New approaches to understanding soil biodiversity with the Smart Soil Organism Detector

Soil biodiversity assessment typically quantifies taxonomic composition (which taxa are present) without accounting for dimensions that shape ecological interpretation. We use the Smart Soil Organism Detector (Smart SOD) to describe three surprising aspects of soil biodiversity: life stages, substrate, and canopy soil.  The Smart Soil Organism Detector uses machine learning to provide high throughput, high resolution imagery and assessments of individual organisms extracted from the soil.  Through large-scale sampling efforts from 1,500 soil samples across California we have recovered more than 10,000 organisms spanning Nematoda, Arachnida, Tardigrada, and Collembola.  With this diversity, we have discovered that life stages of organisms not only vary extensively morphologically, but also affect the ability and extent of taxonomic identification.  Soil substrate also plays an important role in determining what organisms are able to be extracted and identified from the soil - including those embedded into soil particles.  These factors, explored with a combination of high throughput imagery and morphological scans, suggest a breadth and depth of soil biodiversity that is not quantified using traditional methods.  These findings are reinforced by the exploration of the surprising diversity of organisms found in the soil of the canopies of old-growth redwood trees.  Together, these findings showcase an unexplored arena of soil biodiversity that has largely been inaccessible and overlooked through traditional sampling methods, while showcasing what could be possible through more expansive exploration.