KeywordsArthropod-plant-symbiont interaction, bacterial symbiont, fitness effects, host-plant use, mutualism, parasitism.
Abstract 1In the last decades, many studies had revealed the potential role of arthropod bacterial 2 endosymbionts in shaping the host range of generalist herbivores and their performance on different 3 host plants, which, in turn, might affect endosymbiont distribution in herbivores populations. We 4 tested this by measuring the prevalence of endosymbionts in natural populations of the generalist 5 spider mite Tetranychus urticae on different host plants. Focusing on Wolbachia, we then analysed 6 how symbionts affected mite life-history traits on the same host-plants in the laboratory. Overall, the 7 prevalences of Cardinium and Rickettsia were low, whereas that of Wolbachia was high, with the 8 highest values on bean and eggplant and the lowest on purple, tomato and zuchini. Although most 9 mite life-history traits were affected by the plant species only, Wolbachia infection was detrimental 10 for egg hatching rate on purple and zucchini, and led to a more female-biased sex ratio on purple and 11 eggplant. These results suggest that endosymbionts may affect the host range of polyphagous12 herbivores, both by aiding and hampering their performance, depending on the host plant and on 13 the life-history trait that affects performance the most. Conversely, endosymbiont spread may be14 facilitated or hindered by the plants on which infected herbivores occur.15 42 (Hackett et al., 2013), and anti-herbivory defences (Barr et al., 2010, Su et al., 2015. Conversely, 43 symbiont-mediated decreased host performance on particular plants might be due to the nutrient 44 profile (e.g., specific amino acids and nitrogen content) of these plants, which promotes deleterious 45 symbiont traits and disturbs the host control over bacterial abundance (Wilkinson et al., 2007, 46 Chandler et al., 2008.