Plants and insect herbivores are two of the most diverse multicellular groups in the world, and both are strongly influenced by interactions with the belowground soil microbiome. Effects of reciprocal rapid evolution on ecological interactions between herbivores and plants have been repeatedly demonstrated, but it is unknown if (and how) the soil microbiome could mediate these eco-evolutionary processes on a shared host plant. We tested the role of a plant-beneficial soil bacterium (Acidovorax radicis) in altering eco-evolutionary interactions between different aphid genotypes (Sitobion avenae; genotypes Sickte and Fescue) feeding on barley (Hordeum vulgare). We measured fecundity, longevity and population growth of two aphid genotypes reared separately or together (population mixture) on three different barley varieties that were inoculated with or without A. radicis. Results showed that across all plant varieties A. radicis increased plant growth and suppressed aphid populations via reduced longevity and fecundity. The strength of effect was dependent on aphid genotype and barley variety, while the direction of effect was altered by aphid population mixture. Using Lotka-Volterra modelling, we demonstrated that while A. radicis inoculation decreased growth rates for both aphid genotypes it increased the competitiveness of one genotype against the other. In general, in the presence of A. radicis, the Fescue aphid genotype became more inhibitory of Sickte aphids, while Sickte aphids facilitated the growth of Fescue aphids. Our work demonstrates that plant rhizosphere microbiomes exert community-level influences by mediating eco-evolutionary interactions between herbivores and host plants. By altering competitive interaction outcomes among aphids and thus impacting processes such as rapid evolution, soil microbes contribute to the short- and long-term structure and functioning of terrestrial habitats.

Data were collected from a glasshouse pot experiment in Jan-Feb 2020. We used (1) three barley cultivars [Barbarella (Elsoms Seeds), Chevallier (New Heritage Barley Ltd), Irina (KWS)]; (2) two genotypes of the English grain aphid Sitobion avenae (L.) that have been maintained as low density stock populations on barley variety ‘Chanson’, including a dark brown genotype ‘Fescue’ and a pale pinkish coloured genotype ‘Sickte’; and (3) the rhizobacteria Acidovorax radicis N35. 

Experiment 1: Population growth. Using a fully-factorial design of the factors above, and including no-aphid, single aphid genotype, and a mixed aphid genotype population, the 24 treatments were replicated 8 times each. Individual barley plants in 10cm diameter pots were grown in a glasshouse (average temperature of 20°C), infested with four 3rd instar aphids (either 4 of one genotype or two of each genotype for the mixed populations), every plant covered with an air-permeable cover, and aphid numbers counted on days 7, 14, and 21. Plant height and biomass was also measured. Data were analysed in R using linear models and mixed effect models (repeated measure over time) for effects of experimental variables.

Experiment 2: Survival and fecundity. Using the same factorial design as above, the 24 treatments were replicated 6 times each. Individual barley plants were infested with four 3rd instar aphids as in expt1, five days later after aphid reproduction all adults were removed alongwith all except four 1st instar nymphs (either 4 of one genotype or two of each genotype for the mixed populations). Aphids were followeduntil adulthood and then offspring produced counted every 1-2 days with offspring removal, with adults followed and offspring removed until aphid death. Aphid lifetime fecundity, daily fecundity and longevity (survival) were caluclated and effect of experimental variables on these analysed using linear models and mixed effect models (repeated measure over time).

Modelling: We modelled intergenotypic competition between a population of the aphid genotypes growing on the same host plant, via a Lotka-Volterra model. We fitted the data from the first experiment to this model, in order to quantify the effects of host genotype and the presence/absence of bacteria on the growth and interactions of the two aphid genotypes.