introduced cadavers within the first 12 hours of exposure, we closed the nests by using cotton.
141
Figure C.2- Schematic representation of the nest architecture and cadaver treatments used for Camponotus pennsylvanicus. We made the nest using plaster, for which we then used a mold to imprint our nest design. (A) Depicts the nest treatments able to remove the introduced cadavers within the first 12 hours of exposure. The ant images within these nests are exemplar samples we introduced into the nest. (B) Depicts the nest treatments unable to remove the introduced cadavers within the first 12 hours of exposure, we closed the nests by using cotton.
142
Figure C.3- Confirmation of conidia production on external mycosis cadavers. The insert displays image of Beauveria bassiana conidia, while the larger image is a depiction of conidia on hyphal structures emerging from the antennae of a cadaver used within the external mycosis treatment.
143 Appendix D: Supplementary material for Chapter 5
Figure D.1- Before and after photographs of an exemplar cadaver used for the “nest
+ no workers” treatment. Image (A) depicts the average cadaver used for our observations before placing it within an empty nest. Note the extensive external hyphal growth and conidia production on the hyphae. Image (B) depicts the same cadaver after
24 hours of being within an empty nest. The fungal tissue has desiccated and compressed closer to the ant’s cuticle.
144
Figure D.2- Before and after photographs of an exemplar cadaver used for the
“open nest + workers” treatment. Images (A) and (B) illustrate how the average cadaver used for our observation looked like. Note the extensive external hyphal growth and conidia production on the hyphae. Images (C) and (D) portray the same cadaver after
24 hours of exposure to workers within an open nest arena. The cadaver has been broken into pieces and the external hyphal structures have been severely damaged.
145
Figure D.3- Before and after photographs of an exemplar cadaver used for the
“closed nest + workers” treatment. Image (A) depicts fungal development before exposure to a colony unable to remove the cadaver. The cadaver has extensive external hyphal growth and conidia production. Image (B) shows the same cadaver after 24 hours of exposure to workers unable to remove the cadaver from the nest. The gaster of the cadaver has been torn, the hyphal tissue has become closer to the cadaver’s cuticle.
Workers have also used cotton, along with pieces of red modeling clay used to join the semi-natural nest to the foraging arena.
146 References
Abbott, J. (2014). Self-medication in insects: current evidence and future perspectives. Ecological Entomology, 39(3), 273-280. doi: 10.1111/een.12110
Akino, T., Knapp, J. J., Thomas, J. A., & Elmes, G. W. (1999). Chemical mimicry and host specificity in the butterfly Maculinea rebeli, a social parasite of Myrmica ant colonies. Proceedings of the Royal Society B-Biological Sciences, 266(1427), 1419-1426.
Andersen, S. B., Gerritsma, S., Yusah, K. M., Mayntz, D., Hywel-Jones, N. L., Billen, J., . . . Hughes, D. P. (2009). The life of a dead ant: the expression of an adaptive extended phenotype. American Naturalist, 174(3), 424-433. doi: 10.1086/603640
Araújo, J. P. M., & Hughes, D. P. (2016). Diversity of entomopathogenic fungi: which groups conquered the insect body? Advances in Genetics, 94, 1-39. doi: http://dx.doi.org/10.1016/bs.adgen.2016.01.001
Attygalle, A. B., & Morgan, E. D. (1984). Chemicals from the glands of ants. Chemical Society Reviews, 13(3), 245-278. doi: 10.1039/cs9841300245
Aubert, A., & Richard, F. J. (2008). Social management of LPS-induced inflammation in Formica polyctena ants. Brain Behavior and Immunity, 22(6), 833-837. doi: 10.1016/j.bbi.2008.01.010
Baracchi, D., Fadda, A., & Turillazzi, S. (2012). Evidence for antiseptic behaviour towards sick adult bees in honey bee colonies. Journal of Insect Physiology, 58(12), 1589-1596. doi: 10.1016/j.jinsphys.2012.09.014
Bates, D., Maechler, M., Bolker, B. M., & Walker, S. C. (2015). Fitting linear mixed-effects models using lme4. Journal of Statistical Software, 67(1), 1-48.
Beros, S., Jongepier, E., Hagemeier, F., & Foitzik, S. (2015). The parasite's long arm: a tapeworm parasite induces behavioural changes in uninfected group members of its social host. [10.1098/rspb.2015.1473]. Proceedings of the Royal Society of London B: Biological Sciences, 282(1819).
Blum, M. S., Walker, J. R., Callahan, P. S., & Novak, A. F. (1958). Chemical, insecticidal, and antibiotic properties of fire ant venom. Science, 128(3319), 306-307. doi: 10.1126/science.128.3319.306-a
Boomsma, J. J., Schmid-Hempel, P., & Hughes, W. O. H. (2005). Life histories and parasite pressure across the major groups of social insects. Insect evolutionary ecology. Proceedings of the Royal Entomological Society's 22nd Symposium., 139-175. doi: 10.1079/9780851998121.0139
147 Bos, N., Lefevre, T., Jensen, A. B., & d'Ettorre, P. (2012). Sick ants become unsociable. Journal of Evolutionary Biology, 25(2), 342-351. doi: 10.1111/j.1420-9101.2011.02425.x
Bos, N., Sundstrom, L., Fuchs, S., & Freitak, D. (2015). Ants medicate to fight disease. Evolution, 69(11), 2979-2984. doi: 10.1111/evo.12752
Bot, A. N. M., Currie, C. R., Hart, A. G., & Boomsma, J. (2001). Waste management in leaf- cutting ants. Ethology Ecology & Evolution, 13(3), 225-237.
Bouamama, N., Vidal, C., & Fargues, J. (2010). Effects of fluctuating moisture and temperature regimes on the persistence of quiescent conidia of Isaria fumosorosea. Journal of Invertebrate Pathology, 105(2), 139-144. doi: 10.1016/j.jip.2010.05.014
Bourke, A. F. G. (2011). Principles of Social Evolution. New York: Oxford Univ Press.
Bourke, A. F. G., & Franks, N. R. (1995). Social evolution in ants., i-xiii, 1-529.
Bouwma, A. M., Howard, K. J., & Jeanne, R. L. (2005). Parasitism in a social wasp: effect of gregarines on foraging behavior, colony productivity, and adult mortality. Behavioral Ecology and Sociobiology, 59(2), 222-233. doi: 10.1007/s00265-005-0028-5
Bozic, J., & Valentincic, T. (1995). Quantitative-analysis of social grooming behavior of the honey bee Apis mellifera carnica. Apidologie, 26(2), 141-147. doi: 10.1051/apido:19950207
Chapuisat, M. (2010). Social evolution: sick ants face death alone. Current Biology, 20(3), R104- R105. doi: 10.1016/j.cub.2009.12.037
Chouvenc, T., Robert, A., Semon, E., & Bordereau, C. (2012). Burial behaviour by dealates of the termite Pseudacanthotermes spiniger (Termitidae, Macrotermitinae) induced by chemical signals from termite corpses. Insectes Sociaux, 59(1), 119-125. doi: 10.1007/s00040-011-0197-3
Chouvenc, T., & Su, N.-Y. (2012). When subterranean termites challenge the rules of fungal epizootics. Plos One, 7(3). doi: 10.1371/journal.pone.0034484
Chouvenc, T., Su, N.-Y., & Grace, J. K. (2011). Fifty years of attempted biological control of termites - Analysis of a failure. Biological Control, 59(2), 69-82. doi: 10.1016/j.biocontrol.2011.06.015
Cole, L. K., Blum, M. S., & Roncadori, R. W. (1975). Antifungal properties of insect alarm pheromones, citral, 2-heptanone, and 4-methyl-3-heptanone Mycologia, 67(4), 701-708. doi: 10.2307/3758330
Cremer, S., Armitage, S. A. O., & Schmid-Hempel, P. (2007). Social immunity. Current Biology, 17(16), R693-R702. doi: 10.1016/j.cub.2007.06.008 de Bekker, C., Ohm, R. A., Loreto, R. G., Sebastian, A., Albert, I., Merrow, M., . . . Hughes, D. P. (2015). Gene expression during zombie ant biting behavior reflects the complexity
148 underlying fungal parasitic behavioral manipulation. BMC Genomics, 16(1), 620-642. doi: 10.1186/s12864-015-1812-x de Bekker, C., Quevillon, L. E., Smith, P. B., Fleming, K. R., Ghosh, D., Patterson, A. D., & Hughes, D. P. (2014). Species-specific ant brain manipulation by a specialized fungal parasite. BMC Evolutionary Biology, 14. doi: 10.1186/s12862-014-0166-3
De Souza, D. J., Van Vlaenderen, J., Moret, Y., & Lenoir, A. (2008). Immune response affects ant trophallactic behaviour. Journal of Insect Physiology, 54(5), 828-832. doi: 10.1016/j.jinsphys.2008.03.001
Diez, L., Deneubourg, J.-L., & Detrain, C. (2012). Social prophylaxis through distant corpse removal in ants. Naturwissenschaften, 99(10), 833-842. doi: 10.1007/s00114-012-0965-6
Diez, L., Lejeune, P., & Detrain, C. (2014). Keep the nest clean: survival advantages of corpse removal in ants. Biology Letters, 10(7). doi: 10.1098/rsbl.2014.0306
Elizalde, L., & Julia Folgarait, P. (2012). Behavioral strategies of phorid parasitoids and responses of their hosts, the leaf-cutting ants. Journal of Insect Science, 12.
Elmes, G. W., Thomas, J. A., & Wardlaw, J. C. (1991). Larvae of Maculinea rebeli, a large-blue butterfly, and their Myrmica host ants: wild adoption and behaviour in ant-nests. Journal of Zoology, 223, 447-460.
Elmes, G. W., Wardlaw, J. C., & Thomas, J. A. (1991). Larvae of Maculinea rebeli, a large-blue butterfly, and their Myrmica host ants: patterns of caterpillar growth and survival. Journal of Zoology, 224, 79-92.
Erler, S., & Moritz, R. F. A. (2016). Pharmacophagy and pharmacophory: mechanisms of self- medication and disease prevention in the honeybee colony (Apis mellifera). Apidologie, 47(3), 389-411. doi: 10.1007/s13592-015-0400-z
Esponda, F., & Gordon, D. M. (2015). Distributed nestmate recognition in ants. Proceedings of the Royal Society B-Biological Sciences, 282(1806), 42838-42838.
Evans, H. C., Elliot, S. L., & Hughes, D. P. (2011). Hidden diversity behind the zombie-ant fungus Ophiocordyceps unilateralis: four new species described from carpenter ants in Minas Gerais, Brazil. Plos One, 6(3). doi: 10.1371/journal.pone.0017024
Fan, Y., Pereira, R. M., Kilic, E., Casella, G., & Keyhani, N. O. (2012). Pyrokinin beta- neuropeptide affects necrophoretic behavior in fire ants (S. invicta), and expression of beta-NP in a mycoinsecticide increases its virulence. Plos One, 7(1). doi: 10.1371/journal.pone.0026924
Fargues, J., & Luz, C. (2000). Effects of fluctuating moisture and temperature regimes on the infection potential of Beauveria bassiana for Rhodnius prolixus. Journal of Invertebrate Pathology, 75(3), 202-211. doi: 10.1006/jipa.1999.4923
149 Farish, D. J. (1972). The evolutionary implications of qualitative variation in the grooming behaviour of the Hymenoptera (Insecta). Animal Behaviour, 20(4), 662-676. doi: 10.1016/s0003-3472(72)80139-8
Fernandez-Marin, H., Zimmerman, J. K., Rehner, S. A., & Wcislo, W. T. (2006). Active use of the metapleural glands by ants in controlling fungal infection. Proceedings of the Royal Society B-Biological Sciences, 273(1594), 1689-1695. doi: 10.1098/rspb.2006.3492
Foitzik, S., DeHeer, C. J., Hunjan, D. N., & Herbers, J. M. (2001). Coevolution in host-parasite systems: behavioural strategies of slave-making ants and their hosts. Proceedings of the Royal Society B-Biological Sciences, 268(1472), 1139-1146.
Franks, N. R., Bryant, S., Griffiths, R., & Hemerik, L. (1990). Synchronization of behavior within nests of the ant Leptothorax acervorum (Fabricius)- I. Discovering the phenomenon and its relation to the level of starvation. Bulletin of Mathematical Biology, 52(5), 597-612. doi: 10.1007/bf02462100
Fujii, J. K. (1975). Effects of an entomogenous nematode, Neoaplectana carpocapsae weiser, on the formosan subterranean termite, Coptotermes formosanus shiraki, with ecological and biological studies on C. formosanus. Ph.D, University of Hawaii, Egyptian academic journal of biological sciences. (1)
Gordon, D. M. (1989). Dynamics of task switching in Harvester ants. Animal Behaviour, 38, 194- 204. doi: 10.1016/s0003-3472(89)80082-x
Guillem, R. M., Drijfhout, F., & Martin, S. J. (2014). Chemical deception among ant social parasites. Current Zoology, 60(1), 62-75.
Hamilton, C., Lejeune, B. T., & Rosengaus, R. B. (2011). Trophallaxis and prophylaxis: social immunity in the carpenter ant Camponotus pennsylvanicus. Biology Letters, 7(1), 89-92. doi: 10.1098/rsbl.2010.0466
Hamilton, W. D. (1963). Evolution of altruistic behavior. American Naturalist, 97(896), 354-356. doi: 10.1086/497114
Hamilton, W. D. (1964a). Genetical evolution of social behaviour I. Journal of Theoretical Biology, 7(1), 1-16. doi: 10.1016/0022-5193(64)90038-4
Hamilton, W. D. (1964b). Genetical evolution of social behaviour II. Journal of Theoretical Biology, 7(1), 17-52. doi: 10.1016/0022-5193(64)90039-6
Hamilton, W. D. (1987). Kinship, recognition, disease and intelligence: constraints of social evolution. Animal societies. Theories and facts., 81-102.
Heinze, J., & Walter, B. (2010). Moribund ants leave their nests to die in social isolation. Current Biology, 20(3), 249-252. doi: 10.1016/j.cub.2009.12.031
150 Hong, T. D., Gunn, J., Ellis, R. H., Jenkins, N. E., & Moore, D. (2001). The effect of storage environment on the longevity of conidia of Beauveria bassiana. Mycological Research, 105, 597-602. doi: 10.1017/s0953756201004026
Howard, D. F., & Tschinkel, W. R. (1976). Aspects of necrophoric behavior in red imported fire ant, Solenopsis invicta. Behaviour, 56, 157-180. doi: 10.1163/156853976x00334
Hughes, D. P., Andersen, S. B., Hywel-Jones, N. L., Himaman, W., Billen, J., & Boomsma, J. J. (2011). Behavioral mechanisms and morphological symptoms of zombie ants dying from fungal infection. BMC Ecology, 11(13), 11-13. doi: 10.1186/1472-6785-11-13
Hughes, D. P., Pierce, N. E., & Boomsma, J. J. (2008). Social insect symbionts: evolution in homeostatic fortresses. Trends in Ecology & Evolution, 23(12), 672-677. doi: 10.1016/j.tree.2008.07.011
Hölldobler, B., & Engel-Siegel, H. (1984). On the metapleural gland of ants. Psyche (Cambridge), 91(3-4), 201-224. doi: 10.1155/1984/70141
Hölldobler, B., & Wilson, E. O. (1990). The ants. Cambridge, Mass.: Belknap Press of Harvard University Press.
Julian, G. E., & Cahan, S. (1999). Undertaking specialization in the desert leaf-cutter ant Acromyrmex versicolor. Animal Behaviour, 58, 437-442. doi: 10.1006/anbe.1999.1184
Kermarrec, A., Febvay, G., & Decharme, M. (1986). Protection of leaf-cutting ants from biohazards: is there a future for microbiological control? Fire ants and leaf-cutting ants. Biology and management.: Westview Press, Boulder & London.
Kistner, D. H. (1982). The social insect bestiary. Hermann, H. R. (Ed.). Social Insects, Vol. 3. Academic Press, Inc.: New York, N.Y., USA; London, England. Illus, P1-244.
Konrad, M., Grasse, A. V., Tragust, S., & Cremer, S. (2015). Anti-pathogen protection versus survival costs mediated by an ectosymbiont in an ant host. Proceedings of the Royal Society B-Biological Sciences, 282(1799). doi: 10.1098/rspb.2014.1976
Konrad, M., Vyleta, M. L., Theis, F. J., Stock, M., Tragust, S., Klatt, M., . . . Cremer, S. (2012). Social transfer of pathogenic fungus promotes active immunisation in ant colonies. Plos Biology, 10(4). doi: 10.1371/journal.pbio.1001300
Lachaud, J.-P., Lenoir, A., & Witte, V. (2012). Ants and their parasites. Psyche (Cambridge), 2012, 1-5.
Lenth, R. (2016). Least-Squares Means R package lsmeans. Comprehensive R Archive Network, version 2.23.
Li, S., Jin, X., & Chen, J. (2012). Effects of piperidine and piperideine alkaloids from the venom of red imported fire ants, Solenopsis invicta Buren, on Pythium ultimum Trow growth in vitro and the application of piperideine alkaloids to control cucumber damping-off in the greenhouse. Pest Management Science, 68(12), 1546-1552. doi: 10.1002/ps.3337
151
Little, A. E. F., Murakami, T., Mueller, U. G., & Currie, C. R. (2006). Defending against parasites: fungus-growing ants combine specialized behaviours and microbial symbionts to protect their fungus gardens. Biology Letters, 2(1), 12-16. doi: 10.1098/rsbl.2005.0371
Lopez-Riquelme, G. O., & Fanjul-Moles, M. L. (2013). The funeral ways of social insects. Social strategies for corpse disposal. Trends in Entomology, 9, 71-129.
Loreto, R. G., Elliot, S. L., Freitas, M. L. R., Pereira, T. M., & Hughes, D. P. (2014). Long-term disease dynamics for a specialized parasite of ant societies: a field study. Plos One, 9(8). doi: 10.1371/journal.pone.0103516
Loreto, R. G., & Hughes, D. P. (2016). Disease dynamics in ants: a critical review of the ecological relevance of using generalist fungi to study infections in insect societies. Advances in Genetics, 94.
Masri, L., & Cremer, S. (2014). Individual and social immunisation in insects. Trends in Immunology, 35(10), 471-482. doi: 10.1016/j.it.2014.08.005
Mc Cabe, S., Farina, W. M., & Josens, R. B. (2006). Antennation of nectar-receivers encodes colony needs and food-source profitability in the ant Camponotus mus. Insectes Sociaux, 53(3), 356-361. doi: 10.1007/s00040-006-0881-x
Mersch, D. P., Crespi, A., & Keller, L. (2013). Tracking individuals shows spatial fidelity is a key regulator of ant social organization. Science, 340(6136), 1090-1093. doi: 10.1126/science.1234316
Mitra, A., Ramachandran, A., & Gadagkar, R. (2014). Nestmate discrimination in the social wasp Ropalidia marginata: chemical cues and chemosensory mechanism. Animal Behaviour, 88, 113-124. doi: 10.1016/j.anbehav.2013.11.017
Murakami, A. S. N., Nunes, T. M., Desuo, I. C., Shima, S. N., & Mateus, S. (2015). The cuticular hydrocarbons profiles in the colonial recognition of the neotropical eusocial Wasp, Mischocyttarus cassununga (Hymenoptera: Vespidae). Sociobiology, 62(1), 109-115.
Myles, T. G. (2002). Alarm, aggregation, and defense by Reticulitermes flavipes in response to a naturally occurring isolate of Metarhizium anisopliae. Sociobiology, 40(2), 243-255.
Nascimento, D. L., & Nascimento, F. S. (2012). Acceptance threshold Hypothesis is supported by chemical similarity of cuticular hydrocarbons in a stingless bee, Melipona asilvai. Journal of Chemical Ecology, 38(11), 1432-1440. doi: 10.1007/s10886-012-0194-7
Neoh, K.-B., Yeap, B.-K., Tsunoda, K., Yoshimura, T., & Lee, C.-Y. (2012). Do termites avoid carcasses? Behavioral responses depend on the nature of the carcasses. Plos One, 7(4). doi: 10.1371/journal.pone.0036375
Nouvian, M., Reinhard, J., & Giurfa, M. (2016). The defensive response of the honeybee Apis mellifera. Journal of Experimental Biology, 219(22), 3505-3517. doi: 10.1242/jeb.143016
152
Oi, D. H., & Pereira, R. M. (1993). Ant behavior and microbial pathogens (Hymenoptera, Formicidae). Florida Entomologist, 76(1), 63-74. doi: 10.2307/3496014
Oster, G. F., & Wilson, E. O. (1978). Caste and ecology in the social insects. Monographs in Population Biology, i-xv, 1-352.
Otterstatter, M. C., & Thomson, J. D. (2007). Contact networks and transmission of an intestinal pathogen in bumble bee (Bombus impatiens) colonies. Oecologia, 154(2), 411-421. doi: 10.1007/s00442-007-0834-8
Pinter-Wollman, N., Bala, A., Merrell, A., Queirolo, J., Stumpe, M. C., Holmes, S., & Gordon, D. M. (2013). Harvester ants use interactions to regulate forager activation and availability. Animal Behaviour, 86(1), 197-207. doi: 10.1016/j.anbehav.2013.05.012
Pohlert, T. (2016). Calculate pairwise multiple comparisons of mean rank sums R package PMCMR. Comprehensive R Archive Network, version 4.1.
Poinar, G., Jr., & Yanoviak, S. P. (2008). Myrmeconema neotropicum ng, nsp, a new tetradonematid nematode parasitising South American populations of Cephalotes atratus (Hymenoptera : Formicidae), with the discovery of an apparent parasite-induced host morph. Systematic Parasitology, 69(2), 145-153. doi: 10.1007/s11230-007-9125-3
Pontoppidan, M.-B., Himaman, W., Hywel-Jones, N. L., Boomsma, J. J., & Hughes, D. P. (2009). Graveyards on the move: the spatio-temporal distribution of dead Ophiocordyceps- infected ants. Plos One, 4(3). doi: 10.1371/journal.pone.0004835
Poulsen, M., Bot, A. N. M., Nielsen, M. G., & Boomsma, J. J. (2002). Costs and benefits of a general antibiotic defense mechanism in the leaf-cutting ant Acromyrmex octospinosus. 52, 151-157.
Qiu, H.-L., Lu, L.-H., Zalucki, M. P., & He, Y.-R. (2016). Metarhizium anisopliae infection alters feeding and trophallactic behavior in the ant Solenopsis invicta. Journal of Invertebrate Pathology, 138, 24-29. doi: http://dx.doi.org/10.1016/j.jip.2016.05.005
Quevillon, L. E., Hanks, E. M., Bansal, S., & Hughes, D. P. (2015). Social, spatial, and temporal organization in a complex insect society. Scientific Reports, 5, 13393. doi: 10.1038/srep13393
Reber, A., & Chapuisat, M. (2012). Diversity, prevalence and virulence of fungal entomopathogens in colonies of the ant Formica selysi. Insectes Sociaux, 59(2), 231-239. doi: 10.1007/s00040-011-0209-3
Reber, A., Purcell, J., Buechel, S. D., Buri, P., & Chapuisat, M. (2011). The expression and impact of antifungal grooming in ants. Journal of Evolutionary Biology, 24(5), 954-964. doi: 10.1111/j.1420-9101.2011.02230.x
153 Reid, W. M., & Nugara, D. (1961). Description and life cycle of Raillietina georgiensis n. sp. a tapeworm from wild and domestic turkeys. Journal of Parasitology, 47(6), 885-889. doi: 10.2307/3275008
Richard, F. J., & Errard, C. (2009). Hygienic behavior, liquid-foraging, and trophallaxis in the leaf-cutting ants, Acromyrmex subterraneus and Acromyrmex octospinosus. Journal of Insect Science, 9, 9.
Richard, F. J., & Hunt, J. H. (2013). Intracolony chemical communication in social insects. Insectes Sociaux, 60(3), 275-291. doi: 10.1007/s00040-013-0306-6
Rolff, J., & Siva-Jothy, M. T. (2003). Invertebrate ecological immunology. Science, 301(5632), 472-475. doi: 10.1126/science.1080623
Rosengaus, R. B., Maxmen, A. B., Coates, L. E., & Traniello, J. F. A. (1998). Disease resistance: a benefit of sociality in the dampwood termite Zootermopsis angusticollis (Isoptera : Termopsidae). Behavioral Ecology and Sociobiology, 44(2), 125-134. doi: 10.1007/s002650050523
Rosengaus, R. B., Traniello, J. F. A., & Bulmer, M. S. (2011). Ecology, behavior and evolution of disease resistance in termites. Biology of Termites: a Modern Synthesis, 165-191. doi: 10.1007/978-90-481-3977-4_7
Rosengaus, R. B., Traniello, J. F. A., Chen, T., Brown, J. J., & Karp, R. D. (1999). Immunity in a social insect. Naturwissenschaften, 86(12), 588-591. doi: 10.1007/s001140050679
Rosengaus, R. B., Traniello, J. F. A., Lefebvre, M. L., & Maxmen, A. B. (2004). Fungistatic activity of the sternal gland secretion of the dampwood termite Zootermopsis angusticollis. Insectes Sociaux, 51(3), 259-264. doi: 10.1007/s00040-004-0749-x
Ross, K. G., & Matthews, R. W. (1991). The social biology of wasps: Cornell University Press.
Sadd, B. M., & Schmid-Hempel, P. (2006). Insect immunity shows specificity in protection upon secondary pathogen exposure. Current Biology, 16(12), 1206-1210. doi: 10.1016/j.cub.2006.04.047
Sanchez, M. I., Ponton, F., Schmidt-Rhaesa, A., Hughes, D. P., Misse, D., & Thomas, F. (2008). Two steps to suicide in crickets harbouring hairworms. Animal Behaviour, 76, 1621- 1624. doi: 10.1016/j.anbehav.2008.07.018
Scharf, I., Modlmeier, A. P., Beros, S., & Foitzik, S. (2012). Ant societies buffer individual-level effects of parasite infections. American Naturalist, 180(5), 671-683. doi: 10.1086/667894
Schmid-Hempel, P. (1998). Parasites in social insects. Princeton, N.J.: Princeton University Press.
Sendova-Franks, A. B., & Franks, N. R. (1995). Demonstrating new social interactions in ant colonies through randomization tests: separating seeing from believing. Animal Behaviour, 50, 1683-1696. doi: 10.1016/0003-3472(95)80021-2
154
Sharma, K. R., Enzmann, B. L., Schmidt, Y., Moore, D., Jones, G. R., Parker, J., . . . Ray, A. (2015). Cuticular hydrocarbon pheromones for social behavior and their coding in the ant antenna. Cell Reports, 12(8), 1261-1271. doi: 10.1016/j.celrep.2015.07.031
Sherman, P. W., Seeley, T. D., & Reeve, H. K. (1998). Parasites, pathogens, and polyandry in honey bees. American Naturalist, 151(4), 392-396. doi: 10.1086/286127
Silva, M. E., & Diehl-Fleig, E. (1988). Avaliação de diferentes linhagens de fungos entomopatogênicos para controle da formiga Atta sexdens piriventris (Santschi, 1919) (Hymenoptera: Formicidae). . Anais da Sociedade Entomollogica do Brasil, 17(2), 263- 269.
Singer, T. L., & Espelie, K. E. (1992). Social wasps use nest paper hydrocarbons for nestmate recognition. Animal Behaviour, 44(1), 63-68. doi: 10.1016/s0003-3472(05)80755-9
Spivak, M., & Gilliam, M. (1998). Hygienic behaviour of honey bees and its application for control of brood diseases and varroa Part I. Hygienic behaviour and resistance to American foulbrood. Bee World, 79(3), 124-134.
Storey, G. K., Vandermeer, R. K., Boucias, D. G., & McCoy, C. W. (1991). Effect of fire ant (Solinopsis invicta) venom alkaloids on the invitro germination and development of selected entomogenous fungi. Journal of Invertebrate Pathology, 58(1), 88-95. doi: 10.1016/0022-2011(91)90166-n
Sun, Q., Haynes, K. F., & Zhou, X. (2013). Differential undertaking response of a lower termite to congeneric and conspecific corpses. Scientific Reports, 3. doi: 10.1038/srep01650
Theis, F. J., Ugelvig, L. V., Marr, C., & Cremer, S. (2015). Opposing effects of allogrooming on disease transmission in ant societies. Philosophical transactions of the Royal Society of London. Series B, Biological sciences, 370(1669). doi: 10.1098/rstb.2014.0108
Thomas, J. A., Schonrogge, K., & Elmes, G. W. (2005). Specialization and host associations of social parasites of ants. Insect evolutionary ecology. Proceedings of the Royal Entomological Society's 22nd Symposium., 479-518. doi: 10.1079/9780851998121.0479
Thomas, M. B., & Jenkins, N. E. (1997). Effects of temperature on growth of Metarhizium flavoviride and virulence to the variegated grasshopper, Zonocerus variegatus. Mycological Research, 101, 1469-1474. doi: 10.1017/s0953756297004401
Tragust, S. (2016). External immune defence in ant societies (Hymenoptera: Formicidae): the role of antimicrobial venom and metapleural gland secretion. Myrmecological News, 23, 119- 128.
Tragust, S., Mitteregger, B., Barone, V., Konrad, M., Ugelvig, L. V., & Cremer, S. (2013). Ants disinfect fungus-exposed brood by oral uptake and spread of their poison. Current Biology, 23(1), 76-82. doi: 10.1016/j.cub.2012.11.034
155 Tragust, S., Ugelvig, L. V., Chapuisat, M., Heinze, J., & Cremer, S. (2013). Pupal cocoons affect sanitary brood care and limit fungal infections in ant colonies. BMC Evolutionary Biology, 13. doi: 10.1186/1471-2148-13-225
Traniello, J. F. A., Rosengaus, R. B., & Savoie, K. (2002). The development of immunity in a social insect: evidence for the group facilitation of disease resistance. Proceedings of the National Academy of Sciences of the United States of America, 99(10), 6838-6842. doi: 10.1073/pnas.102176599
Trumbo, S. T., & Robinson, G. E. (1997). Learning and task interference by corpse-removal specialists in honey bee colonies. Ethology, 103(11), 966-975.
Turillazzi, S. (2012). The biology of Hover wasps. Berlin: Springer.
Ugelvig, L. V., & Cremer, S. (2007). Social prophylaxis: Group interaction promotes collective immunity in ant colonies. Current Biology, 17(22), 1967-1971. doi: 10.1016/j.cub.2007.10.029
Visscher, P. K. (1983). The honey bee way of death- necrophoric behavior in Apis Mellifera colonies. Animal Behaviour, 31(NOV), 1070-1076. doi: 10.1016/s0003- 3472(83)80014-1
Westhus, C., Ugelvig, L. V., Tourdot, E., Heinze, J., Doums, C., & Cremer, S. (2014). Increased grooming after repeated brood care provides sanitary benefits in a clonal ant. Behavioral Ecology and Sociobiology, 68(10), 1701-1710. doi: 10.1007/s00265-014-1778-8
Wilson, E. O. (1968). Ergonomics of castes in social insects. American Naturalist, 102(923P), 41- &. doi: 10.1086/282522
Wilson, E. O., Durlach, N. I., & Roth, L. M. (1958). Chemical releasers of necrophoric behavior in ants. Psyche, 65((4)), 108-114. doi: 10.1155/1958/69391
Wilson, E. O., & Regnier, F. E. (1971). The evolution of the alarm-defense system in formicine ants. American Naturalist, 105(943), 279-&. doi: 10.1086/282724
Yek, S. H., & Mueller, U. G. (2011). The metapleural gland of ants. [Article]. Biological Reviews, 86(4), 774-791. doi: 10.1111/j.1469-185X.2010.00170.x
Zhukovskaya, M., Yanagawa, A., & Forschler, B. T. (2013). Grooming behavior as a mechanism of insect disease defense. Insects, 4(4), 609-630.
VITA
Emilia Solá Gracia
Education
PhD in Ecology, August 2017
The Pennsylvania State University, USA 2012-2017
B.S. in Ecology and Evolutionary
University of Rochester, USA 2008-2012
Grants and Awards
Fund for Excellence in Graduate Recruitment (Penn State) August 2012
Bunton-Waller Award (Penn State) August 2012
Graduate Research Fellowship (NSF) May 2012
Gates Millennium Scholarship (HSF) May 2008
Publications
Wisenden, B. D., Martinez-Marquez, J. Y., Gracia, E. S., and McEwen, D.C. (2012) High intensity and prevalence of two species of trematode metacercariae in the fathead minnow (Pimephales promelas) with no compromise of minnow anti-predator competence. Journal of Parasitology: August 2012, Vol. 98, No. 4, pp. 722-727.