International audienceThe host-associated differentiation (HAD) hypothesis states that higher trophic levels in parasitic associations should exhibit similar divergence in case of host sympatric speciation. We tested HAD on populations of Aphidius ervi the main parasitoid of the pea aphid Acyrthosiphon pisum, emerging from host populations specialized on either alfalfa or red clover. Host and parasitoid populations were assessed for genetic variation and structure, while considering geography, host plant and host aphid protective symbionts Regiella insecticola and Hamiltonella defensa as potential covariables. Cluster and hierarchical analyses were used to assess the contribution of these variables to population structure, based on genotyping pea aphids and associated A. ervi with microsatellites, and host aphid facultative symbionts with 16S rDNA markers. Pea aphid genotypes were clearly distributed in two groups closely corresponding with their plant origins, confirming strong plant associated differentiation of this aphid in North America. Overall parasitism by A. ervi averaged 21.5 % across samples, and many parasitized aphids producing a wasp hosted defensive bacteria, indicating partial or ineffective protective efficacy of these symbionts in the field. The A. ervi population genetic data failed to support differentiation according to the host plant association of their pea aphid host. Potential for parasitoid specialization was also explored in experiments where wasps from alfalfa and clover aphids were reciprocally transplanted on alternate hosts, the hypothesis being that wasp behaviour and parasitic stages should be most adapted to their host of origin. Results revealed higher probability of oviposition on the alfalfa aphids, but higher adult emergence success on red clover aphids, with no interaction as expected under HAD. We conclude that our study provides no support for the HAD in this system. We discuss factors that might impair A. ervi specialization on its divergent aphid hosts on alfalfa and clover

Insect parasitoids and their insect hosts represent a wide range of parasitic trophic relations that can be used to understand the evolution of biotic diversity on earth. Testing theories of coevolution between hosts and parasites is based on factors directly involved in host susceptibility and parasitoid virulence. We used controlled encounters with potential hosts of the Aphidius ervi wasp to elucidate behavioral and other phenotypic traits of host Acyrthosiphon pisum that most contribute to success or failure of parasitism. The host aphid is at an advanced stage of specialization on different crop plants, and exhibits intra-population polymorphism for traits of parasitoid avoidance and resistance based on clonal variation of color morph and anti-parasitoid bacterial symbionts. Randomly selected aphid clones from alfalfa and clover were matched in 5 minute encounters with wasps of two parasitoid lineages deriving from hosts of each plant biotype in a replicated transplant experimental design. In addition to crop plant affiliation (alfalfa, clover), aphid clones were characterized for color morph (green, pink), Hamiltonella defensa and Regiella insecticola symbionts, and frequently used behaviors in encounters with A. ervi wasps. A total of 12 explanatory variables were examined using redundancy analysis (RDA) to predict host survival or failure to A. ervi parasitism. Aphid color was the best univariate predictor, but was poorly predictive in the RDA model. In contrast, aphid host plant and symbionts were not significant univariate predictors, but significant predictors in the multivariate model. Aphid susceptibility to wasp acceptance as reflected in host attacks and oviposition clearly differed from its suitability to parasitism and progeny development. Parasitoid progeny were three times more likely to survive on clover than alfalfa host aphids, which was compensated by behaviorally adjusting eggs invested per host. Strong variation of the predictive power of intrinsic (body color) and extrinsic traits (symbionts, host plant), indicate that host variables considered as key predictors of outcomes strongly interact and cannot be considered in isolation.