Between-mound sharing and colony genetic structure
All treated mounds showed elevated \(\delta\)15N values, indicating that our methods were successful in enriching the isotope values of individual mounds. Counter to our expectations, treated mounds shared very little with neighboring mounds, regardless of social form (Fig 3) and within-mound relatedness between workers (Appendix S5). Six of the 57 untreated mounds showed evidence of sharing with the treated mound (three monogyne and three polygyne). Sharing was independent of the social form of the treated mound (χ2 = 0.0091, df = 1, P = 0.924), the social form of the untreated mound (χ2 = 0.0001, df = 1,P = 0.992), and by the interaction between these variables (χ2 = 0.0061, df = 1, P = 0.938). Moreover, sharing was independent of within-mound relatedness between workers in the treated mound (χ2 = 0.0681, df = 1, P = 0.794), by relatedness between workers in the untreated mound (χ2 = 0.7718, df = 1, P = 0.380), and by the interaction between these variables (χ2 = 0.000, df = 1, P = 1.000).
There was a significant effect of distance on whether or not sharing was detected in untreated mounds (χ2 = 10.0858, df = 1,P = 0.001). All untreated mounds with elevated\(\delta\)15N values were within 5 m of the treated mound (Fig 4a). However, not all mounds within 5 m of the treated mound shared (Appendix S6), indicating that distance was not the only component influencing sharing between mounds. There were 13 untreated mounds within 5 m from the treated mound that did not share with the treated mound, one of which was only 0.4 m from the treated mound. Sharing between mounds did not vary by site (χ2 = 2.0408, df = 5, P = 0.843).
Treated mounds were more genetically similar to the untreated mounds with which they shared compared to those with which they did not share. Pairwise F ST values between the treated and untreated mounds were significantly lower on average and near zero in mounds that shared (mean and standard errors: 0.028 ± 0.015) than in mounds that did not share (0.228 ± 0.021; χ2 = 14.5562, df = 1, P < 0.001; Fig 4b) whereF STs were equivalent toF STs between different colonies. Likewise, untreated mounds that shared could not be significantly genetically differentiated from the treated mound according to the log-likelihood (G)-based test of differentiation. There was no significant interaction between distance and pairwise F ST values (χ2 = 0.3395, df = 1, P = 0.560). PairwiseF ST between polygyne mounds within each site was lower than between monogyne mounds (mean and standard errors of 0.144 ± 0.008 and 0.356 ± 0.010 respectively), but there was no significant interaction between pairwise F ST and social form (χ2 = 0.0000, df = 3, P = 1.000) on sharing between mounds.