Fungal spent media affect growth and metabolite production in rhizobia
The inhibitory effects ofTrichoderma spent media against the four rhizobial strains resulted in different extents of inhibition (Figure 1 ). In general, the overall profiles of rhizobial growth response among the spent media obtained after 1 week (1W), 2 weeks (2W), and 3 weeks (3W) of fungal growth were different. In fact, Mantel tests performed on the matrices representing the pairwise inhibition values did not show significant correlations (W1 vs. W2 R=-0.496, p-value=0.92; W1 vs. W3 R=-0.402, p-value=0.92; W2 vs. W3 R=0.618, p-value=0.12), indicating that the fungal spent media from 1W, 2W and 3W exert a different effect on rhizobia strain growth. The analysis of the variance (ANOVA) on the growth of individual strains in response to the different fungal spent media showed that all three weeks discriminate among strains (Supplementary Table S1) . However, 1W spent media had the highest levels of significance, suggesting that here the four fungal species maximize their differential effects in giving rise to rhizobial strains differential growth inhibition.
Regarding the effects on the S. meliloti physiology, as growth and plant-growth-promoting activities, such as auxins, EPS (total and reducing sugars), and biosurfactants production, on the overall (Figure 1 ) we observed effects spanning from growth inhibition to EPS production which were different in relation to both fungal spent media and to the rhizobial strain. The effect detected for total sugars (total EPS) was similar to the growth inhibition effect, while reducing sugars clustered with the effect on biosurfactant production. The effect on biofilm grouped apart from the other phenotypes. A two-way ANOVA (Supplementary Table S2 ) indicated that most of the phenotypes were influenced by both strain, fungus and their interaction, the highest F-values in the interaction being IAA production and growth inhibition. Considering single phenotypes separately (Supplementary Figure S1 ), for growth inhibition (Figure 1, Supplementary Figure S1a ), S. meliloti 1021 and cis -hybrid strains clustered together, while AK83 and BL225C formed a second cluster. This clustering is not surprising since thecis -hybrid and 1021 share about ⅔ of the genome (they differ for the pSymA megaplasmid only) and suggest that the response to spent media is mainly residing on chromosomal loci (and chromid). Concerning fungi,T. gamsii showed the strongest inhibiting effect towards the four strains, while T. tomentosum and T. velutinum had a milder effect. The BL225C strain was generally less inhibited by the four spent media. Regarding auxins production (Figure 1, Supplementary Figure S1b ), differential patterns of production related to either rhizobial strains or Trichoderma species were recorded. In particular, auxin production by 1021 and BL225C S. melilotistrains was more inhibited by T. gamsii, T. harzianum andT. velutinum, while spent media from these fungi had milder effects on the AK83 and cis-hybrid S. meliloti strains. T. tomentosum had the lowest effect on the four strains. Interestingly, the 1021 and the cis -hybrid S. meliloti strains clustered separately, suggesting that the mobilization of pSymA in thecis -hybrid strain could affect fungal spent media-related auxin production. Regarding biosurfactants (Figure 1, Supplementary Figure S1c ), no or very weak production was detected in presence of fungal spent media with 1021 and cis -hybrid S. melilotistrains, while AK83 and BL225C S. meliloti strains showed a different pattern of production under spent media treatment. A pattern related to the combination of fungal species and S. melilotistrains (fungal x rhizobial interaction) was also clear for the total EPS production (Figure 1, Supplementary Figure S1d ) and the biofilm formation on root (Figure 1, Figure S1e ), while reducing sugar production (Figure 1, Supplementary Figure S1e ) only distinguished T. tomentosum spent medium response from those of the other fungi.