Data analysis
We used logistic regression to determine the effects of spatial
distance, genetic differentiation (using pairwiseFST values), social form, within-mound
relatedness coefficients between workers, and site on whether or not
untreated mounds shared with the treated mound. To do this, we
constructed generalized linear models with a binomial distribution using
the glm function in base R statistical software v3.6.1 (R Core
Team, 2019). Distance from the treated mound, pairwise
FST values (compared between the treated and untreated
mounds), social form of both treated and untreated mounds (i.e.,
monogyne or polygyne), within-mound relatedness coefficients between
workers in both treated and untreated mounds, and site were treated as
independent variables. The sharing status of the untreated mounds (i.e.,
“shared with the treated mound” or “did not share with the treated
mound”) was the dependent, binary variable. Mounds that were identified
as having shared with the treated mound had\(\delta\)15N values greater than 20‰, as these values
were far higher than any natural abundance isotope values observed at
our field sites (mean natural abundance \(\delta\)15N
values before tracer treatment: 5.00‰ ± 0.15‰). Untreated mounds could
only have attained \(\delta\)15N values greater than
20‰ by freely exchanging workers and/or resources with the treated
mound. All other mounds were designated as “did not share with the
treated mound.” All plots were generated using ggplot2 (Wickham,
2016).