Within-colony sharing
To evaluate within-colony sharing, we conducted a laboratory experiment testing polygyne worker discrimination between larvae that are close siblings and larvae that are not siblings (Fig 2). To do this, we established six single-lineage experimental colonies by collecting mated polygyne queens following mating flights. Fire ant queens are typically singly mated (Ross, 1993; Ross & Fletcher, 1985; but see Fritz, Vander Meer, & Preston, 2006; Lawson, Vander Meer, & Shoemaker, 2012), so by maintaining colonies with an individual queen (rather than several queens), we were able to test worker discernment between brood that are close siblings (i.e., from the same mother) and brood that are not close siblings (i.e., from a different mother).
All colonies were kept in standardized laboratory conditions for at least two years to ensure they were large enough to be divided into smaller experimental colonies. Colonies were confirmed as the polygyne social form by screening workers for the presence of theGp-9b allele using the same methods as described above (see Genetic analysis in the methods section). Once incipient colonies were large enough, we used food dye to label the brood of each colony. We dyed brood by giving workers two separate tubes of 15 mL of water and 15 mL of artificial nectar each containing 0.9 ml of food coloring (McCormick® Food Colors & Egg Dye, McCormick & Company, Inc., 18 Loveton Circle, Sparks, MD). We gave three colonies (colonies A, B, and C) yellow food coloring and three colonies (colonies D, E, and F) green food coloring (Table 1; Fig 2). Colors were randomly assigned. During this six-day period of brood dyeing, we did not give the ants any proteinaceous food so that dye would be highly visible in the guts of larvae.
Next, we created experimental colonies by combining 0.1g of workers (~120 workers) with 50 larvae from the same natal colony as the workers (i.e., close siblings) and 50 larvae from a different colony (i.e., not siblings; see Table 1 for complete family combinations). Not all permutations of families were logistically possible in this experiment, so only larval combinations of different colors were combined so that all possible two-color combinations were created (Table 1). In all, there were 18 experimental colonies.
To quantify the feeding of larvae by workers, experimental colonies were given 7.5 ml artificial nectar containing 0.2 g of non-toxic, fluorescent dye (DFDRY-C0 UV Dye from Risk Reactor, 2676 S. Grand Ave., Santa Ana, CA) for 18 hours. After 18 hours, we recorded the number of larvae remaining from each family and used a black light to count the number of larvae from each family fed the fluorescent dye. In order to ensure accurate results for potentially variable behaviors, data for experimental colonies were averaged across the three iterations of this experiment. This allowed us to remove within colony temporal variation and estimate the general behaviors of each experimental colony instead of only looking at a single snapshot of their behavior.
All data were analyzed using JMP® 9. Percentage data were arcsine-square-root transformed. All graphs were produced with untransformed data. A more detailed description of the methods can be found in Appendix S4.