6 Taxonomy

Integrating distribution data from several providers requires resolving taxonomic identity across naming systems, so that models of the same species merge rather than being counted as different species. The MST uses a multi-authority matching pipeline in which every source name resolves to one canonical identifier.

6.1 Two taxonomic namespaces

Taxa resolve into one of two authorities, not one:

Birds → BirdLife (BOTW:{sisid}). WoRMS coverage of birds is poor and its isMarine flag is unreliable for seabirds, so birds are not resolved through WoRMS.
Everything else → WoRMS (WORMS:{aphia_id}), resolved from the scientific name via the ITIS↔︎WoRMS crosswalk, then exact WoRMS matching, then the WoRMS REST API for the remainder.

A merged model is keyed by that identity — ms_merge|WORMS:137209, ms_merge|BOTW:22694927 — which is also why species must be counted on the (authority, id) pair: the two namespaces number independently, so counting bare ids collides birds with invertebrates.

6.2 Taxonomic authorities

Table 6.1: Taxonomic authorities used for name resolution.

Authority	Role	Key
WoRMS	World Register of Marine Species — canonical identity for all non-bird marine taxa, and the source of the class/phylum used to assign a species category	`worms_id` (AphiaID)
BirdLife BOTW	Birds of the World — canonical identity and ranges for birds	`botw_id` (SIS ID)
GBIF	Backbone taxonomy, used as a crosswalk into WoRMS	`gbif_id`
ITIS	US federal standard; the first match attempted where a source supplies a TSN	`itis_tsn`
IUCN Red List	Conservation status, and expert range maps	`iucn_id`

At each stage deprecated names resolve to their accepted synonyms, with the original source name kept alongside the accepted one.

6.3 Species categories

Table 6.2: Species categories in v7. Scored is read from the metric registry — a category is scored only where the release published an extrisk_{cat} metric for it. Valid counts the taxa eligible for scoring.

Category	Taxa	Valid	Scored
`invertebrate`	8,243	8,062	Yes
`fish`	6,679	5,943	Yes
`other`	1,129	1,125	Yes
`bird`	614	476	Yes
`coral`	775	457	Yes
`mammal`	89	84	Yes
`turtle`	6	6	Yes
`reptile`	26	0	No

Categories present in the taxonomy but not scored in v7: reptile. They are carried in the taxon table — the identity is real — but no metric aggregates them, so they contribute nothing to any score.

How categories are assigned

Sea turtles are their own category (turtle), separated from reptile so that the multiplicative turtle merge and the turtle metric have somewhere to live (see Chapter 8). Non-turtle reptiles — sea snakes, marine iguanas — remain in the taxonomy but are not scored.
other collects taxa this release’s assignment could not place. It is a symptom rather than a category, and later releases removed the need for it by assigning from the taxonomic hierarchy instead — see Releases.

6.4 Marine relevance

Not every taxon in the source data belongs in a marine sensitivity analysis, and the obvious test does not work.

WoRMS isMarine fails for birds. Birds are resolved through BirdLife, so most have no WoRMS record at all, and those that do are inconsistently flagged; relying on it would drop nearly every seabird. Earlier releases sidestepped this with a purely spatial test — a taxon counted as marine if its distribution overlapped the study area — which admits any coastal or inland bird whose range happens to reach the coast.

This release predates the explicit marine-relevance cull. Marine status comes from WoRMS isMarine, with birds admitted through BirdLife instead, so the bird counts on this page are broader than in later releases.

6.5 From source taxa to valid species

Table 6.3: Validity gates applied cumulatively to v7’s published taxon table. The final row applies the release’s own validity flag to the whole table — it is the authoritative count, and not simply the conjunction of the diagnostic gates above it.

Filter step	Removed	Taxa remaining
Source taxa (all datasets)	—	17,561
Resolved to a taxon id	—	17,561
Has a merged distribution	−1,173	16,388
Not extinct	−21	16,367
Marine (WoRMS, or a BirdLife bird)	−48	16,319
Valid for scoring (the release’s `is_ok`)	−166	16,153

This yields 16,153 valid species across the study area.

Of those, 9,230 have modeled distributions inside the 20 BOEM Program Areas of the current program cycle. That is a spatial result, obtained by asking the zone tables which species occur there — not a property of the validity flag.

Validity is not the same question as “occurs in a Program Area”

Through v6 the validity flag silently required overlap with a BOEM Program Area, conflating “this is a real, mapped marine species” with “this species happens to fall inside the areas of the current program cycle” — and so hiding biodiversity outside that cycle. From v7 the two are separate: validity means the taxon has a modeled distribution in the study area, and Program-Area membership is a zone query answered afterwards.

This is why the headline count is not comparable across releases: it rose sharply at v7 without a single species being added, because the definition changed. Per-release counts are on the Releases page.

6.6 Data quality

6.6.1 Duplicate resolution

Several datasets may supply a model for the same species. Resolving every source name to one canonical identifier is what makes those models merge — taking the maximum per cell — instead of appearing as separate species and counting the taxon more than once in every score.

6.6.2 Synonym handling

Names change as taxa are reclassified. The pipeline resolves through acceptedNameUsageID in each authority, keeps the original source name alongside the accepted one, and updates deprecated identifiers to their current accepted form.

6.7 Implementation

Name resolution is msens::match_taxa(), which runs the identifier cascade and returns the unified taxon table. Category assignment, the marine cull and the validity flags are applied in the merge_taxon workflow (see Chapter 15).