No abstract

Background Host-associated microorganisms are now recognised as being critical for eukaryotic host functioning; however, most studies to date have focused on descriptive approaches or have used model systems, usually in the laboratory, to understand host-microbiome interactions. To advance our understanding of host-microbiome interactions and their wider ecological impacts, we need (i) robust experimental frameworks to explore causality in host-microbiome interactions and (ii) protocols that apply to model systems but also to often highly diverse natural systems. Results We used a dominant habitat-forming seaweed, Hormosira banksii, to explore a widely applicable framework for experimentally testing host-microbiome interactions. The experimental protocols were particularly designed to try and disentangle microbially-mediated effects on hosts from direct effects on hosts associated with the methods employed to manipulate host-microbiota. This was done through a combination of antimicrobial treatments, which have widespread use in holobiont research, and inoculations, in mesocosms and in the field. Three different antibiotic treatments were used to disrupt seaweed-associated microbial communities to test whether such microbiome disturbances would negatively affect host performance. Responses of microbiomes to these disturbances were complex and differed substantially among treatments. However, by comparing the temporal sequence of antibiotic treatments, changes in microbial diversity, and decreases in host performance, a consistent effect of the microbiome on host performance was observed in some treatments. To further test these effects, we used gene sequencing to identify microbial taxa that were both correlated and uncorrelated with poor host performance following antibiotic treatment. These were then isolated and used in inoculation experiments, independently or in combination with the previously used antibiotic treatments. Negative effects on host condition were strongest where specific microbial disturbances (by particular antimicrobials) were combined with inoculations of strains correlated with poor host performance. For these treatments, negative host effects persisted the entire experimental period (12 days), even though treatments were only applied at the beginning of the experiment. Host condition recovered in all other treatments. Conclusions This experimental framework allows for causal relationships to be determined within ecologically important holobionts. This should allow for better predictions of how these systems will respond to, and potentially mitigate, environmental disturbances in their natural context.