Diversity, connectivity and negative interactions define robust microbiome networks across land, stream, and sea

In this era of rapid global change, factors influencing the stability of ecosystems and their functions have come into the spotlight. For decades the relationship between stability and complexity has been investigated in modeled and empirical systems, yet results remain largely context dependent. To overcome this we leverage a multiscale inventory of fungi and bacteria ranging from single sites along an environmental gradient, to habitats inclusive of land, sea and stream, to an entire watershed. We use networks to assess the relationship between microbiome complexity and robustness and identify fundamental principles of stability. We demonstrate that while some facets of complexity are positively associated with robustness, others are not. Beyond positive biodiversity x robustness relationships we find that the number of "gatekeeper" species or those that are highly connected and central within their networks, and the proportion of predicted negative interactions are universal indicators of robust microbiomes. With the potential promise of microbiome engineering to address global challenges ranging from human to ecosystem health we identify properties of microbiomes for future experimental studies that may enhance their stability. We emphasize that features beyond biodiversity and additional characteristics beyond stability such as adaptability should be considered in these efforts.