In animals, group living comes at the cost of increased pathogen exposure. In kin groups, social immune behaviors offset that cost and reach their most complex expression in eusocial insect societies. In the nests of these societies, collective social behaviors can modify patterns of individual interactions across space, reducing the ability of pathogens to reach the reproductive core of the colony (organizational immunity). To be effective, these behaviors must separate infected and uninfected individuals, implying that the efficacy of social immune behaviors may depend upon nest structure. The role of nest space has received little attention, and most knowledge of social immune behavior in social insects is based on the study of generalist entomopathogenic fungi. We examine the social immune behaviors involved in the interaction between the supercolonial, invasive tawny crazy ant (Nylanderia fulva), and its specialist, intracellular, microsporidian pathogen Myrmecomorba nylanderiae, to ask how nest structure influences social immunity. By manipulating nest structure, we demonstrate that preventing pathogen transmission to the colony core requires a multi-chambered nest. Without which, social immune function was lost, and disease transmission was universal. To understand how nest space enhances social immune efficacy, we first confirm that workers within tawny crazy ant nests form spatially and behaviorally segregated social sub-networks. We then find that infected ants introduced into the colony core migrate to the colony periphery, while uninfected ants do not. Behavioral tests indicate that, despite the infection being internal, uninfected ants can detect the infection status of a worker; thus, behaviors enforcing spatial segregation could be triggered by either party. Additionally, infected ants alter the behavioral tasks they perform, assuming more corpse removal tasks, particularly infected corpse removal, and reducing their efforts in foraging and brood care. With some exceptions, the social immune behaviors expressed by this supercolonial ant in response to microsporidian infection correspond to immune defense behaviors employed to defend against generalist entomopathogenic fungi. These behaviors appear to be conserved, generalized responses to pathogen infection among social insects.

Methods

We test whether nest structure alters immune function. Then, we examine the mechanisms that contribute to this effect. First, we evaluate whether workers from uninfected colony fragments occupy spatially distinct social networks shortly after moving into a new nest. Second, we analyze how these spatially defined subnetworks differ in interaction types and task profiles. Third, we test whether infection status changes how workers segregate amongst these subnetworks. Fourth, we assess whether infection influences the interactions workers engage in or the tasks they perform, and finally, we evaluate whether infection status affects how aggressively workers interact.

Infection Assessment: Colony Fragments

The infection status of colony fragments was determined by homogenizing groups of 20 workers in 300 µl of DI H2O, fixing and staining a 0.3 µl aliquot of the tissue homogenate, and counting the Type II DK M. nylanderiae spores present [1].  It is not possible to artificially infect individual adult workers, as they acquire the infection as larvae, nor can we assess the infection status of individual workers non-destructively. Introduced ants were assigned the infection status of their natal colony fragments.  Spore counts of individuals from colony fragments with similar spore burdens show consistently high infection prevalences, with 85 to 95% of workers spore-positive with variable infection intensities. A fraction of the infected workers used in assays were likely uninfected or had low-level infections. Therefore, tests herein may underestimate differences between infected and uninfected workers.

Multi-chamber nest-array

We used a multi-chambered nest array to test how nest structure affects social immune function; to characterize both the social networks of colony fragments absent introduced individuals, as well as the behaviors and tasks assumed by infected and uninfected ants introduced into the arrays; and finally to evaluate how infected and uninfected ants assort spatially. Nest-arrays provide a multi-chamber structure allowing the ants to sort themselves spatially while keeping the queen and brood in the chamber farthest from the outside environment.

Influence of nest structure on infection transmission

To see if nest structure affects whether a colony fragment can resist infection, we introduced infected ants into nest-arrays and simple nest boxes. Simple boxes (24 x 14 x 9 cm) were translucent plastic with a water tube for the queen and brood. Nest-arrays had 396 cm² of floor area, and nest boxes had 336 cm². In both, the ants clustered in a small part of the nest area, making inter-individual spacing insensitive to slight differences in nest area.

Uninfected colony fragments, with 600 workers, two queens, and 1/8 cm³ of brood (including eggs and larvae, but excluding pupae), were housed in nest-arrays or simple nest boxes. Ants housed within each were termed ‘resident,’ while introduced ants were ‘aliens.’  Aliens originated from local-supercolonies far from the collection site of resident fragments. Three treatments were applied: 1) infected-alien workers introduced into the outer chamber of the nest array; 2) infected-aliens introduced into the queen chamber; and 3) infected-aliens introduced into a simple nest box. Pupae were collected weekly for infection testing. All available pupae were harvested and tested to confirm infections arose from direct transmission by introduced ants.

Social networks of resident ants in nest-arrays

To quantify this social network in tawny crazy ant colony fragments, we introduced a colony fragment (resident fragment) consisting of 350 workers, 1/8 cm³ of mixed-stage brood, and two queens into the queen chamber of a nest-array. This composition was consistent across all subsequent experiments using the nest array. After 72 hours for the resident fragment to spatially assort, 35 workers were removed from both the peripheral and queen chambers. Workers were identified as Peripheral Chamber Cohort (PCC) or Queen Chamber Cohort (QCC) workers by applying a drop of colored nail polish to their head above the eyes while ants were briefly anesthetized with CO₂. Pairwise interactions were classified as bump (ants meet and part), antennate (antennal tap for over one second), trophallaxis (regurgitative food sharing), or allogrooming (ants use mouthparts to groom the thorax or abdomen of another ant). These categories were grouped by disease transmission risk (the potential to ingest microsporidian spores): “high-risk" (trophallaxis, allogrooming) and “low-risk" (bump, antennate). Interactions between marked ants were categorized by cohorts: QCC-QCC, QCC-PCC, PCC-PCC. Data were collected across seven nest-arrays. Replicates were defined as the sum of each interaction category observed in a 3-minute focal survey (N = 162 surveys). In a second three-minute survey, we recorded the cohort identity of ants performing tasks. Tasks included corpse removal – carrying dead workers, brood care – touching brood with mouthparts or antennae, and foraging – feeding at either the sucrose tube or cricket. Finally, the number and cohort composition of all groups of ants in the chamber was tallied. Groups were ants in proximity, typically touching, that were not moving or interacting. Replicates were surveys within arenas. Only surveys where specific tasks were observed were included in the analyses: corpse removal (N = 54), brood care (N = 92), foraging (N = 96).

Dispersal of infected and uninfected ants introduced into nest-arrays

To understand how nest structure influences disease spread, we tested whether worker spatial distribution varies with worker source, point of introduction into the nest-array, or worker infection status.  Resident ants (workers, brood, and queens) were introduced into the nest-array queen chamber 72 hours before the introduction of experimental ants. Three types of experimental ants were introduced: 1) nestmates (from the resident, uninfected ant colony fragment), 2) uninfected-aliens, and 3) infected-aliens. Thirty-five experimental ants were introduced into either the queen chamber or the peripheral chamber, making a total of six experimental groups (e.g., nestmates introduced into the peripheral chamber etc.). All experimental ants were color coded with nail polish in the manner previously described.  Typically, two groups were introduced into each array: infected or uninfected alien ants into either the peripheral chamber or the queen chamber, and nestmates into both chambers. In some replicate arrays, nestmates were only introduced into the chamber where alien ants were introduced. Arrays with nestmates introduced into both chambers were used to evaluate the effect of introduction location on the final position of workers, independent of infection status or ant source. The ants were allowed to integrate into the Nest-array for four days, after which the array was disassembled, restricting ants to the chambers where they were located. The final positions of the ants were recorded. To standardize the influence of feeding on ant distribution, fragments were fed a dead cricket 30 minutes before disassembly.  For each array, the response of the experimental group was the average distance of workers from the queen, measured in ‘chamber steps’, with each step represented by a chamber and the connecting nest tube before it.

Behavioral profiles of infected and uninfected workers introduced into nest-arrays

We conducted an additional nest-array experiment, to determine how infection status affects social tasks performed. These followed earlier methods for characterizing social networks of resident ants, with the following exceptions. Introduced, marked ants were from alien colony fragments that were either infected with M. nylanderiae or uninfected. Seventy marked workers (35 infected and 35 uninfected) were introduced into the queen chamber of each array. Data came from 12 nest-arrays, with replicates defined as surveys within each arena. Only surveys in which a specific task was observed were included (corpse removal: N = 112, brood care: N = 240, foraging: N = 238).

Interactions Between Infected and Uninfected Ants

We performed aggression assays to test if infection status affects behaviors during interactions. It involved conflicts between five workers from each nest in non-neutral arenas. Except for the non-neutral arena, the methods followed Roulston et al. 2003. Seventy-two hours before the assay, a 5 cm diameter petri dish lid was placed in an uninfected N. fulva nest box to collect nest odors. A 5 cm plastic ring with 1 cm high walls, painted with Fluon^TM to prevent ants from climbing out, was placed inside the lid. Five resident ants were placed into the arena via a paintbrush, and five test colony ants were introduced by placing them in glass vials and inverting the vial over the arena. During observation, the nest box was set on a Vivosun^TM heat mat maintained at 26 °C. Interactions were surveyed at 1, 3, 5, and 7 minutes for 30 seconds, and rated on a seven-point scale; scores of 4 (biting) or greater indicated aggression (adapted from: LeBrun et al. 2019). The average interaction score characterized the arena outcome. Test ants were either nestmates isolated from the resident nest for 24 hours (Isolated Nestmates), uninfected ants from a different local-supercolony (Uninfected-aliens), or workers from an infected colony fragment, from a different local-supercolony (Infected-aliens). Three arenas were run for each pairing, and the mean score from these was used as the response variable. Observers were blind to the origin or infection status of the introduced ants.

1. LeBrun, E.G., K. Ottens, and L.E. Gilbert, *Intracolony transmission of the microsporidian pathogen (Myrmecomorba nylanderiae) and its impact upon the growth of tawny crazy ant (*Nylanderia fulva) colonies. Journal of Applied Entomology, 2017. 142: p. 114-124.