The insect-killing fungi are studied for crop protection and for medicine — across a family most of which is unstudied.

This is the family of the prized medicinal Cordyceps, of Beauveria bassiana — the white fungus sprayed on crops to kill pests — and of the depsipeptide toxins that let them do it. COMPASS links each of ~654 accepted species to OpenAlex, tags every study through a domain-expert panel with adversarial cross-model judging, and geocodes each to its study site. The result is a census of a lineage that sits at the intersection of agriculture, medicine, and one of biology’s strangest life histories.

The methods follow suit: culturing and antibiosis bioassays (1,858 studies) and fermentation and bioprocessing (1,220) dominate. Nearly six in ten studies are experimental. This is an applied science — a field asking, above all, how do we use them?

Beauveria, the biocontrol staple, is the best-studied genus — only a fifth of its species unrecorded. But even Cordyceps is 70% unstudied; Isaria, 91%; Torrubiella, 80%; Samsoniella, 79%. Attention centres on the “zombie-ant fungus” and the tonic mushroom, but most of the family — hundreds of species of insect-specific parasites — is unstudied.

Conservation framing barely registers: threats appear in just 1% of studies, and where they do, the dominant one is that these fungi are themselves an invasive concern — a reflection of the biocontrol lens, not a worry for the family’s own survival. A lineage we deploy by the tonne against crop pests, and harvest by the tonne for medicine, is one we have made almost no attempt to conserve — or even to count as at risk.

India (83 sites) and China (61) lead — India and Brazil through tropical entomopathogen diversity and biocontrol research, China through the deep cultural and commercial pull of medicinal Cordyceps. Korea and Colombia punch above their size. But most study sites are national-scale — this is a lab and cultivation science as much as a field one, and much of the family’s tropical range is barely sampled.

This atlas is one view of a larger queryable corpus. A single genus’s evidence profile, the biocontrol-versus-medicine split for any species, or a shortlist of unstudied taxa ready for survey are each a query away, on the same reproducible foundation.