Nonsensical choices? Fall armyworm moths choose seemingly best or worst hosts for their larvae, but neonate larvae make their own choices.

Selecting optimal host plants is critical for herbivorous insects, such as fall armyworm (Spodoptera frugiperda), an important maize pest in the Americas and Africa. Fall armyworm larvae are presumed to have limited mobility, hence female moths are presumed to be largely responsible for selecting hosts. We addressed host selection by fall armyworm moths and neonate and older (3rd-instar) larvae, as mediated by resistance and herbivory in maize plants. Thus, we compared discrimination among three maize cultivars with varying degrees of resistance to fall armyworm, and between plants subjected or not to two types of herbivory. The cultivars were: (i) susceptible, and deficient in jasmonic acid (JA) production and green leaf volatiles (GLV) emissions (inbred line B73-lox10); (ii) modestly resistant (B73), and; (iii) highly resistant (Mp708). The herbivory types were: (i) ongoing (= fall armyworm larvae present), and; (ii) future (= fall armyworm eggs present). In choice tests, moths laid more eggs on the highly resistant cultivar, and least on the susceptible cultivar, though on those cultivars larvae performed poorest and best, respectively. In the context of herbivory, moths laid more eggs: (i) on plants subject to versus free of future herbivory, regardless of whether plants were deficient or not in JA and GLV production; (ii) on plants subject versus free of ongoing herbivory, and; (iii) on plants not deficient in compared to deficient in JA and GLV production. Neonate larvae dispersed aerially from host plants (i.e. ballooned), and most larvae colonized the modestly resistant cultivar, and fewest the highly resistant cultivar, suggesting quasi-directional, directed aerial descent. Finally, dispersing older larvae did not discriminate among the three maize cultivars, nor between maize plants and (plastic) model maize plants, suggesting random, visually-oriented dispersal. Our results were used to assemble a model of host selection by fall armyworm moths and larvae, including recommendations for future research.