The present study provides detailed

The present study provides detailed data on the foraging strategy of an ant that obligatorily exploits an unpredictable food source. To harvest fast-decaying mushrooms E. procera established a complex network of remarkably stable trails, which were used
only temporarily. Subsections of the trail network were abandoned over longer time spans and later followed again with high trail fidelity, leading us to reject our initial hypothesis of a fully flexible trail system consisting only of temporary trails. This hypothesiswas based on the assumption that mushrooms represent a spatiotemporally unpredictable food source. Our detailed, fine-scaled investigation of mushroom growth patterns in a Malayan tropical rainforest revealed, however, that mushroom growth was to some degree predictable. Predictability of mushrooms in space and time depended on the fungal species and on the timescale under consideration. The presence of most mushrooms harvested by the ants appeared unpredictable only when short time periods were
considered. Persisting for only a few days, they appeared to be
randomly distributed in space at any given point in time. Looking at
longer time periods, however, mushroom occurrence was to some
degree predictable because the same species frequently reappeared
in close proximity to its previous growth location within days or
weeks. Consequently, many mushroom species may be better
classified as a periodically regenerating resource with reasonable
spatial but low temporal predictability.
Our detailed examination of E. procera’s food preferences supports
the view of mushrooms being a regenerating food source for
the ants. Exclusively soft-bodied and mainly fast-decaying mushrooms
were harvested; these showed both a higher tendency to
reappear and a higher frequency of regrowth locally. Indeed, there
were almost always mushrooms available at our study site, offering
a reliable food supply to an animal that shows a rather generalistic
mushroom feeding habit, such as E. procera. Besides providing a
sufficient food supply, a general mushroom feeding habit is probably
also beneficial for achieving a balanced diet for an obligate mycophagist since fungal species vary considerably in nutrient
content (Claridge & Trappe, 2005). Support for this assumption
comes from studies of mammalian obligate mycophagists (vole,
Clethrionomys californicus: Maser, Nussbaum, & Trappe, 1978;
squirrel, Glaucomys sabrinus: Maser, Maser, Witt, & Hunt, 1986;
bettong, Bettongia gaimardi: Johnson, 1994), which also routinely
consume a wide variety of mushrooms. The evolution of a rather
general diet in animals exploiting food sources that vary strongly in
occurrence and that are difficult to predict in time and space seems
to be a general trend as indicated by theoretical and experimental
studies (Kephart & Arnold, 1982; Ostfeld & Keesing, 2000;
Raubenheimer & Jones, 2006; Stephens et al., 2007; Yang, Bastow,
Spence, & Wright, 2008).
As outlined above, food characteristics such as the periodical
regrowth of mushrooms we found in this study represent important
evolutionary forces shaping an animal’s foraging strategy. In a
competitive environment, natural selection should favour animals
that evolve patterns of returning regularly to spatially fixed, periodically
renewing food sources such as the mushrooms we studied
(Stephens et al., 2007). Indeed, a variety of animals revisit food
locations with such characteristics on a regular basis (e.g. gorilla,
Gorilla beringei: Watts, 1998; capuchin monkey, Cebus apella nigritus:
Di Bitetti, 2001; rat, Rattus norvegicus: Babb & Crystal, 2005;
pied wagtails, Motacilla alba: Davies & Houston, 1981; hummingbird,
Selasphorus rufus: Healy & Hurly, 2013). Many animals displaying
such systematic revisiting behaviour have well-developed
spatial memory skills (Stephens et al., 2007). Ants, however, are
expected to have less developed idiosyncratic memory capabilities
than higher vertebrates (for notable memory capabilities in ants
see Mangan &Webb, 2012; Sommer, von Beeren, &Wehner, 2008).
Nevertheless, social insects are able to solve complex tasks such as
optimal foraging through cooperation (Hölldobler &Wilson, 2008).
By use of pheromone trails (stigmergy) ants release information in
the environment which can be used by other individuals to create
intelligent collective solutions, without any need of centralized
control (Couzin, 2009; Garnier, Gautrais, & Theraulaz, 2007). In the
ant species we studied, a permanent trail network served as a
shared external colony memory allowing the ants to re-exploit
reappearing mushroom patches quickly without the necessity of
foragers relying on complex idiosyncratic memory. Regarding the
discovery of regrowing mushrooms, the pre-existing trail network
may offer guidance for scouting ants to find former mushroom
growth sites. This hypothesis is supported by two observations:
individual workers sporadically used abandoned trails (C. von
Beeren, S. Lizon à l’Allemand, R. Hashim & V. Witte, personal
observation) and they were regularly detected at former fungal
growth patches, although mushrooms were absent at that time (S.
Lizon à l’Allemand, personal observation).
Comparedwithotherants, E. procera’s trail network shows several
unusual features, which we consider to be adaptations to their unusual
dietary specialization. The extensive network of available trails
is used flexibly by abandoning and reactivating trail sections with
high fidelity. Such dynamic use of pre-existing trails in response to
changes in food availability is exceptional among ants because
abandoned trails are usually not reactivated after food depletion (for
exceptions see Gordon, 1991; Hölldobler, 1976; Hölldobler &Wilson,
2011). Euprenolepis procera’s dynamic use of trail sections can be
regarded as intermediate between a permanent trail use, which is
found in ant species exploiting predictable, long-lasting and stationary
food sources (e.g. Lasius fuliginosus foraging on trophobiont
exudates: Quinet, de Biseau, & Pasteels, 1997), and a highly flexible
trail use, seen in species exploiting randomly appearing, quickly
depleting food sources (e.g. Pheidole oxyops foraging on larger insects:
Czaczkes, Nouvellet, & Ratnieks, 2011). We found that mushrooms
combine both attributes, spatial predictability and short
availability, highlighting the interdependency of foraging and food
resources.We propose that these specific growth characteristics have
placed selection on E. procera’s foraging systemshaping theirunusual
foraging strategy. Permanence (route fidelity over time) can be
considered to be an adaptation to the spatially fixed regrowth of
mushrooms, whereas flexibility (activation/deactivation of trail sections)
allows the ants to respond to the permanently changing temporal
availability of mushrooms in their environment.
As in many ant species the appearance of E. procera’s trail
network is highly dendritic. Such structures are widespread in
nature. Dendritic shapes can be found at various levels of biological
organization (e.g. neuronal cells, amoeboid organisms, growth of
trees). Generally, dendritic networks are expected to cover the
surroundings of a central place efficiently for various purposes, for
example communication among neuronal cells (Gulledge, Kampa,
& Stuart, 2005; Wen & Chklovskii, 2008) or food transport in
slime moulds (Tero et al., 2010) or ants (Hölldobler &Wilson, 1990).
The convergent evolution of dendritic structures in different organisms
and at different organizational levels provides evidence of
its adaptive value. In E. procera, the dendritic trail network probably
facilitates efficient food discovery and harvest, features that seem
to be of particular importance for a quick response to the brief
availability of mushrooms.
An interesting point remains to be discussed regarding the efficiency
of food retrieval: the paths from the nest to the food were
convoluted rather than straight, which may suggest inefficiency.
The observed convolution of trails can be explained only partly by
natural barriers such as trees or other vegetation (see Fig. 2a). We
suggest that although not representing a beeline, convoluted trails
may still constitute the fastest way to a given food source. Euprenolepis
procera workers do not clear their trails from obstacles to
increase foraging efficiency, as other ant species do (Greenaway,
1981; Hölldobler, 1976; Hölldobler & Wilson, 1990, 2011). Instead,
they prefer to move over even objects in the litter such as twigs or
large leaves (C. von Beeren, S. Lizon à l’Allemand, R. Hashim & V.
Witte, personal observation and information from maps). By using
a path through the natural environment with low resistance they
may increase speed and lower the overall energetic costs, similar to
carpenter ants (Loreto et al., 2013).
In summary, the present study shows how a temporally unpredictable,
periodically regrowing food source can be successfully
exploited through collective foraging. The unusual foraging system
of the ant E. procera consists of a stable trail network, which can be
found in many other ants, that is used in a highly flexible way by
deactivating and reactivating trail sections. In thisway, fast-decaying
mushrooms can be detected and exploited reliably, particularly since
we demonstrated that their growth locations are not as unpredictable
as commonly believed. Our findings demonstrate once again
the high potential of social insect societies to develop sophisticated
solutions for utilizing various ecological niches.