Eusocial insect colonies can function like a single organism, a trait sometimes referred to as “superorganism”. Although superorganisms greatly benefit from division of labor, they may be susceptible to infectious diseases due to increased opportunities for pathogen transmission. The mechanisms behind insect societies’ ability to successfully defend against infectious diseases are of significant interest in evolutionary biology. Next to their innate immune systems, eusocial insects such as termites boast sophisticated collective defences, termed social immunity, which are thought to have evolved in response to selective pressure from pathogens. Social immunity comprises a multi-layer assembly of physiological and behavioural adaptations which effectively supress the probability of exposure and transmission of infectious diseases within a colony. Although a significant body of research has explored the broad repertoire and effectiveness of termite collective defences, the underlying mechanisms of termite social immunity remain largely unexplored. This thesis exploits termite-pathogen step-wise infection dynamics to investigate the underlying mechanisms of collective defence against different entomopathogenic agents, as well as explore the link between innate defences and collective behaviours to offset increased disease pressure. The thesis is primarily based on research conducted in the eastern subterranean termite Reticulitermes flavipes and the entomopathogenic fungus Metarhizium anisopliae; a natural host–pathogen system. In Chapter I, we seek to explore the underpinnings of the “care-kill dichotomy” in termite social immunity. The care-kill switch represents a generalized social immune response found in many advanced insect societies, whereby compromised nestmates are cared for when possible but sacrificed if necessary. In the case of fungal infection, we ask whether termites can detect and respond to infected individuals when pathogens (or infectious conidia) are absent from their cuticle; the rationale being to experimentally manipulate the type of cues that are presented to interacting nestmates. In this chapter, we sought to identify potential triggers for the switch from sanitary care to elimination defence (cannibalism), which represents the so-called care-kill transition. We sought to demonstrate how R. flavipes termites detect and respond to internal M. anisopliae infection at different stages of infection progression, as well as test the effect of pathogen viability on social immunity. By injecting fungal blastospores directly into the hemocoel of individuals, we removed the pathogen as a direct cue that could be detected by responding nestmates. Injection, regardless of blastospore viability, led to slightly increased rates of grooming, but also rapid transition to cannibalism (even at early stages of an internal infection), particularly when termites became visibly moribund following injection with viable blastospores. Surprisingly, however, cannibalism was still observed when termites were injected with dead blastospores and were not terminally ill, indicating that the threshold at which elimination behaviour is triggered can be reached at a very early stage during internal M. anisopliae infection, before viability or even terminal disease status is known. The faster cannibalism response to viable blastospore-injected termites could be due to the active synthesis of virulence factors from the pathogen such as destruxins, although this remains to be tested. Since termites may be expected to communicate with their nestmates via chemical compounds called cuticular hydrocarbons (CHCs), alterations in the CHCs profile associated with internal fungal pathogen presence, could represent important signals for responding nestmates. We used gas chromatography mass spectrometry (GC–MS) in the second part of Chapter I to identify cues associated with disease status and explore potential CHC signal(s) that may be responsible for triggering elimination behaviours. We found that CHC profiles were significantly altered in individuals injected with viable but not dead blastospores, and at 15 but not at 12 hours post infection. More specifically, we detected significant increases in four exclusively methyl-branched CHCs 15 hours after injection with viable blastospores compared to control-injected individuals, which we speculate could be correlated with the advanced state of moribundity of these challenged individuals, and which may contribute as possible chemical cues triggering the high levels of observed cannibalism. Although a direct link between the identified CHC compounds and altered social immune behaviour remain to be tested, these data suggest that termites could employ chemical signals provoked by early internal immune activation to trigger cannibalism. In Chapter II we expand on the role of social immunity against potentially novel infection threats, as well as further explore the conditions that trigger specific behavioural defences. To understand the interaction between R. flavipes and the non-native entomopathogenic bacterium Pseudomonas entomophila and compare termites’ collective defences between a native fungal entomopathogen and a non-native bacterium as well as characterize associated cuticular hydrocarbon (CHC) changes. We injected termites with different doses of either viable / dead P. entomophila or viable / dead M. anisopliae blastospores. As expected, injection regardless of severity and pathogen type led to slightly increased rates of grooming but rapid transition to cannibalism. Cannibalism was particularly evident when infected individuals were injected with high doses of viable P. entomophila and M. anisopliae blastospores (causing 100% mortality). Such individuals showed external signs of disease and were close to death, but as described in Chapter I, cannibalism was still exhibited following injection with dead blastospores, which elicited only 40% mortality. Interestingly, injection with an equivalent dose of viable P. entomophila (causing 40% mortality) did not elicit similarly elevated amounts of cannibalism, suggesting that triggers stimulating elimination behaviours may be pathogen-specific in this termite. We hypothesize that termites may have evolved greater sensitivity to fungal versus non-native bacterial infection, due to the likely long evolutionary association with the former. These results nevertheless show that collective defences of R. flavipes are effective against both fungal and bacterial challenges, with elimination behaviours being transferable to diverse and potentially novel infection threats (like P. entomophila). Analysis of CHC profiles from termites injected with different bacterial and fungal doses revealed unique patterns of CHCs, with a discriminant analysis showing a particularly distinct profile in termites injected with viable blastospores − treatment associated with the strongest cannibalism response. In Chapter III, we examined the inhibition effect of the external antifungal activity of the Gram-negative binding protein 2 (GNBP-2) and collective behaviours after exposure of single individuals to M. anisopliae. Termites may depend on the innate immune system for defence against pathogens. Therefore, immune effectors have been co-opted from an internal role to an external role to prevent infection from entomopathogenic fungi that can evade innate immune defences after penetration of the cuticle. Termicins and GNBP-2 associated β-1,3- glucanase activity are involved in external defence against M. anisopliae. The effectiveness of this external defence strategy likely depends on or is bolstered by collective behaviours. Through the suppression by an inhibitor (D-d-gluconolactone (GDL)) of the termite GNBP-2 β- 1,3-glucanase activity that is capable of degrading entomopathogenic fungi, we found that collective defences such as grooming are not triggered, but instead cannibalistic behaviours are reduced. This suggests that the internal immune system or the use of antimicrobial secretions may be linked to certain collective immune behaviours in termites. Understanding the molecular basis of termite defence mechanisms may also be relevant for the development of sustainable control strategies against pest termite species. Finally, on the side of the pathogen, we require efficient transformation methods for entomopathogenic fungi to develop essential molecular tools for elucidating the function of genes involved in fungus-insect interactions. For example, by inserting or deleting genetic elements in the genome of the strain of interest, it is possible to modify the expression of targeted endogenous genes, and thereby experimentally test their role in the pathogen’s infection strategy, and how they may influence host fitness, immunity, and ultimately, social immune traits. However, developing methods to deliver foreign nucleic acid into fungal cells represents a major stumbling block in fungal genetics. In Chapter IV we address this issue by testing suitable selection markers: glufosinate ammonium (GFS) and chlorimuron ethyl (CME) for transformation of M. anisopliae with green fluorescence protein gene (gfp). Since M. anisopliae can produce blastospores through yeast-like budding in liquid culture, as well as being thin-walled and unicellular, an efficient blastospore-based transformation system was developed for the introduction of “bar-gfp” and “sur-gfp” constructs using the LiAc/ssDNA/PEG method. These genetic constructs conferred resistance to GFS and CME, respectively. We efficiently achieved integration of these genes into the fungal genome, resulting in resistant transformants, MA-0001 (baR-gfp) and MA-0002 (suR-gfp), which expressed high levels of green fluorescence in conidia, hyphae, blastospores, as well as on termite cadavers after injection of blastospores within the host hemolymph. The generation of these stable MA-0001 and MA-0002 strains allows us to monitor the internal infection process that begins after the fungus penetrates the cuticle and proliferates inside the termite hemolymph as hyphal bodies. This research represents a proof-of-principle that the genetic manipulation of M. anisopliae for studying gene function and for elucidating the relevant factors for pathogenic interactions with insects, is feasible. Targets for future work could include the generation of strains lacking virulence factors such as destruxins and examining the individual and social immune consequences of infection with genetically modified fungi. The present work highlights the importance of the effective innate immune responses in addition to physiological and behavioural defences in the termite R. flavipes, and how these are mediated by precise chemical communication which also contribute to social immunity. This allows R. flavipes to be less susceptible to pathogen infections and facilitates its evolutionary success.