Extended Data Fig. 6: Assays of spore adherence and dispersal.
To compare dispersal of sporulating and non-sporulating strains, individuals of F. candida (n=12) were fed for 3 days with developed mycelium of S. coelicolor strain M145 or its congenic non-sporulating whiG mutant J2400. CFUs that could be washed off (0.05% Tween-20) from individual animals were quantified by plating on SFM agar. Significantly more spores than hyphae had adhered to the springtail body (a, two-tailed Mann Whitney test, P<0.001 **). To monitor the release of viable cells by defaecation, each washed springtail was allowed to roam for 20 h on MM agar before colonies were allowed to develop. Significantly more CFU were excreted from springtails that had been feeding on the sporulating Streptomyces strain (b, two-tailed Mann Whitney test, P<0.001 **). The boxplots show the medians, lower- and upper quartiles as well as the span of data points (smallest to largest). In a dispersal assay (c), S. coelicolor strains J2192/pIJ10646 (geoA mibAB mutant with geoA and mibAB genes provided in trans on integrated plasmid) and J2192/pIJ10770 (geoA mibAB mutant with empty vector) were allowed to develop and sporulate (5 days) in small spots in center of MM agar plates. Two springtails (F. candida) (n=13) were released for 2 h on each plate. Animals were then removed, plates incubated, and Streptomyces colonies counted. The presence of geoA and mibAB genes (strain J2192/pIJ10646) resulted in significantly more CFU being dispersed by springtails (c, two-tailed Mann Whitney test, P=0.022 *). Error bars show the standard error of the mean. Volatiles from cultures of these strains were collected for 24 h on an air filter, eluted with hexane, and analysed by GC-MS. Chromatograms (d) show that geosmin and 2-MIB were present in strain J2192/pIJ10646 but not detectable (n.d.) in the non-complemented J2192/pIJ10770 strain. The analyses were independently repeated three times with similar results.
