The Sparsely Sampled Ubiquity of Global Fungal Endophytes

The paradigm that ‘all plants harbor fungal endophytes’ is a widely accepted axiom in plant ecology, yet this claim has never been tested. Here, we use a machine learning pipeline to screen 21,891 research abstracts spanning nearly a century. We developed a high-sensitivity screening tool to find and manually validate all potential claims of endophyte absence. This comprehensive validation revealed no evidence of a verifiable endophyte-free plant taxon within the entire scientific literature surveyed. This finding confirms that endophyte ubiquity is a robust pattern wherever plants have been studied. However, we show that this “ubiquitous” paradigm is a generalization from an extraordinarily biased sample: analysis of the 19,119 relevant primary studies reveals that only 0.8% of described plant species have been examined for fungal endophytes, with 77% of this research concentrated in the Global North. Our analysis reframes a foundational paradigm, revealing that our understanding of a “global” symbiosis is based on a tiny fraction of biodiversity. This sampling bias represents a critical bottleneck in biodiversity research, obscuring the discovery of novel biotechnologies and climate-resilient symbioses in the world’s most threatened ecosystems.