1. Floral plantings adjacent to crop fields can recruit populations of natural enemies by providing flower nectar and non-crop prey to increase natural pest regulation. Observed variation in success rates might be due to changes in the unseen community of endosymbionts hosted by many herbivorous insects, of which some can confer resistance to natural enemies, e.g. parasitoid wasps. Reduced insect control may occur if highly protective symbiont combinations increase in frequency via selection effects, and this is expected to be stronger in lower diversity systems.
2. We used a large-scale field trial to analyse the bacterial endosymbiont communities hosted by cereal aphids (Sitobion avenae) collected along transects into strip plots of barley plants managed by either conventional or integrated (including floral field margins and reduced inputs) methods. In addition, we conducted an outdoor pot experiment to analyse endosymbionts in S. avenae aphids collected on barley plants that were either grown alone or alongside one of three flowering plants, across three time points.
3. In the field, aphids hosted up to four symbionts. The abundance of aphids and parasitoid wasps was reduced towards the middle of all fields while aphid symbiont species richness and diversity decreased into the field in conventional, but not integrated, field-strips. The proportion of aphids hosting different symbiont combinations varied across cropping systems, with distances into the fields, and was correlated with parasitoid wasp abundances.
4. In the pot experiment, aphids hosted up to six symbionts. Flower presence increased natural enemy abundance and diversity, and decreased aphid abundance. The proportion of aphids hosting different symbiont combinations varied across the flower treatment and time and were correlated with varying abundances of the different specialist parasitoid wasp species recruited by different flowers.
5. Synthesis and applications. Floral plantings and flower identity had community-wide impacts on the combinations of bacterial endosymbionts hosted by herbivorous insects, which correlated with natural enemy diversity and abundance. We recommend that integrated management practices incorporate floral resources within field areas to support a more functionally diverse and resilient natural enemy community to mitigate selection for symbiont-mediated pest resistance throughout the cropping area.

Field data were collected from The CSC platform at Balruddery Farm near Dundee, Scotland in 2016 and 2017. Conventional and integrated management treatments were randomly allocated to each half, and (for the barley fields) in each field, half three different barley varieties were planted as plot strips. Sitobion avenae aphids were collected on 6th July 2016 and 3rd July 2017, at four distances (5, 15, 30, 50 m) from the edge of the field into each barley variety strip. Aphids were stored in 70% ethanol and shipped to Technical University of Munich for symbiont analysis. At each sampling location (distance into the field strip), a maximum of five aphids (one per colony) were collected from three adjacent infested tillers in every barley strip (variety); aphids reproduce asexually, and females deposit a group of offspring in one patch before moving away, this approach minimises the chance of collecting aphids from the same clonal mother. Total aphid number was also counted for these three tillers. In total, there were four sampling distances for three barley varieties, repeated across the two management systems, and across two years. Additional data were collected on % cover of the crop and weed plants in the sampled areas using a 1 m2 quadrat to estimate the projected area of ground covered by each vegetation type in mid-July. Data on parasitoid and generalist natural enemy abundance were obtained from two yellow sticky traps, one placed at the top and one at the bottom of the plant canopy, at each sampling location, in mid-to-late July.

Data were collected from an outside pot experiment in 2018. We used four barley cultivars and three flowering plants (Fagopyrum esculentum, Vicia faba, Trifolium pratense), with flower treatments of no-flower, one flower (each for the three species) and mixed flowers. Treatments were replicated six times and the experiment set up as a complete randomised block design. All insects were allowed to naturally colonise the plants and observations were taken weekly (2018) from May until seed harvest in July. Data collected included all aphid and natural enemy species per plant. Aphids were collected for endosymbiont analysis on 23rd June (late June), 11th July (early July) and 25th July (late July). One aphid per colony on every plant was collected and stored in 70% ethanol.

Data were analysed in R using linear models, mixed effect models, and structural equation models. 

The Excel file contains an explanation of each of the variables and the two datasets in three worksheets:

(1) ReadME with variable explanations

(2) DATA_PotExperiment

(3) DATA_FIeldExperiment