Digital phenotyping accelerates soil biodiversity discovery

Soils are among the most biodiverse ecosystems on the planet, home to staggering amounts of small organisms. Many of these organisms are not known to science. This biodiversity 'dark matter' remains largely unexplored because identifying small soil organisms requires either specialized taxonomic expertise or expensive molecular methods, creating a throughput bottleneck that restricts ecosystem-scale monitoring. To explore this dark matter, we developed a high-throughput digital phenotyping approach using multispectral flow cytometry to create digital 'fingerprints' of soil organisms. Analyzing 2318 organisms spanning nematodes, collembola, mites, and tardigrades, we show that these digital fingerprints distinguish taxonomic groups with high accuracy and capture phylogenetic signal that explains 91% of variance in DNA barcode relationships. Machine learning alignment enables us to assess genetic similarity based solely on digital fingerprints, allowing prediction of relationships without sequencing. Smart sampling strategies guided by these projections achieve 6-fold improvements in species discovery efficiency compared to traditional approaches, with advantages that compound as sampling increases. Our smart sampling approach has applications across domains and provides a scalable pathway for rapid biodiversity assessment with immediate applications in agriculture, conservation, and ecosystem monitoring.