GBIF holds 1,060,583 Philippine occurrence records that anyone may reuse without conditions. Institutions in the United States publish 66.7% of them. Philippine institutions publish 5,016 — 6th place, behind Switzerland.
GBIF records carry whatever licence their publisher chose, so a country slice is a mixture of three. Only the CC0 ones can be reused with no attribution or non-commercial condition attached — and taking only those discards 72.71% of what GBIF holds for the Philippines.
That is the cost of the rule this page is built to, stated rather than hidden: 59.44% of the Philippine record is CC BY and 13.27% is CC BY-NC, and neither appears in any figure below. Every number on this page describes the 27.29% that is free of conditions.
24 countries publish Philippine biodiversity records. The Philippines is 6th, with 5,016 records against the United States' 706,959.
860,076 of the records are preserved museum specimens. Only 13,655 are human observations — somebody seeing a living thing and writing it down.
records are dated 1908 alone, one of several early-century collecting peaks. The single biggest year is 2023 with 168,934 — but 96.6% of that one year is a single expedition, which the next section is about.
149,896 records carry no year at all, so the chart of collecting over time is drawn on the 85.9% that do.
One record is catalogued as collected in 1612 — a weevil, in a genus not described until 1842. GBIF attaches four quality flags to that record. None of them is about the date.
The archive does flag dates: 119 Philippine records carry
RECORDED_DATE_UNLIKELY. This one is not among them —
0 of the records from that year are flagged. So the plausibility
bound on this page is its own, set at 1700, and anything below it is published as an
anomaly rather than drawn quietly into a chart of collecting over time.
That is the general lesson of an upstream quality flag: it tells you what the publisher thought to check, which is not the same as what is wrong.
Half the records — 540,011, or 50.9% — carry coordinates. Counting those into half-degree cells across the country gives a map of collecting effort, which is not a map of biodiversity.
of all mappable records sit in a single half-degree cell. The top ten cells hold 48.6%.
of 864 cells queried hold no reusable record at all. 464 hold at least one. Much of that blank is open sea — but not all of it.
keyless requests back this page. GBIF's bulk download needs an account, so nothing here downloads a record: each cell is one count query, and the map is the answers.
The busiest cell is at 8.5°N 119.5°E — Tubbataha Reefs, in the Sulu Sea. It holds 164,171 records, and 99.4% of them come from one dataset.
That dataset is a UNESCO environmental-DNA expedition of April 2023: 163,186 records from filtering seawater and sequencing what is in it. One method, one site, one month — and it is 30.5% of every mappable reusable record in the country, and 96.6% of the biggest collecting year on the chart above.
So the concentration on that map is not simply "people collect near cities". It is mostly one modern technique that turns a single site visit into six figures of records. Strip that expedition out and the map is flatter, older, and far more thinly spread — which is the honest picture of how much of the Philippines has actually been surveyed.
Four limits, and the first one bounds everything else.
The download endpoint answers 403 without an account and the search endpoint stops paging at 100,000 rows, so this page is aggregates only. No per-record inspection, no joins to anything outside GBIF.
The publishing country is the institution that uploaded the record. A specimen collected in Luzon and held in Leiden counts as Dutch here. That is the point, but it is not a statement about legal ownership.
A cell with many records is a cell people visited. An empty cell may be unstudied rather than lifeless, and the map cannot tell the two apart.
GBIF grows daily. Every response is cached so these figures are reproducible, but a fresh run will not reproduce them exactly.
Every figure on this page traces to one of these, through a CSV in
data/ph-gbif/. Each is checked against its source query on every
build.
Every Philippine occurrence record GBIF aggregates, and for the CC0 subset its publisher, licence, basis of record, collection year, taxonomic class, quality issues and position — Keyless: a plain GET, no account, no token. Queried with limit=0 so it returns counts and no rows, and with facet= so a whole distribution arrives in one request. 1,755 cached responses back this page.
The 1,060,583 Philippine records carrying a CC0 waiver, out of 3,886,391 in total — GBIF records carry the licence their publisher chose, so a country slice is a mixture. Filtering to CC0 leaves data with no attribution or non-commercial condition attached. It also discards 72.7% of what GBIF holds for the Philippines, which is published on the page rather than left implicit.
The name behind each numeric classKey returned by the occurrence facets — Facets return taxonomic keys, not names. Each key is resolved once and cached; a row whose name is still its key fails a check rather than reaching a chart axis.
Would have provided the individual records rather than their counts — Answers 403 without a registered GBIF account, and the search endpoint stops paging at 100,000 records, so the million-row slice cannot be retrieved without registering. That is what an account is, so it is out under the no-registration rule. The whole page is therefore built from aggregates the API will answer for free, which is a narrower instrument -- no per-record inspection, no joins to anything outside GBIF.
I wrote a companion post — “Who Keeps The Record Of Philippine Nature? Mostly, Not The Philippines.” — in simple, everyday words.