Camponotusfloridanus Ants Incur a Trade-Off between Phenotypic Development and Pathogen Susceptibility from Their Mutualistic Endosymbiont Blochmannia

Various insects engage in microbial mutualisms in which the reciprocal benefits exceed the costs. Ants of the genus Camponotus benefit from nutrient supplementation by their mutualistic endosymbiotic bacteria, Blochmannia, but suffer a cost in tolerating and regulating the symbiont. This cost suggests that the ants face secondary consequences such as susceptibility to pathogenic infection and transmission. In order to elucidate the symbiont’s effects on development and disease defence, Blochmannia floridanus was reduced in colonies of Camponotus floridanus using antibiotics. Colonies with reduced symbiont levels exhibited workers of smaller body size, smaller colony size, and a lower major-to-minor worker caste ratio, indicating the symbiont’s crucial role in development. Moreover, these ants had decreased cuticular melanisation, yet higher resistance to the entomopathogen Metarhizium brunneum, suggesting that the symbiont reduces the ants’ ability to fight infection, despite the availability of melanin to aid in mounting an immune response. While the benefits of improved growth and development likely drive the mutualism, the symbiont imposes a critical trade-off. The ants’ increased susceptibility to infection exacerbates the danger of pathogen transmission, a significant risk given ants’ social lifestyle. Thus, the results warrant research into potential adaptations of the ants and pathogens that remedy and exploit the described disease vulnerability.


Ant culture
Queenright colonies of C. floridanus were collected from Gainesville, Florida (29.6520° N, 82.3250° W). The colonies were housed in Fluon-lined containers, equipped with a cotton-plugged test tube of water and kept in the laboratory at 25°C with a 12-hour light/dark cycle. Four large colonies each containing >100 workers were selected for the experiment: two colonies were assigned an ordinary diet of Tenibrio molitor larvae and honey water (50% deionized water and 50% honey), and the remaining two colonies were assigned an antibiotic diet of T. molitor larvae and honey water with the antibiotic rifampicin (49.5% deionized water, 49.5% honey, 1% rifampicin). The ants were fed three times a week and maintained on the respective experimental diets for 6 months before melanisation was measured.

Assessment and analysis of cuticular melanisation
Thirty randomly selected ants from each colony were used to assess cuticular melanisation of the thorax and gaster. Ants were photographed under a stereo microscope and the photos analyzed in ImageJ according to the protocol described in the main text. A general linear mixed model was used to determine the fixed effect of diet and morphological location on the log-transformed melanisation level, specifying colony as a random effect. The lmer function and packages lme4 and lmerTest were used to create the model, and backwards model reduction was conducted (p>0.05) to obtain the minimum adequate model. Post hoc pairwise comparisons of diet and morphological location were conducted using least-squared means and Tukey p-value adjustment, employing the lsmeans package. The analysis was completed in R using R-Studio editor.

Results
The interaction between diet and morphological location had a significant effect on ant melanisation (Figure S1; LMM; overall LR χ 2 = 18.116, df= 1, p<0.001). Ants receiving the antibiotic diet had a significantly lighter thorax and gaster compared to the respective areas of ants on the ordinary diet (post host comparisons, p=0.001, p<0.001 respectively). Due the similar effect of diet at both morphological locations we only measured the melanisation of the thorax in the main experiment. Figure S1. Cuticular melanisation of ants with reduced and ordinary symbiont levels. The colour of the thorax and gaster (displayed on a grey scale) of ants on the antibiotic diet (orange dots) compared to those on the ordinary diet (blue dots). Overlaying black boxes indicate the mean ± SE; letters specify significant post hoc groups [α = 0.05].

Symbiont reduction in callow ants and age-dependent melanisation
We tested if the reduction of B. floridanus had an effect on the head width and melanisation of callow minor workers (10 days post-eclosion) and found that while head width decreased, melanisation was not significantly different. Second, we confirmed that cuticular melanisation is agedependent in C. floridanus, with complete melanisation occurring between days 10 and 30 posteclosion.

Ant culture
In addition to the two age groups described in the main text, we assessed the head width and cuticular melanisation of a third age group of minor workers from the same colonies (note: one antibiotic-treated colony was substituted due to low fecundity and thus insufficient sample sizes). This group comprised minor workers developing from larvae to adult on the experimental diets sampled 10 days post-eclosion, at which point they were still callow workers.

Assessment and analysis of cuticular melanisation and head width
Callow workers were randomly selected from each colony (immature-treated: nantibiotic=9, nordinary=9) and the cuticular melanisation and head width were assessed according to the protocols described in the main text. General linear mixed models were used to determine the fixed effect of diet on head width and melanisation level, specifying colony as a random effect. A third model was conducted to determine the fixed effect of age on the melanisation of ants on the ordinary diet, with colony as a random effect. The lmer function and packages lme4 and lmerTest were used to create the models, and backwards model reduction was conducted (p>0.05) to obtain minimum adequate models. Due to re-analysis of the data, p-values of the overall models were adjusted according to the Benjamini-Hochberg procedure. In all three cases, p-value adjustment did not alter the significance of the overall models (Table S1). A post hoc pairwise comparison of age was conducted using leastsquared means and Tukey p-value adjustment, employing the lsmeans package. All analysis was completed in R using R-Studio editor.

Colony growth and worker polymorphism
We tested if the reduction of B. floridanus had an effect on colony growth and worker polymorphism.

Assessment of colony size and worker poymorphism
A census was obtained four months into the experimental diet period, the time required for an effect of the antibiotic treatment. Another census was completed across colonies after seven months of receiving the respective experimental diets, with one of the antibiotic-diet colonies exempt due to an ant escape during the experimental period. The overall number of workers removed for the measurement of head width and melanisation were added to the total number of minors in the census. These two census, allowed us to follow how colony size changed due to the different diets. At the second census, the number of major and minor workers were also obtained, enabling an estimate of worker polymorphism.

Results
Colonies from the respective diets were of approximately similar size four months into the treatment (Table S2; average colony sizes: ordinary diet: 69, antibiotic diet: 65). In contrast, seven months into the experimental diet period, colonies on the ordinary diet had on average more than doubled in size, whereas colonies on the antibiotic diet had stagnated (average colony sizes: ordinary diet: 189, antibiotic diet: 60). Table S2. Colony census obtained four (Census 1) and seven months (Census 2) into the experimental diet period. In Census 2, the number of majors, minors, and minors removed for head width and melanisation measures, were recorded (shown in italics).