Maternally transmitted bacterial symbionts can be important mediators of the interactions between insect herbivores and their foodplants. These symbionts are often facultative (present in some host individuals but not others) and can have large effects on their host’s phenotype, thus giving rise to heritable variation upon which selection can act. In the cowpea aphid (Aphis craccivora) it has been established that the facultative endosymbiont Arsenophonus improves aphid performance on black locust trees (Robinia pseudoacacia) but not on fava (Vicia faba). Here, we tested whether this fitness differential translated into contemporaneous evolution of aphid populations associated with the different plants. In a laboratory study lasting 16 weeks, we found that the frequency of Arsenophonus-infected individuals significantly increased over time for aphid populations on black locust but declined for aphid populations on fava. By the end of the experiment, Arsenophonus infection was >3× more common on black locust than fava, which is comparable to previously described infection frequencies in natural field populations. Our results clearly demonstrate that aphid populations with mixed facultative symbiont infection status can rapidly evolve in response to the selective environments imposed by different host plants. This selection differential may be a sufficient explanation for the global association between Arsenophonus-infected cowpea aphids and black locust trees, without invoking additional assortative mechanisms. Because the aphid and plant originate from different parts of the world, we further hypothesize that Arsenophonus infection may have acted as a preadaptation that has promoted functional specialisation of infected aphids on a novel host plant.

All replicate aphid clones derived from a single Arsenophonus infected female, LW, originally described in Wagner et al. 2015 Functional Ecology 29: 1402-1410. The metadata for the accession and the curing process used to generate the Arsenophonus- isoline are described in that paper.  Twelve replicate populations were initiated, each with 20 Ars+ and 20 Ars- aphids at time 0.  Half of these populations (1-6) were caged on fava (Vicia faba) and half (7-12) on locust (Robinia pseudoacacia). 

Every two weeks (until week 16) one subset of 20 aphids per population were transferred to a fresh plant of the same species, and a second subset was transferred to 95% ethanol and had their DNA extracted.  Tab1 provides the data associated with each individual extraction. Extraction codes (column E) indicate the population (column A) and individual specimen (column C). Each extraction was checked for the presence (column F) or absence (column G) of Arsenophonus via diagnostic PCR for a segment of the Arsenophonus 23S gene, and were checked for extraction quality via diagnostic PCR for the aphid COI gene.  Any extractions  that did not yield COI product were considered failed extractions (column H). Each week's (column D) set of aphids per population represent 20 new aphids; aphid numbers (Column C) were randomly assigned and bear no particular significance, and no relationship to one another across populations, treatments or weeks.