Cheating/infection experiments. Two-way interaction graph illustrating the impact of experimental infection and cheating ants or mutualistic fungi on the garden biomass in two types of fungus-growing ant colonies. A) Trachymyrmex cf.zeteki sub-colonies (n = 40), treated with crossing the presence/absence of fungus garden infection with the presence/absence of cheating ants, shows significantly higher loss of garden biomass when exposed to cheating and infection simultaneously than in all other treatments (2-way ANOVA p < 0.001 df = 3). B) Atta colombica sub-colonies (n = 80), crossing the presence absence of fungus garden infection with the presence/absence of cheating fungi, shows a significantly greater loss in garden biomass when cheating and infection are experienced simultaneously than in all other treatments (2-way ANOVA p < 0.001 df = 3). Non-parallel lines shown in 'A' and 'B' illustrate significance (2-way ANOVA analyses) of the interaction between infection and cheating treatments.