Effects of microbial evolution dominate those of experimental host-mediated indirect selection

Microbes ubiquitously inhabit animals and plants, often affecting their host's phenotype. As a result, even in a constant genetic background, the host's phenotype may evolve through indirect selection on the microbiome. 'Microbiome engineering' offers a promising novel approach for attaining desired host traits but has been attempted only a few times. Building on the known role of the microbiome on development in fruit flies, we attempted to evolve earlier-eclosing flies by selecting on microbes in the growth media. We carried out parallel evolution experiments in no- and high-sugar diets by transferring media associated with fast-developing fly lines over the course of four selection cycles. In each cycle, we used sterile eggs from the same inbred population, and assayed mean fly eclosion times. Ultimately, flies eclosed seven to twelve hours earlier, depending on the diet, but microbiome engineering had no effect relative to a random-selection control treatment. 16S rRNA gene sequencing showed that the microbiome did evolve, particularly in the no sugar diet, with an increase in Shannon diversity over time. Thus, while microbiome evolution did affect host eclosion times, these effects were incidental. Instead, any experimentally enforced selection effects were swamped by uncontrolled microbial evolution, likely resulting in its adaptation to the media. These results imply that selection on host phenotypes must be strong enough to overcome other selection pressures simultaneously operating on the microbiome. The independent evolutionary trajectories of the host and the microbiome may limit the extent to which indirect selection on the microbiome can ultimately affect host phenotype. Random-selection lines accounting for independent microbial evolution are essential for experimental microbiome engineering studies.