Summary Herbivores often defend themselves from predation by transmitting toxic plant‐produced chemicals to their enemies. Polyphagous herbivores sometimes exhibit differential toxicity when found on various host plant species, which is generally assumed to reflect variation in plant chemistry. Here, however, we provide evidence that host‐associated herbivore lineages can intrinsically differ in their toxic properties. Lineages of Aphis craccivora originating from black locust (Robinia pseudoacacia) are unsuitable food for the ladybeetle Harmonia axyridis, resulting in death of both larvae and adults, whereas aphid lineages originating from alfalfa (Medicago sativa) support larval development and adult reproduction. We show that locust‐origin aphids remain toxic and alfalfa‐origin aphids remain non‐toxic when reared on any of three legume plants (fava, alfalfa or locust). Furthermore, toxicity is not a function of the facultative bacterial symbiont Arsenophonus, which is naturally present in locust‐origin aphid lineages and facilitates aphid use of locust. Experimentally cured locust‐origin lineages remain toxic, and an experimentally transinfected alfalfa‐origin lineage remains non‐toxic to H. axyridis. Instead, Arsenophonus plays an indirect role in the distribution of toxic aphid lineages by facilitating aphid use of locust. It is the parthenogenetic coinheritance of Arsenophonus and the toxic trait that observationally correlates locust‐feeding with toxicity in A. craccivora, rather than host plant chemistry per se. Our results clearly demonstrate that aphid lineages intrinsically vary in their toxic properties in a way that neither plant chemistry nor bacterial symbionts can explain. A more inclusive paradigm is needed for understanding variation in herbivore defence against predators.

Introduced species have been linked to declines of native species through mechanisms including intraguild predation and exploitative competition. However, coexistence among species may be promoted by niche partitioning if native species can use resources that the invasive species cannot. Previous research has shown that some strains of the aphid Aphis craccivora are toxic to a competitively dominant invasive lady beetle, Harmonia axyridis. Our objective was to investigate whether these aphids might be an exploitable resource for other, subdominant, lady beetle species. We compared larval development rate, survival, and adult weight of five lady beetle species in nochoice experiments with two different strains of A. craccivora, one of which is toxic to H. axyridis and one that is nontoxic. Two lady beetle species, Cycloneda munda and Coleomegilla maculata, were able to complete larval development when feeding on the aphid strain that is toxic to H. axyridis, experiencing only slight developmental delays relative to beetles feeding on the other aphid strain. One species, Coccinella septempunctata, also was able to complete larval development, but experienced a slight reduction in adult weight. The other two lady beetle species, Hippodamia convergens and Anatis labiculata, demonstrated generally low survivorship when consuming A. craccivora, regardless of aphid strain. All five species showed increased survival and/or development relative to H. axyridis on the "toxic" aphid strain. Our results suggest that this toxic trait may act as a narrow-spectrum defense for the aphids, providing protection against only some lady beetle enemies. For other less-susceptible lady beetles, these aphids have the potential to provide competitive release from the otherwise dominant H. axyridis. K E Y W O R D SCoccinellidae, cowpea aphid, exploitative competition, generalist predators, invasive species, niche partitioning

Characterization of intraspecific variation is required to assess the potential nontarget effects of augmentative releases of Hippodamia convergens (Guerin) (Coleoptera: Coccinellidae) from the Western United States on local populations of this species in the Eastern United States. Adults of this predatory lady beetle species overwinter in adult reproductive diapause, thus examining responses to photoperiods can characterize geographic variation influencing their seasonal biology. This laboratory study quantified the induction and duration of adult reproductive diapause in five North American populations of H. convergens in response to four constant photoperiods (16:8, 14:10, 12:12, and 10:14 [L:D] h) at 22°C. Three populations were collected over a range of latitudes (31° N to 42° N) in the central portion of the United States (Texas, Kansas, and Iowa); two populations were purchased from commercial sources in the Western United States. All populations exhibited a long-day response to photoperiods: ≤17% of females reared at 16:8 (L:D) h entered diapause, whereas shorter photoperiods (12:12 and 10:14 [L:D] h) induced diapause in 82 to 100% of females. The response to 14:10 (L:D) h showed significant variation among the populations, ranging from 0 to 89% of females in reproductive diapause. Both the phenotypic variation in response to diapause inducing photoperiods and the genetic variation in North American populations of H. convergens document the geographic variability in this widely distributed predatory species.