ABSTRACT: Insects harbor a variety of maternally inherited bacterial symbionts. As such, variation in symbiont presence/absence, in the combinations of harbored symbionts, and in the genotypes of harbored symbiont species provide heritable genetic variation of potential use in the insects' adaptive repertoires. Understanding the natural importance of symbionts is challenging but studying their dynamics over time can help to elucidate the potential for such symbiont-driven insect adaptation. Toward this end, we studied the seasonal dynamics of six maternally transferred bacterial symbiont species in the multivoltine pea aphid (Acyrthosiphon pisum). Our sampling focused on six alfalfa fields in southeastern Pennsylvania, and spanned 14 timepoints within the 2012 growing season, in addition to two overwintering periods. To test and generate hypotheses on the natural relevance of these non-essential symbionts, we examined whether symbiont dynamics correlated with any of ten measured environmental variables from the 2012 growing season, including some of known importance in the lab. We found that five symbionts changed prevalence across one or both overwintering periods, and that the same five species underwent such frequency shifts across the 2012 growing season. Intriguingly, the frequencies of these dynamic symbionts showed robust correlations with a subset of our measured environmental variables. Several of these trends supported the natural relevance of lab-discovered symbiont roles, including anti-pathogen defense. For a seventh symbiont-Hamiltonella defensa-studied previously across the same study periods, we tested whether a reported correlation between prevalence and temperature stemmed not from thermally varying host-level fitness effects, but from selection on co-infecting symbionts or on aphid-encoded alleles associated with this bacterium. In general, such "hitchhiking" effects were not evident during times with strongly correlated Hamiltonella and temperature shifts. However, we did identify at least one time period in which Hamiltonella spread was likely driven by selection on a co-infecting symbiont-Rickettsiella viridis. Recognizing the broader potential for such hitchhiking, we explored selection on co-infecting symbionts as a possible driver behind the dynamics of the remaining six species. Out of twelve examined instances of symbiont dynamics unfolding across 2-week periods or overwintering spans, we found eight in which the focal symbiont underwent parallel frequency shifts under single infection and one or more co-infection contexts. This supported the idea that phenotypic variation created by the presence/absence of individual symbionts is a direct target for selection, and that symbiont effects can be robust under co-habitation with other symbionts. Contrastingly, in two cases, we found that selection may target phenotypes emerging from symbiont co-infections, with specific species combinations driving overall trends for the focal dynamic symbionts, without correlated change under single infection. Finally, in three cases-including the one described above for Hamiltonella-our data suggested that incidental co-infection with a (dis)favored symbiont could lead to large frequency shifts for "passenger" symbionts, conferring no apparent cost or benefit. Such hitchhiking has rarely been studied in heritable symbiont systems. We propose that it is more common than appreciated, given the widespread nature of maternally inherited bacteria, and the frequency of multi-species symbiotic communities across insects.
