a) Baseline aggression
Neither uninfected nor infected M. scabrinodis workers showed any aggression towards dummies with removed CHC. This confirmed the role of CHC-profiles in the discrimination process and demonstrated the suitability of the applied CHC-removal procedure in producing ‘odourless’ dummies for further experimental manipulations.
The application of CHC-extracts originating from either infected or uninfected workers to CHC-removed dummies resulted in responses by alive individuals comparable to corresponding unmanipulated assays, confirming the accuracy of the CHC-transfer method (figure 2): U–I untreated vs U–I extract GLM z = 0.6, p = 0.54 (figure 2b,e); U–U untreated vs U–U extract GLM z = -0.9, p = 0.35 (figure 2a,d); I–I untreated vs I–I extract GLM z = -1.96, p = 0.051 (figure 2h,k); I–U untreated vs I–U extract GLM z = -1.1, p = 0.27 (figure 2g,j).
In unmanipulated assays, the level of aggression was significantly lower when both ants were infected compared to uninfected counterparts (I–I untreated vs U–U untreated GLM z = 3.7, p = 0.0002; figure 2a,h). Generally, infected ants were less aggressive towards infected corpses than healthy workers (GLM z = 2.66, p = 0.007; figure 2b,h), but such differences were not observed when it came to uninfected corpses (GLM z = 1.93, p = 0.06; figure 2a,g). Infected ants also treated infected corpses less aggressively compared to uninfected ones (GLM z = 2.34, p = 0.02; figure 2g,h), while healthy workers did not differentiate between cadavers of different health status (GLM z = 1.39, p = 0.16; figure 2a,b).
b) Manipulation of the CHC-profiles
The aggression of infected workers towards dummies was significantly lower towards dummies supplemented with cuticular extract from infected individuals in comparison to assays with uninfected counterparts (GLM U-U extract vs I-I extract GLM z = 2.94, p = 0.003; figure 2d,k). Contrary to the results of the unmanipulated assays, uninfected ants this time responded with a significantly reduced aggression towards hexane-washed dummies treated with cuticular extract from infected individuals than to uninfected dummies (GLM U–I extract vs U–U extract GLM z = 2.49, p = 0.01; figure 2d,e). No significant differences were detected in any other comparisons: ( I–I extract vs I–U extract; I–U extract vs U–U extract; I–I extract vs U–I extract; see Supplementary results, Table S1 for the exact statistical values).
The addition of n-C23 to uninfected dummies, which simulates the effect of the parasitic fungus on the CHC profile of infected ants (see figure 1.4), resulted in striking changes in the outcome of encounters involving uninfected workers. The level of aggression displayed by healthy ants decreased significantly (GLM U–U+n-C23 vs U–U z = 3.27, p = 0.001; figure 2a,c). In contrast, no differences were observed between the assays when the live counterpart was already infected (GLM I-U vs I–U+n-C23 z = 1.76, p = 0.07; figure 2g,i). Similar results were obtained when instead of untreated dummies, CHC-stripped dummies were used that were treated with CHC-extracts from untreated ants and additional n-C23 (see figure 2).
The supplementation with n-C23 of hexane washed dummies with CHC extract from healthy ants significantly changed the outcome of interactions both in uninfected and infected ants: GLM U–U extract vs U–U extract + n-C23 z = -3.87, p = 0.001 (figure 2d,f); I–U extract vs I–U extract + n-C23 GLM z = 2.13, p = 0.03 (figure 2j,l).
c) Chemical analyses of treatment efficiency
Control extracts from dummies that had been washed five times with hexane contained only traces of four to five CHCs out of 38 being present in non-washed workers (figure S1). These residues were from those compounds with the highest abundance in non-washed ants . The abundance of these major compounds, however, was 10-15 times lower than in the unwashed ants. Therefore, we concluded that the applied method was suitable for obtaining dummies without relevant amounts of cuticular compounds.
The relative abundance of n-C23 in uninfected ants treated with the synthetic alkane was on average 22.4 ± 6.0% (mean ± SD) and did not differ from the relative abundance of n-C23 in untreated infected ants (LM t = 1.87, p = 0.11), thus confirming that the topical application of n-C23 on uninfected ant corpses was successful to achieve the relative abundance of this linear alkane present on infected ants. In untreated, uninfected ants, the relative abundance of n-C23 was only 13.8 ± 1.3% and differed significantly from the relative abundance present on uninfected n-C23-treated ants (LM t = 4.46, p < 0.001).