Studies over time
Where it was studied · fieldwork location
Where it is published · journal
The two countries are highlighted on the map; everything else is dimmed. Focus goes further — it zooms to them and narrows every figure here (and the gap matrix) to those two alone. Click any row to filter the map to that tag.
Each cell counts the study sites matching both — any two of this atlas’s tags, or crossed against where the work was done and when it was published. A dashed, empty cell is a pair no indexed study in this selection has combined — those are the openings. Click any cell to colour the study sites carrying both — everything else is dimmed. Focus then narrows the map to those sites alone and rescopes every figure here; pressing it again shows everything.
Species with no indexed research, ranked by how consistently their studied relatives report the property. This is how bioprospecting has always worked — Strophanthus gave us ouabain because its relatives were known to be cardioactive.
Shares of geolocated, indexed studies in the current selection. Research published domestically or outside the major indexes is under-represented, which is a coverage artefact, not an absence of effort.
Spatial discovery · beta
Where the research happens
Pick a field and explore its research footprint — the study sites, mapped from the text of each paper, not the authors’ institutions.
Each dot is a geolocated study site.
Measured satellite data, drawn beneath the study sites.
Top of the list draws on top
Where the study sites are, cell by cell; how much of each country’s species list has a study site inside that country; and how the environment has changed in the cells with research. Drawn beneath the dots; click a cell or a country for its figures and a link to the place explorer.
Keeps a site when any of its listed studies (up to six per site in this map data) was published in that decade. The share of all located research from that decade is stated below when one is chosen.
How thoroughly a study is characterised across the codebook. Citation counts aren’t in the map data — open a study to see them.
Everything you set here — tags, country, years, the gap-matrix pair — is in the address. The link reopens this exact view.
Click a dimension to open its tags, then click tags to filter the map. Combine several.
Each dot = a geolocated study site larger = more studies
Earth observation
The satellite datasets on this atlas
Each layer is a published, peer-reviewed dataset served through Google Earth Engine and drawn underneath the study sites. COMPASS ID does not process or reinterpret the imagery — it draws the dataset as published, so what you see is the source, not our reading of it.
About this map · beta
A map of where the research happens
Each dot is a study site taken from the text of a paper — the place the work describes, not the author’s university. Pick an atlas, open a tag dimension to filter, or search a place, then click any dot to read the studies behind it.
This is a beta with partial coverage. An empty region means no indexed study was found for that field there — not that no research exists. The map shows what the corpus currently holds, and we try to be clear about its limits.
Planned — not live yet
- Natural-language search. Ask a plain question, such as “antivenom studies in West Africa,” and get the matching studies. It works by having Claude read the tagged corpus.
- Live environmental layers. Satellite-derived data from Google Earth Engine — vegetation, land degradation, surface temperature — overlaid so you can compare where research happens against conditions on the ground.
- More fields. Additional plant and animal families, and cross-cutting themes, as their corpora are tagged.
- Compare atlases. View more than one atlas together without the points overlapping confusingly.
- Full corpus browser. Search and filter every paper in a corpus directly, not just the patterns summarized in the atlases.