LAUSR

Bacterial endophytes may be fairly host‐specific; nonetheless, an important subset of taxa may be shared among numerous host species forming a community‐wide core microbiome. Moreover, other key factors, particularly the supply of limiting macronutrients and disturbances, may supersede the importance of host identity. We tested the following four non‐mutually exclusive hypotheses: (a) The Host Identity Hypothesis: endophytes vary substantially among different host‐plant species. (b) The Core Microbiome Hypothesis: a subset of microbial taxa will be shared among all host‐plant species. (c). The Soil Resource Supply Hypothesis: endophytes vary substantially among habitats with experimentally elevated levels of macronutrients. (d) The Disturbance–Disruption Hypothesis: disturbances created by the periodic application of antibiotics structure bacterial endophyte communities. We tested these hypotheses by characterizing endophytes using high‐throughput sequencing among seedlings of five phylogenetically diverse tree species nested within a long‐term, full factorial nitrogen, phosphorus and potassium soil fertilization experiment. We artificially disturbed one of our focal species by applying antibiotics every 10–14 days for 29 months within the soil (N, P, K) fertilization experiment. While we detected a significant effect of host identity and soil nutrient additions, together they explained little variation in endophyte community composition (<10%). We found unequivocal evidence for a core microbiome shared by all species. Specifically, we inferred that nine OTUs were present among 95% or more of all control saplings, representing a third of the total reads. These bacterial taxa belonged to the Actinobacteria. In contrast, disturbance (antibiotics) explained more endophyte variation than all nutrient addition combinations combined and twice the variation explained by host identity for all five tree species. Synthesis. Our results challenge the idea that host identity is a primary filter shaping bacterial endophyte communities; this suggests that many of the same bacterial taxa occur inside plant hosts even if those host plants are phylogenetically diverse. Moreover, we documented a distinct core microbiome shared among our five focal tree species. Finally, we found that disturbance rather than host identity and soil nutrient availability was an important driver of microbial community composition, which parallels the importance of disturbance in other areas of community ecology.