Entomopathogenic fungi from the order Entomophthorales are highly specialized, host-specific
and obligatory pathogens, which infect, consume and eventually kill their host insect within a few
days. Established infection can effectively wipe out the majority of a host population. Social
insects, on the other hand, are remarkably efficient at preventing disease, a trait which
necessarily arose together with social organization. In the one known example of social insects, in
this case wood ants of the genus Formica, being attacked by an entomophthoralean fungus –
Pandora formicae, social behaviors provide a game changing component that shapes the
interaction in quite a unique way.
This thesis explores some of the aspects of biology of host-pathogen interaction between red
wood ants, F. polyctena, and the fungus P. formicae. First, the taxonomy of the fungus is studied
and some nomenclatural issues clarified in the light of morphological observations of fungal
features. Some basic studies of life cycle and prevalence of the fungus contributing to the general
knowledge of this system are also completed and for the first time winter survival structures,
resting spores, are found in this species.
Next, various aspects of the interaction with a social insect host are studied. Like a number of
other entomophthoralean fungi, P. formicae manipulates pre-death behavior of its host to secure
favorable position for transmission of actively discharged conidia to new hosts. Before dying,
infected ants climb vegetation and bite down on it, locking their mandibles and securing their
legs around a twig or grass blade, so that they die at an elevated position. However, unlike nonsocial
insects, healthy ants exhibit a behavioral response which helps to limit the disease entering
their colony. The observation that ants actively remove fungus-killed cadavers is documented
and further explored in this thesis. I establish the effect of this behavior on raw numbers of
cadavers present around an ant colony by detailed mapping of colony surroundings for three
subsequent days, twice a day, three times during the season. The pattern of cadaver distribution
is associated with ant density, i.e. fungus killed cadavers stay close to the nest and foraging trails.
This indicates that manipulation is adaptive for the fungus, and that social exclusion of sick
individuals, which has previously been reported in social insects, does not take place in this
interaction. Nonetheless, removal rates are high and prevent the disease from overtaking the
colony. The data also indicate that cadavers that already show fungal outgrowth and produce
deadly spores are removed at a higher rate.
Cadaver removal behavior is further studied in the context of social immunity and response to
disease. Ants normally remove dead nestmates and all other insects from their environment as
prophylactic hygienic measure, and to use as protein source. I used behavioral choice
experiments to assess whether persistent removal of Pandora-killed cadavers is prophylactic, or
a display of a more specific response to a disease threat. Results showed that ants prefer to
remove fungus-killed cadavers rather than uninfected control cadavers killed by freezing. This
effect holds both for cadavers freshly killed by the fungus and not yet infective, and those already
sporulating and highly infective; the effect is however more obvious for sporulating cadavers
which pose a more immediate threat. This is in line with the data on removal obtained from
mapping.
The preference ants show for removal of infected cadavers indicates a detection mechanism. I
compare chemical profiles of cuticular hydrocarbons of three cadaver types: uninfected controls,
infected non-sporulating, and infected sporulating. These three groups can be clearly separated
by discriminant analysis of their cuticular chemical profiles, which suggests that ants might use
this cue to detect disease.
Lastly, a gene expression study of the fungus in two final stages of infection was conducted,
including: soon after host death, but before breaching the barrier of cuticle, and during
outgrowth on ant surface and sporulation. Expression of many known pathogenicity-related
genes: proteases, lipases and chitinases related with host breakdown, and many genes related to
morphological and developmental reorganization, was detected. Differences in expression of
these genes mark this developmental switch, regulation of serine proteases being a clear
example.
Results obtained in this PhD project with a variety of classical natural history, and state-of-theart
molecular methods significantly contribute to the knowledge of this unique, but understudied
host-pathogen system, and lay a foundation for its further exploration. With growing interest to
study host manipulating parasites, as well as specialized pathogens of social insects, this system
is especially attractive, even more so because it’s easily available to study here in Denmark.