Many insects host vertically-transmitted microbes, which can confer benefits to their hosts but are costly to maintain and regulate. A key feature of these symbioses is variation: for example, symbiont density can vary among host and symbiont genotypes. However, the evolutionary forces maintaining this variation remain unclear. We studied variation in symbiont density using the pea aphid (Acyrthosiphon pisum) and the bacterium Regiella insecticola, a symbiont that can protect its host against fungal pathogens. We found that relative symbiont density varies both between two Regiella phylogenetic clades and among aphid ‘biotypes’. Higher-density symbiont infections are correlated with stronger survival costs, but variation in density has little effect on the protection Regiella provides against fungi. Instead, we found that in some aphid genotypes, a dramatic decline in symbiont density precedes the loss of a symbiont infection. Together, our data suggest that the optimal density of a symbiont infection is likely different from the perspective of aphid and microbial fitness. Regiella might prevent loss by maintaining high within-host densities, but hosts do not appear to benefit from higher symbiont numbers and may be advantaged by losing costly symbionts in certain environments. The standing variation in symbiont density observed in natural populations could therefore be maintained by antagonistic coevolutionary interactions between hosts and their symbiotic microbes. 

Regiella strain densities - Density data for the Regiella strain variation experiment (Figure 1A). Column A (Sample) gives Regiella genotype names. Each genotype's phylogenetic clade (either 1 or 2) is shown in Column B (Clade). Column C (neg_delta_Ct) gives Regiella densitiy as qPCR output, which can be interpreted on a log2 scale. The neg_delta_Ct value is equal to 1 - {(Ct value for ef1a [aphid]) - (Ct value for hrpA [Regiella])}. 

Density_survival_correlation - Shows the values used to calculate the correlation betwen Regiella density and aphid survival (Figure 1B). Symbiont Sample and Clade are given in Columns A and B, respectively. Density is given as the average negative delta Ct value as above. Survival_Cost data are taken from Parker et al. (2017) Evolution. 

G_host x G_symbiont densities - Density data for the G_host by G_symbiont experiment (Figure 1C). Sample names are given in Column A. Host genotype and Symbiont genotype associated with each sample are given in Columns B and C, respectively, along with Symbiont clade in Column D. Symbiont density is reported as neg_delta_Ct as above (Column E). 

Host genotype densities  - Density data for the host biotype experiment (Figure 2A). Aphid genotype and the biotype from which each aphid line comes from are given in Columns A and B, respectively. Density data is shown in column C as negative delta Ct values, as above. 

Sporulation & Survival - Gives raw data for the fungal infection data (Figures 2B and 2C). Aphid line names are given in Column A, and the aphid genotype, biotype of each aphid line, and symbiont status (control or with Regiella) are shown in columns B, C, and D, respectively. Treatment is given in column E (Control or Fungus). The random number assigned to each Petri dish is given in Column F. The Number of aphids placed in each dish at the start of the experiment, and the number of aphids that survived the humidity treatment by Day 2 of the infection, are given in Columns G and H. The Number of aphids from each dish that produced a sporulating cadaver by the end of the experiment is given in Column I. 

correlations - Values used to calculate correlations between symbiont density, survival, and fungal resistance (Figures 2D and 2E). Aphid genotype and biotype are given in Columns A and B. Density is as above, in Column C. Columns D and E give the survival of each genotype without and with Regiella, respectively, and the difference between these values in Column F (survival cost). Columns G and H give the percent sporulated aphids of each genotype without and with Regiella, respectively, with the difference between these values (protection) given in Column I. 

Regiella time-course densities - Density data in the symbiont loss experiment (Figure 3). Experimental Block is given in Column A, with aphid genotype in Column B (either 317 or 663). Regiella density (negative delta Ct values as above) are shown in Columsn C and D at generations 2 and 4, respectively. Whether a line ultimately lost or retained symbionts by the end of the experiment (Generation 5) is shown in Column E.