Plate 0Origin & Mission
Born at the microscope. Built for the field.
Foraminiferal identification is slow and error-prone — not because researchers lack skill, but because the knowledge infrastructure is fragmented. This platform exists to change that.
⟢ The Problem
A genuine scientific bottleneck, lived at the bench.
Foraminiferal species identification is one of the quieter bottlenecks in marine science. Each specimen is identified by eye under a stereomicroscope. The researcher must hold in memory the morphological profiles of hundreds of species across dozens of genera — then search manually across disconnected legacy catalogs, publications, and institutional archives to find a description that matches what they see.
The taxonomic literature is vast and fragmented. Morphological terminology varies across authors and languages — the same character may be described as "foliate" in one paper and "foliáceo" in another, with no computational bridge between the two. There is no single place where this knowledge is concentrated and searchable.
Time Cost
A single master’s identification campaign — 50 slides, more than 300 specimens each, using only books, paper catalogs, and dispersed journal searches — can consume roughly eighteen months of a two-year programme, leaving less than six for the quantitative analyses and writing.
Error Compounding
Fatigue, deadline pressure, inconsistent access to reference materials, and the absence of a shared morphological vocabulary all amplify misidentification. Expert disagreement on the same physical material is not uncommon.
Expertise Barrier
Foram identification requires high specialisation built over years. Labs that lose a senior specialist — or onboard a new researcher — face a steep re-entry cost with no standardised support infrastructure.
Reproducibility Gap
Foraminifera are key proxies for reconstructing past climates and monitoring present ocean conditions. Inconsistent taxonomy undermines the reproducibility of paleoclimate reconstructions and biostratigraphic correlations.
"I can see the individual. I recognise the aperture, the coiling, the ornamentation, the wall type. I want to select those characters and narrow the possible species — without having to guess the order, family, or genus."
⟢ The Insight
The real bottleneck is infrastructure, not a shortage of papers.
The existing literature is large. WoRMS maintains the taxonomic tree. Mikrotax, Foraminifera.eu, and regional atlases hold partial character data for planktonic groups. Ellis & Messina holds an enormous legacy catalog. The knowledge exists — it simply is not connected, standardised, or queryable in a way that serves the researcher at the bench.
The answer, therefore, is not to produce more papers or train more experts in isolation. It is to build the shared data infrastructure: a morphological character matrix with a controlled vocabulary, a provenance-tracked catalogue keyed by WoRMS AphiaID, and a filter that turns an observation into a ranked candidate list. Then open it.
A second insight shaped the solution: rather than image-based deep learning — which requires high-resolution equipment, GPUs, and infrastructure far beyond most labs — a character-based filter works with any stereomicroscope and democratises access across the entire field.
- 1
Observe specimen
Stereomicroscope · any resolution
- 2
Select characters
Coiling · aperture · wall · ornament · outline
- 3
Testaria filter
Faceted morphological search replaces manual catalog search
- 4
Ranked candidates
Each with description, synonymies, images, and full provenance
- 5
Researcher confirms
Expert judgement retained; identification is citable and auditable
"The platform’s role is humble on taxonomy — sync, never fork — and ambitious on morphology: to originate the character matrix that no other database covers."
⟢ Where We Stand
Pre-MVP. Beginning a long arc.
The platform is early-stage and states this clearly. The conviction behind it is not. The timing — AI and automation now accessible without industrial infrastructure — makes a working prototype possible. That prototype opens doors.
Character-based, not image-based
Works with any stereomicroscope. Democratises access to labs that cannot afford high-resolution imaging rigs.
Shallow-water and Recent priority
Focuses where Mikrotax and planktonic atlases are weakest — shallow-platform benthics and extant material.
AI-native ingestion pipeline
Nomia processes scientific PDFs at scale, rather than relying on manual data entry, making the catalogue continuously updatable.
Explicit reliability methodology
Every identification carries a confidence record and full provenance chain — auditable, reproducible, citable.
Omnichannel access
Browser, REST/MCP APIs, voice at the microscope — the filter meets the researcher where they work.
Open and community-governed
No paywall, no proprietary lock-in. Aligned with POSI principles for open scholarly infrastructure.
A 5–10+ year mission
The infrastructure is being built. The door is open.
Foraminifera.org is at the beginning of its arc. If you work in micropaleontology, marine biology, or adjacent fields, your contribution — data, images, feedback, collaboration — matters now.